Repositorio Institucional de la Universidad Autónoma de Madrid https://repositorio.uam.es Esta es la versión de autor del artículo publicado en: This is an author produced version of a paper published in: The American Journal of Human Genetics 107.2 (2020): 234 – 250 DOI: https://doi.org/10.1016/j.ajhg.2020.06.006 Copyright: © 2020 Cell Press El acceso a la versión del editor puede requerir la suscripción del recurso Access to the published version may require subscription 1 The genetic landscape and epidemiology of phenylketonuria 2 Alicia Hillert1, Yair Anikster2, Amaya Belanger-Quintana3, Alberto Burlina4, Barbara K. Burton5, Carla Carducci6, Ana E Chiesa7, John Christodoulou8, Maja Đorđević9, Lourdes R. Desviat10, Aviva Eliyahu2, Roeland A.F. Evers11, Lena Fajkusova12, François Feillet13, Pedro Eduardo Bonfim Freitas14, Maria Giżewska15, Polina Gundorova16, Daniela Karall17, Katya Kneller2, Sergey I. Kutsev16, Vincenzo Leuzzi18, Harvey L. Levy19, Uta Lichter-Konecki20, Ania C. Muntau21, Fares Namour13, Mariusz Oltarzewski22, Andrea Paras5, Belen Perez10, Emil Polak23, Alexander V. Polyakov16, Francesco Porta24, Marianne Rohrbach25, Sabine SchollBürgi17, Norma Spécola26, Maja Stojiljković27, Nan Shen28, Luiz Carlos Santana da Silva14, Anastasia Skouma29, Francjan van Spronsen11, Vera Stoppioni30, Beat Thöny25, Friedrich K. Trefz1, Jerry Vockley20, Youngguo Yu31, Johannes Zschocke32, Georg F. Hoffmann1, Sven Garbade1*, Nenad Blau1,21* 3 4 5 6 7 8 9 10 11 12 13 14 1Dietmar-Hopp Metabolic Center, University Children’s Hospital, Heidelberg, Germany; 15 2Metabolic Disease Unit, Edmond and Lily Safra Children’s Hospital, Sheba Medical Center, 16 Tel-Aviv University, Tel-Aviv, Israel; 3Unidad de Enfermedades Metabolicas, Servicio de 17 Pediatria, Hospital Ramon y Cajal, Madrid, Spain; 4Division of Inherited Metabolic Diseases, 18 Department of Woman's and Child's Health, University Hospital, Padua, Italy; 5Ann & Robert 19 H. Lurie Children’s Hospital of Chicago, Chicago, IL, USA; 6Department of Experimental 20 Medicine, Sapienza University of Rome, Italy; 7Fundación de Endocrinología Infantil (FEI). 21 Buenos Aires, Argentina; 8Murdoch Children's Research Institute and Department of 22 Pediatrics, University of Melbourne, Melbourne, Australia; 9Institute of Mother and Child 23 Healthcare "Dr. Vukan Čupić", Belgrade, Serbia; 10Centro de Diagnóstico de Enfermedades 24 Moleculares, Centro de Biología Molecular CSIC-UAM, Universidad Autónoma de Madrid. 25 CIBERER, IdiPAz, Madrid, Spain; 11University of Groningen, University Medical Center 26 Groningen, Beatrix Children’s Hospital, Section of Metabolic Diseases, Groningen, the 27 Netherlands; 12Centre of Molecular Biology and Gene Therapy, University Hospital Brno, 28 Brno, Czech Republic; 13Reference Center for Inherited Metabolic Diseases, University 29 Hospital of Nancy, Vandoeuvre-lès-Nancy, France; 14Laboratory of Inborn Errors of 30 Metabolism, Institute of Biological Sciences, Federal University of Pará, Belém, Brazil; 31 15Department of Pediatrics, Endocrinology, Diabetology, Metabolic Diseases and Cardiology, 32 Pomeranian Medical University, Szczecin, Poland; 16Research Centre for Medical Genetics, 33 Moscow, Russia; 17Clinic of Pediatrics, Division of Inherited Metabolic Disorders, Medical 34 University of Innsbruck, Innsbruck, Austria; 18Department of Human Neuroscience, Sapienza 1 35 University of Rome, Italy; 19Division of Genetics and Genomics, Boston Children's Hospital, 36 Harvard Medical School, Boston, MA, USA; 20University of Pittsburgh, Children's Hospital of 37 Pittsburgh, Pittsburgh, PA, USA; 21University Children's Hospital, University Medical Center 38 Hamburg Eppendorf, Hamburg, Germany; 22Department of Screening and Metabolic 39 Diagnostics, Institute of Mother and Child, Warsaw, Poland; 23Comenius University, Faculty 40 of Natural Sciences, Department of Molecular Biology, Bratislava, Slovak Republic; 41 24Department of Pediatrics, AOU Citta' della Salute e della Scienza di Torino, Torino, Italy; 42 25Division of Metabolism, University Children’s Hospital, Zürich, Switzerland; 26Unidad de 43 Metabolismo. Hospital de Niños “Sor Ludovica” de La Plata, Buenos Aires, Argentina; 44 27Institute of Molecular Genetics and Genetic Engineering, University of Belgrade, Belgrade, 45 Serbia; 28 Department of Infectious Diseases, Shanghai Children’s Medical Center, Shanghai 46 Jiao Tong University School of Medicine, Shanghai, China; 29Institute of Child Health, Athens, 47 Greece; 30Centro Screening Neonatale Regione Marche, Azienda Ospedaliera Ospedali Riuniti 48 Marche Nord, Fano, Italy; 31Department of Pediatric Endocrinology/Genetics, Xinhua 49 Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai Institute for Pediatric 50 Research, Shanghai, China; 32Institute of Human Genetics, Medical University Innsbruck, 51 Innsbruck, Austria. 2 52 Abstract 53 Phenylketonuria (PKU), caused by variants in the phenylalanine hydroxylase (PAH) gene, is 54 the most common autosomal recessive Mendelian phenotype in amino acid metabolism. We 55 estimated that globally 0.45 million have PKU, with global prevalence 1:23,930 live births 56 (range 1:4,500 [Italy] – 1:125,000 [Japan]). Comparing genotypes and metabolic phenotypes 57 from 16,092 patients revealed differences in diseases severity in 51 countries from 17 world 58 regions, with the global phenotype distribution of 62% classic PKU, 22% mild PKU and 16% 59 mild hyperphenylalaninemia. A gradient in genotype and phenotype distribution existed 60 across Europe, from classic PKU in the east, to mild PKU in the southwest and mild 61 hyperphenylalaninemia in the south. The p.Y414C-associated genotype can be traced from 62 northern to western Europe, from Sweden via Norway, Denmark to central Europe. The 63 frequency of classic PKU increases from Europe (56%) via Middle-East (71%) to Australia 64 (80%). Of 758 PAH variants, p.R408W (22.2%), c.1066-11G>A (6.4%) and p.R261Q (5.5%) 65 were most common and responsible for two prevalent genotypes: p.[R408W];[R408W] 66 (11.4%) and c.[1066-11G>A];[1066-11G>A] (2.6%). Most genotypes (73%) were compound 67 heterozygous, 27% were homozygous and 55% of 3,659 different genotypes occurred in a 68 single patient only. PAH variants were scored using an allelic phenotype value and correlated 69 with pre-treatment blood phenylalanine levels (n=6,796) and tetrahydrobiopterin test 70 results (n=4,381), enabling a genotype-based phenotype (88%) and tetrahydrobiopterin 71 responsiveness (83%) prediction. This study shows that large genotype databases enable 72 accurate phenotype prediction, allowing appropriate targeting of therapies to optimize 73 clinical outcome. 3 74 Introduction 75 Phenylketonuria (PKU; MIM 261600) is the most frequent inborn errors of the amino acid 76 metabolism. It is caused by more than 1,180 biallelic variations in the phenylalanine 77 hydroxylase (PAH) gene located on chromosome 12q22-24.1.1 These autosomal recessive 78 inherited variations lead to a deficiency in the PAH enzyme, which normally hydroxylates 79 phenylalanine into tyrosine, with the help of a cofactor (tetrahydrobiopterin; BH 4), 80 molecular oxygen, and non-heme iron.2 81 The clinical picture is highly heterogenous as it depends on the degree of residual PAH 82 activity and blood phenylalanine (Phe) levels. Usually, lower residual enzyme activity results 83 in higher the blood phenylalanine levels and a more severe PKU phenotype.3 84 The severity of PKU is classified according to the pre-treatment blood phenylalanine 85 concentration and daily dietary phenylalanine tolerance, ranging from the most severe 86 classical PKU (cPKU) with pre-treatment blood phenylalanine levels of >1,200 µmol/L, to mild 87 PKU (mPKU) with pre-treatment blood phenylalanine levels of 600–1,200 µmol/L and mild 88 hyperphenylalaninemia (MHP) with pre-treatment phenylalanine blood levels of 120–600 89 μmol/L.1,4 90 Untreated PKU in newborns can result in global developmental delay or severe irreversible 91 intellectual disabilities, as well as growth failure, hypopigmentation, motor deficits, and 92 ataxia, and seizures.1 The population of patients with PKU is also heterogeneous in terms of 93 treatment history and diet compliance.4,5 Early diagnosis and treatment with a low-Phe diet 94 has enabled an almost normal life for the majority of PKU patients.6 Pharmacological 95 treatment with BH4 (sapropterin) and enzyme substitution therapy with Phe ammonia lyase 96 (PAL) provide alternative treatment options for some patients.7 97 PKU is one of the most frequent inherited disorders in Caucasians, with an incidence of 98 roughly 1:10,000 live births in the USA,8 although the prevalence of PKU varies significantly 99 among ethnicities and geographic regions worldwide. In Europe, the incidence of PKU ranges 100 from 1: 850 in the Karachay-Cherkess Republic (Russia)9 to only 1:112,000 live births in 101 Finland.10 PKU occurs less often Japan, with an incidences of 1:120,000.11 102 A large number of PAH variations (>1,300) give rise to a wide scale of residual PAH enzyme 103 activities that correspond to different PKU phenotypes.12,13 Associations between genotypes 4 104 and in vitro residual PAH activity have been documented for many PAH variations.3 105 Therefore, the molecular genetics of PKU and genotype-based phenotype prediction may be 106 clinically useful, particularly where treatment recommendations are unclear (e.g. due to 107 borderline blood phenylalanine levels) or for genetic counseling of patients’ families. 108 The major goal of this study was the analysis of the largest phenotypes and genotypes 109 database to elucidate the current distribution of PKU worldwide and to create an overview 110 of the severity of PAH variants, genotypes and the resulting phenotypes worldwide, in 111 various geographic regions and respective countries. Furthermore, this work improves the 112 accuracy the genotypic phenotype prediction by the use of allelic phenotype value (APV) 113 value and the genotypic BH4 responsiveness prediction. 114 115 Subjects and methods 116 LITERATURE SEARCH 117 An electronic search using the databases MEDLINE (via Pub Med), the Cochrane library and 118 Web of Science was carried out to compare articles covering the epidemiology and genetics 119 of PKU in different world regions between January 1980 and October 2019. Key words 120 included Phenylketonuria[MeSH Terms] OR "phenylalanine hydroxylase/deficiency"[MeSH 121 Terms] OR hyperphenylalaninaemia[Title/Abstract] OR 122 hyperphenylalaninemia[Title/Abstract] OR PAH deficiency[Title/Abstract] OR phenylalanine 123 hydroxylase deficiency[Title/Abstract] OR phenylketonuri*[Title/Abstract] OR 124 pku[Title/Abstract]) AND ( "1980/01/01"[PDat] : "2019/12/31"[PDat] ) AND 125 Humans[Mesh] Filters: Publication date from 1980/01/01 to 2019/12/31; Humans. 126 This literature search yielded 5,459 records without duplicates, of which 1,118 papers with 127 an appropriate title and abstract were assessed. The final number of relevant records was 128 256. (Supplementary Figure 1) 129 130 131 5 132 DATABASES 133 The PAH locus-specific database PAHvdb, ClinVar, HGMD and LOVD databases were 134 searched for variants. PAHvdb is linked with the genotypes-phenotypes BIOPKU database 135 and was used for analyses in this work. It follows the HGVS nomenclature recommendations. 136 The BIOPKU database encompasses information about more than 16,900 PKU patients from 137 51 countries, providing information on patients’ genotype, corresponding metabolic 138 phenotypes, BH4 responsiveness (where reported) and highest blood Phe levels before 139 starting the treatment (where reported). Individual information was collected from the 140 published literature or anonymized records submitted online. Phenotype information was 141 unknown for 690 patients. Table 1 shows the information included in the database. 142 DEFINITION OF PHENOTYPES 143 We used the term “PKU” to refer to the entire spectrum of phenotypes due to PAH 144 deficiency. Since not all countries use the same nomenclature for the severity of PKU and 145 HPA, we used in this study the following 3 phenotype groups: classical PKU (cPKU; pre- 146 treatment blood Phe >1200 µmol/L); mild PKU (mPKU; pre-treatment blood Phe 600–1200 147 µmol/L) and mild HPA (MHP; pre-treatment blood Phe 120–600). Patients with a moderate 148 PKU (pre-treatment blood Phe 900–1,200 µmol/L) were included in the mPKU group and 149 MHP included MHP-no treatment (blood Phe 120–360 µmol/L) and MHP-gray zone (360–600 150 µmol/L) categories.4 Any classifications, not fitting into one of the above groups (due to 151 different country-specific classifications), were reassigned on the basis of reported pre- 152 treatment blood Phe levels and the genotype. 153 DEFINITION OF THE ALLELIC (APV) AND GENOTYPIC PHENOTYPE VALUE (GPV) 154 APV is a value defining the association of a variant with the corresponding metabolic 155 phenotype, thus defining its severity. APV was calculated for variants occurring in a 156 functionally hemizygous constellation (i.e. in a combination with an inactive 0-allele) in at 157 least 5 patients.14 APVs range between 0 and 10, with following classification definitions 158 cPKU (APV = 0-2.7), mPKU (APV = 2.8-6.6) and MHP (APVs 6.7-10).14 159 The genotypic phenotype value (GPV) was calculated from APVs of both alleles and was 160 assigned to a higher APV (APVmax). This calculation was based on the fact that the milder 6 161 variant (with a higher APV) is always dominant over the severe one.14,15 Possible effects of 162 interallelic complementation and epigenetic factors, which may influence the 163 phenotype16,17, were not considered in this study. 164 DEFINITION OF BH4 RESPONSIVENESS 165 BH4 responsiveness was defined as a ≥30% reduction of blood Phe levels within 24-48 hours 166 after the administration of BH4 (20mg/kg body weight)18,19. A linear discriminant analysis was 167 applied to predict BH4 responsiveness based on the APV for untested patients. 168 STATISTICAL ANALYSIS 169 The statistical analysis was performed using R, an open source software and flexible 170 programming language used for the statistical data analysis as well as graphic creations 171 (https://www.r-project.org/ ). A total of 16,196 records with a complete genotype 172 information (variant 1 and 2 known) were analyzed. 173 174 Results 175 PREVALENCE OF PKU 176 Based on the literature search and reports from national screening centers, the prevalence 177 of PKU was estimated for 64 countries. For parts of Africa, Asia, Southern America and 178 Caribbean there was no information. The estimated total number of PKU patients (all 179 phenotypes) from those 64 countries in 2018 was 360,466. For the remaining 257 countries 180 we were unable to find credible PKU prevalence sources; we used the average regional 181 prevalence of 64 countries, multiplied by their populations, resulting in an additional 94,114 182 PKU patients (total about 0.45 million PKU patients). The global PKU prevalence was 183 estimated to be 1: 23,930 newborns (Figure 1, Supplementary Table 1). 184 The PKU prevalence was highest in European and certain Middle Eastern populations. Italy 185 (1:4,000) and Ireland (1:4,545) had even higher prevalence than Iran and Jordan (both 1: 186 5,000) or Turkey (1:6,667). Saudi Arabia (1:14,245), Iraq (1:14,286) or United Arab Emirates 187 (1:14,493) had lower PKU prevalence, however. 7 188 PKU prevalence was also high in central European countries PKU affect patients at high rates 189 as well e.g. Germany (1:5,360), Czechia (1:5,521) and Slovakia (1:5,753). Slovenia (1:7,143), 190 Austria (1:5,764) or Poland (1:8,309) had similar rates to eastern Europe, e.g. Estonia 191 (1:7,143), Russia (1: 7,714) Belarus (1:7,692), and Croatia (1:8,333). 192 PKU occurred slightly less frequently in western Europe, e.g. in France (1:9,091), United 193 Kingdom (1:10,000), Belgium (1:11,000) or the Netherlands (11,546), and in southern 194 Europe, e.g. Spain (1:10,115) and Portugal (1:12,500). 195 Northern Europe presented with the lowest PKU rates in Europe, e.g. for Norway (1:11,457), 196 Sweden (1: 12,681), Denmark (1: 13,434), or only 1:112,000 in Finland. 197 PKU occurred more frequently In Canada (1:15,000) than in the United States of America 198 (1:25,000) or in Latin American countries, including Chile (1:19,231), Brazil (1:25,000), 199 Mexico (1:27,778), Argentina (1:33,333) or Peru (1:46,970). 200 The lowest PKU prevalence was reported in Asian countries, such as Thailand (1:227,273), 201 Japan (1:125,000), Philippines (1:116,006) or Singapore (1:83,333). One exception was China 202 where the PKU prevalence was 1:15,924, which was comparable to Europe. 203 DESCRIPTIVE ANALYSIS OF THE PHENYLALANINE HYDROXYLASE GENE LOCUS-SPECIFIC DATABASE 204 Substitutions were by far the most frequent variant type In the PAHvdb (80.5%), followed by 205 deletions (12.9%) and duplications (2.1%). Of all variants, 691 were missense variants 206 (58.3%), followed by 165 frameshift variants (13.9%) and 155 splice site (13.1%) variants. 207 Nonsense variants (6.9%), synonymous variants (4.9%) and in-frame variants (1.9%) were 208 less frequent. Extension, complex and unknown variants, as well as polymorphisms, 209 accounted for the remaininder. Exon 6 contained the most variants (14.1%), followed by 210 Exon 7 (12.2%) and Exon 3 (9.9%). Most variants (59.2%) were located in the central catalytic 211 domain, 17.5% in the N-terminal regulatory domain and 5.4% in the C-terminal 212 oligomerization domain of the PAH monomer. The remaining variants (17.9%) were either in 213 the intronic or UTR regions. Only 7.7% of all variants were located in one of the 4 cofactor 214 binding regions (Supplementary Figure 2). 215 The APV was known for 589/1,186 variants. Most variants (441) were defined as severe 0- 216 alleles (APV=0), 32 variants as cPKU phenotype, 52 as mPKU and 64 as mild MHP alleles. 8 217 DESCRIPTIVE ANALYSIS OF THE BIOPKU DATABASE 218 As of October 2019, the BIOPKU database contained anonymized data on more than 16,900 219 PKU patients. A total of 16,196 PKU patients with 3,659 different genotypes were analyzed. 220 Information on the PKU phenotype was available for 16,902 patients. Maximum 221 pretreatment Phe levels were recorded for only 6,796 patients and the BH4 test information 222 was available for 4,381 patients (Table 1). 223 GLOBAL PKU PHENOTYPE AND GENOTYPE DISTRIBUTION 224 Information on the PKU phenotype was available for 16,092 out of 16,196 patients (99%). Of 225 these, most had cPKU (9,923; 61.7%), 3,521 (21.9%) had mPKU, and 2,648 had MHP (16.4%). 226 Information about the phenotype was unavailable for 104 patients 227 The comparison of pre-treatment Phe level with the phenotype (n=6,796) is illustrated in 228 Figure 2A and with the GPVs (n=6,371) in Figure 2B. The interquartile range (n, median, 25 th– 229 75th percentile) was smallest and lowest for the MHP patient group: 1,283, 320 µmol/L, 242– 230 432 µmol/L, and was larger for mPKU and cPKU (1,487, 793 µmol/L, 660–793 µmol/L, and 231 3,599, 1,550 µmol/L, 1,270–1936 µmol/L, respectively). 232 A total of 758 different alleles were identified in this study. The three most prevalent 233 variants were p.R408W, with an allele frequency (AF) =22.2%, c.1066-11G>A (IVS10-11G>A; 234 AF=6.4%) and p.R261Q; AF 5.5%). Figure 3 shows the most frequent phenotype-specific 235 variants. A full list of PAH variants is shown in Supplementary Table 2. 236 Of all patients, 11,810 (72.9%) were compound heterozygotes and 4,386 (27.1%) were 237 homozygotes. Of 3,659 genotypes, 3,446 (94.2%) were compound heterozygotes and 213 238 (5.8%) homozygotes. The three most prevalent genotypes were: p.[R408W];[R408W] with a 239 genotype frequency (GF) =11.4%, followed by c.[1066-11G>A];[1066-11G>A] (GF=2.6) and 240 c.[1222C>T];[1315+1G>A] (p.[R408W];IVS12+1G>A) (GF=1.6). A full list of genotypes is 241 shown in Supplementary Table 3. Strikingly, 54.5% of all genotypes were specific for only one 242 patient and were not used for the phenotype prediction. 243 244 9 245 BH4 RESPONSIVENESS 246 Information on BH4 responsiveness was available for 4,381 patients. About half were 247 classified as BH4 responsive (2,053; 47%). Supplementary Table 4 shows the distribution of 248 BH4 responsiveness in different phenotype groups. As expected, milder forms of PKU 249 appeared more likely to be BH4 responsive, whereas most patients with cPKU were non- 250 responders. 251 Pretreatment blood Phe levels (n, median, 25th-75th percentile) were much lower for BH4 252 responsive patients (1,116, 620 µmol/L, 411–853 µmol/L), compared with non-responders 253 (1,180, 1,361 µmol/L, 1039–1719 µmol/L (Figure 2C). The 11,584 patients not tested for BH4 254 responsiveness were analyzed for potential responsiveness using the GPVs. Patients with an 255 GPV>3.8 (n=3,023; 26%) were assigned as potential BH4 responders (p<0.001 vs. patients 256 assigned as non-responders [n=8,561; 74%]) (Figure 2D). 257 PKU PHENOTYPES AND GENOTYPES IN WORLD REGIONS 258 Overview of global data 259 Patient data originated from 51 countries from 17 world regions and a total of 15,357 260 patients from 33 countries with at least 35 reported cases, were analyzed (Supplementary 261 Tables 5 and 6, Supplementary Figure 3). Most cases were reported from central (18.3%) and 262 eastern (18.1%) Europe, eastern Asia (13.8%), western Europe (9.6%) and the Middle-East 263 (9.2%). Very few patients were reported from the northern and southern parts of Africa. 264 cPKU was the most frequent phenotype in all world regions, with high rates reported in 265 Australia and eastern Europe (81%), and rates of <50% reported only in Serbia, Argentina, 266 Turkey, Netherlands, Sweden, Spain, Italy, Japan, Slovenia, Germany, and Taiwan (Figure 267 3A). Of 2,835 patients in Russia, 81% had cPKU, 9.9 % had mPKU and 9.1% had MHP. Estonia, 268 representing the Baltic region, reported 93.5% of patients with cPKU. The ratio of 269 mPKU:MHP patients were comparable, except for Eastern Asia, North America and South 270 America, where mPKU was more common. The phenotype distribution in world regions is 271 shown in Supplementary Figure 4. 272 Reports from individual regions are summarized in Table 2 and below. 273 10 274 Europe 275 Variant p.R408W was the most common (AF=63.7%), followed by c.1066-11G>A (AF=11%) 276 and p.R261Q (AF=11%). Eastern Europe had the highest AF for p.R408W (54.6%), mostly with 277 homozygous genotype p.[R408W];[R408W] (GF=32.7%); findings were similar for central 278 Europe (AF=44.4% and GF=23.8%). Russia contributed the largest number of records. A 279 single variant, p.R408W (AF=53.7%), was dominant in Russia, followed by p.R261Q 280 (AF=5.6%) and p.P281L (AF=4.1%). The most common genotype was p.[R408W];[R408W] 281 (30.6% of 458 different genotypes). (AF=82.6% for p.R408W). In southeastern Europe 282 p.R408W was also the most frequent allele, but p.[L48S];[R408W] was the most prevalent 283 genotype (GF=10.6%) 284 The distribution of patients with p.R408W on at least one allele (compound heterozygotes 285 and homozygotes) decreased from 98% in Estonia to 89% in Poland, 76% in Russia, 69% in 286 Slovakia, 65% in Czechia, 40% in Austria, 36% in Germany, 10% in France, 6% in Italy, and 287 only 4% in Spain (Table 3, Figure 4). Genotypes with the IVS10-11G>A splice site variant, 288 occurred commonly in Armenia (48%), Turkey (32%), Iran (26%), Israel (21%), Spain (20%) 289 and Italy (19%) (Table 3, Figure 4). The prevalence of p.R261Q was 10–30% in different 290 countries. 291 The distribution of PKU phenotypes was more even in southern Europe (39.5% cPKU, 36.6% 292 MHP, 23.9% mPKU) and the milder phenotype was more predominant, e.g. 37% had MHP in 293 Italy. This was consistent with a high frequency of variants with a substantial residual PAH 294 activity, e.g. p.A403V (AF=8.4), p.A300S (AF=3.6), p.V245A (AF=2.4) or p.Y414C (AF=2.4). 295 Variants c.1066-11G>A and p.R261Q, often called the “Mediterranean mutation” accounted 296 for majority of mutant alleles, and the most frequent genotype was c.[1066-11G>A];[1066- 297 11G>A] (GF=5.2%). In Portugal and Italy c.1066-11G>A and p.R261Q occurred at similar rates 298 (AF=16.2% and AF=10.7%, respectively). Variant p.V388M was also common in Portugal 299 (AF=11.3%) and Spain (AF=6.8%). 300 Low rates of cPKU were found in southwestern Europe (42.7%) and northern Europe 301 (50.5%). In southeastern Europe, Croatia (68%) and Bulgaria (65%) had a higher prevalence 302 of cPKU than Serbia (48%). p.R408W was the most frequent variant in Croatia (AF=31.4%) 303 and Bulgaria (AF=32.4%), but was less common in Serbia (AF=14.7%). The p.L48S variant, 11 304 which was initially identified in Turkey,20 has the highest reported AF (31.3%) in Serbia. The 305 most frequent genotypes in other countries locally were p.[L48S];[L48S] (14.7 %) in Serbia; 306 p.[L48S];[R408W] (14.3 %) in Croatia; and p.[R408W];[R408W] genotype (16.2 %) in Bulgaria. 307 In northern Europe c.1315+1G>A was the most common splice site variant (AF=25%). Most 308 patients were compound heterozygous for p.[R408W];c.[1315+1G>A]. Denmark had the 309 largest sample size in this study, with the highest number of cases of the mild p.Y414C 310 variant in the in the world. Sweden and Norway had a relatively high rate of mPKU (>43%). In 311 Denmark the most frequent variant was c.1315+1G>A with an AF= 27.3%, while p.G46S 312 occurred more commonly in Sweden and Norway. Variants p.R408Q and p.F299C were 313 specific for Norway. 314 The p.R408W variant was also the most frequent variant for all central European countries, 315 except for Switzerland where p.R261Q (AF=15.3%) was more prevalent. Classical PKU was 316 particularly frequent in Poland and Slovakia (>70%), but not in Czechia, consistent with a 317 higher prevalence of the mild variant, p.A403V. There was a wide spectrum of PKU variants 318 in Germany: of 102 distinct variants, p.R408W accounted for only 19.3%, and other variants 319 such as p.Y414C (AF=9.7 %) as well as c.1315+1G>A (AF=8.8 %) were prominent. Almost half 320 (44%) of patients in Germany had the mPKU phenotype, 32% had cPKU and 24% had MHP. 321 France represented the largest European sample (n=1,307): 59.1% had cPKU, and there were 322 a total of 229 different variants, e.g. c.1066-11G>A (AF=7.4%), p.R261Q (AF=6.5%), p.R408W 323 (AF=5.5%), c.1315+1G>A (AF=4.5%) and p.E280K (AF=3.7%). The p.R408W variant was also 324 less common in the Netherlands (AF=2.2%). The predominant genotype in western Europe 325 was p.[E280K];[E280K]. 326 Latin America 327 The most frequent variants here, p.V388M (AF= 13.9%), p.R261Q (AF=10.7%) and c.1066- 328 11G>A (AF=9.4%), were also prominent mutations in southern Europe (see above). 329 Homozygous p.[V388M];[V388M] (GF=4.4%) occurred most frequently. Of the three relevant 330 South American countries Brazil had the highest numbers of cPKU (63%), followed by Mexico 331 (57%) and Argentina (43%). In comparison to Brazil or Argentina, Mexico’s most frequent 332 variant was c.60+5G>T (IVS1+5G>T). 12 333 North America 334 The most prevalent variants in North America also resembled those in European 335 populations: p.R408W (18.5%), c.1066-11G>A (7.9%), c.1315+1G>A (6.9%), as did the 336 genotype distribution p.[R408W];[R408W] (GF=4.0%) 337 Middle East 338 The predominance of c.1066-11G>A variant (AF=20.1%) and it’s homozygous genotype 339 (GF=15.3%) was evident In Iran, Turkey, Israel and Saudi Arabia. Other frequent variants 340 included p.R261Q, c.168+5G>C (IVS2+5G>C), p.P281L and p.R243*. 81% of the Iranian 341 patients had cPKU. In comparison to Iran, Turkey had a similar AF for c.1066-11G>A (22.9 %) 342 and p.R261Q (11.8 %), but strongly different phenotypes: 42% had cPKU, 35% MHP and 23% 343 mPKU. This may be related to the augmented presence of mild variants such as p.A300S 344 (7,4%) or p.E390G (4.2%). In contrast to other Middle Eastern countries, the most frequent 345 variant from Saudi Arabia was p.R252W (AF=27,6%), followed by p.R261* (11.2 %) and 346 p.V388M (10.4 %); 30% of all patients had a p.R252W-genotype (Table 3). 347 Asia 348 Most reported cases from Asia came from the east, with the most prevalent being the 349 missense variant p.R243Q (AF= 21.8%), followed by Ex6-96A>G (9.8%) and p.R241C (8.7%). 350 The most frequent genotype was homozygous p.[R243Q],[R243Q] (GF=5.6%). 351 In China, 62% of patients (n=2008) had a cPKU, 28% mPKU and 10% the MHP phenotype. A 352 total of 234 different variants were reported, out of which five p.R243Q, Ex6-96A>G, 353 p.R241C, p.R111* and p.R413P, each with a frequency of >5%, which accounted for 52.5% of 354 all alleles. Furthermore 679 different genotypes were reported, the most frequent being 355 p.[R243Q];[R243Q]. Korea was the only country that like China has a high rate of cPKU (71%), 356 whereas cPKU in Japan (37%) and Taiwan (25%) was less common. The p.R243Q, Ex6-96A>G, 357 and p.R241C, each with allele frequencies >5%, were detected in Korea, Taiwan, and China. 358 The p.R111* and p. R413P variants were common in Japan, China and Taiwan, nut not Korea. 359 The splice variant IVS4-1G>A was much more prevalent in Korea than in other Eastern Asian 360 countries. 13 361 Genotypes with p.R243Q were common in China (40%) and Korea (27%) (Table 3), while 362 those with p.R241C were prevalent in Taiwan (41%), Japan (27%) and China (15%) (Table 3). 363 Patients with the splice site variant IVS4-1G>A were more common in Korea (18%), followed 364 by China (8%) and Japan (7%). The splice variant Ex6-96A>G (c.611A>), which masquerades 365 as missense variant p.Y204C21), was found 19% of patients in China, 13 in Taiwan and 12 in 366 Korea (Table 3). 367 Africa 368 The low number of reports available from Africa were considered not sufficiently 369 representative of this world region, and we did not analyze these data. 370 371 Discussion 372 The aim of this retrospective study was to elucidate the PKU prevalence and distribution of 373 causative PAH variants worldwide and in different countries. We provided a rough 374 estimation of global PKU prevalence by calculating the number of affected patients for 375 countries based on the provided PKU prevalence and the total population in 2018 22 and an 376 average prevalence for countries not employing newborn screening for PKU. Overall, it 377 seems that there are about 0.45 million PKU patients worldwide, of whom at least two thirds 378 have PKU that requires treatment (most patients had a severe, cPKU phenotype). An 379 improvement in early diagnosis via NBS in states without this should be an urgent priority. 380 Although PKU has been found to be most common in European populations,8 its similar 381 prevalence in certain Middle Eastern countries (particularly Turkey and Iran) was a 382 remarkable finding.23,24 A possible contributing factors may be a tendency to consanguinity 383 in Arabic cultures, especially first cousin marriages, which favors the autosomal recessive 384 inheritance of PKU.25 Prevalence and marriage status was missing for many countries, 385 making it more difficult to assess this proposed causal relationship for all states.26 386 We identified 758 different PKU alleles in this study, emphasizing the strong genetic 387 heterogeneity of PKU. Information of about 10,000 genotypes and phenotypes from PKU 388 patients in Europe enabled the a more detailed study of their distribution across different 389 regions and countries than has been possible previously.27,28 The severe p.R408W variant 14 390 was the most common overall, especially among Eastern European populations, in 391 accordance with previous studies.29,30 Furthermore, p.R408W dominates most Central 392 European populations (Poland, Slovakia, Czechia, Austria, or Germany), also supporting 393 previous studies.31-35 394 Previous research had suggested that some PKU variants appeared to have been carried by 395 migration, e.g. p.R408W, IVS10-11G>A, IVS12+1G>A or p.F299C and R408Q.27 Our study 396 confirms and extends the knowledge on how certain variants appear to have spread within 397 Europe and to different regions and countries. In particular, the distribution of PKU variants 398 across Europe was consistent with successive waves of historical migration, and Figure 4 399 summarizes the likely geographical routes of transmission of these variants within Europe. 400 For example, Germany has occupied a crossroads location during several migration waves 401 throughout history, and displays a broad spectrum of PAH variants, with similarities to some 402 other populations, such as Northern European countries (Denmark, Sweden and Norway). 35 403 This implies a genetic connection of these regions, possibly during the Germanic settlement 404 of Scandinavia. Norway’s most frequent variant, p.G46S, did not occurring in the German 405 group. Furthermore, the occurrence of IVS10-11G>A in Germany was surprisingly high in our 406 analysis, which might be explained by immigration from Turkey. This variant, often described 407 as the “Mediterranean mutation”, may be of Italian origin.36 and has been found mainly in 408 Southern countries or the Middle-East.37,38 Our study showed that IVS10-11G>A has the 409 highest AF in Middle Eastern countries (Turkey, Iran and Israel), and in Portugal and Spain, 410 The p.R408W-associated genotype, the most common variant in people with Slavic roots, 411 followed the east-central-southwest European axis, starting from Estonia, with the highest 412 number of severe PKU patients, via Russia, Poland, Czechia, Slovakia, Germany, France, and 413 Italy, down to Spain. The high prevalence of cPKU in Russia, and the dominance of the severe 414 variant, p.R408W, were striking, given that Russia has a diverse ethnic composition, with 415 >160 ethnic groups.39 416 Another well-documented axis exists for the p.Y414C-associated genotype, from northern to 417 western Europe, i.e. from Sweden via Norway, Denmark and Germany to the Netherlands. 418 Some variants, e.g., IVS12+1G>A, seem to have migrated within Scandinavian countries 419 (Denmark >Sweden >Norway) at high frequency. The same was true for the p.E390G- 420 associated genotype concentrating between Slovenia, Croatia, Serbia and Austria. A typical 15 421 east-west gradient can be seen for IVS10-11G>A, originating from Western Asia and the 422 Middle-East, i.e., from Armenia (48%) via Turkey, Iran, Israel to Spain. Similar consequences 423 of migration were evident for the very mild p.A300S-associated genotypes, migrating from 424 Turkey via Israel, Italy to Spain. No trends for p.V245A and p.A403V distribution across 425 Europe could be seen, however. 426 The influence of migration patterns during history can also be seen in the heterogenous 427 spectrum of PAH variations in Latin America. These countries were strongly marked by 428 immigration of Europeans, especially from southern Europe, during colonial times.40 An 429 example is p.V388M, which was especially prevalent in Brazil and Chile, once colonies of 430 Portugal and Spain. The Argentina cohort was more prevalent for variations frequently 431 occurring in eastern, central and northern Europe; e.g., as p.R408W, p.R261Q and IVS10- 432 11G>A. In contrast to Brazil or Argentina, Mexico was characterized by a high percentage of 433 indigenous population, which could be a reason for the exceptional high AF of the 434 IVS1+5G>T splice variant there.41 435 The USA has one of the most multi-racial and ethnic populations in the world. The strongest 436 ancestral influence is European, mainly from Germany, Ireland, England, Italy and France 42, 437 which explains a comparable distribution of PKU variants in the USA and Europe. 43 African 438 Americans have a much lower PKU incidence than Caucasian Americans,44 and observing 439 changes in the growth of causative variants in USA, major ethnic groups there (Caucasian, 440 African-American, Hispanic, Asian) grow at different rates would be an interesting area for 441 future research. 442 While the overall PKU prevalence in China was 1:15,924, its distribution across the country 443 varies significantly with higher rates in the north in comparison to the south.45 In accordance 444 with previous reports, p.R234Q accounts for 23% of all variants in China. 46 Three additional 445 variants p.R243Q, Ex6-96A>G and p.R241C (AF>5%) were common in Korea, Taiwan as well, 446 indicating commonalities of migration movements and evolution among those nationalities. 447 Prevalent PKU variants for Europe, Middle-East, Latin America and the USA were uncommon 448 in the Asian populations. 449 Preliminary data indicated a value of large patient databases with the genotype and 450 phenotype information and introduction of the APV and GPV for the genotypic phenotype 451 prediction.14 This study extends this previous knowledge with additional information, and 16 452 has confirmed the power of genotyping in phenotype definition and BH4 responsiveness in 453 PKU, thus offering a powerful tool in the personalized medicine of this inherited metabolic 454 disease. 455 This study also documents that functional mild variants (in a compound heterozygous 456 constellation), with a substantial residual PAH activity, were always dominant over inactive 457 severe variants (0-alleles) and gave raise to a milder phenotype and potential BH4 458 responsiveness. Compared with severe classic PKU variants, they were much rarer in 459 number, but determined the milder metabolic phenotype. Interestingly, homozygous mild 460 variants have higher APV and thus milder phenotype than when in a compound 461 heterozygous state with a 0-alle14. Two inactive severe variants are never BH4-responsive, as 462 a rule, although anecdotal reports sometimes suggest this.47 463 In conclusion, this study provides an overview of the current distribution of the most 464 important PAH variants and patients genotypes in various world regions. This information, 465 together with the APV value, helps to predict the PKU phenotype and the possible treatment 466 options. Despite immense progress in diagnosis and treatment of PKU in the last decade 467 there are still too many areas of the world without adequate access to this care. Especially 468 data from Africa, certain Asian and Southern American as well as Caribbean countries was 469 missing. 470 471 Description of Supplemental Data 472 Supplementary Figure 1. Flow-chart of the literature search. 473 Supplementary Figure 2. Physical structure of the PAH gene and variants tabulated in the 474 locus-specific database (http://www.biopku.org/pah/home.asp). Regulatory domain: residues 475 1-142; catalytic domain: residues 143-410; oligomerization domain: residues 411-452. 476 Cofactor (BH4)-binding regions (CBR): CBR1: residues 246-266; CBR2; residues 280-283; 477 CBR3: residues 322-326; CBR4: residues 377-379. The reference accession number for the 478 PAH sequence is ENSG00000171759; RefSeq NM_000277.1. 479 Supplementary Figure 3. Heat maps of the distribution (%) of classic PKU (cPKU), mild PKU 480 (mPKU) and mild hyperphenylalaninemia (MHP) A) in countries with at least 35 reported 17 481 cases and B) heat maps of overall genotypes severity, expressed as genotypic phenotype 482 value (GPV), The lower the GPV, more severe the phenotype. 483 Supplementary Figure 4. Heat maps of the distribution (%) of classic PKU (cPKU), mild PKU 484 (mPKU) and mild hyperphenylalaninemia (MHP) in A) different European regions and C) in 7 485 world regions. Heat maps of overall genotypes severity, expressed as genotypic phenotype 486 value (GPV), in B) European regions and D) in 7 world regions. The lower the GPV, more 487 severe the phenotype. F) Ten most common variants, their allele frequency (AF) and 488 corresponding allelic phenotype values (APV) for the 3 PKU phenotype categories. The 489 reference accession number for the PAH sequence is ENSG00000171759; RefSeq 490 NM_000277.1. 491 Supplementary Table 1. Prevalence of PKU (all phenotypes) in world regions and countries 492 and estimated number of PKU patients. 493 Supplementary Table 2. PAH variations (n=758) found in the study cohort of PKU patients. 494 Supplementary Table 3. Genotypes (n=3659) found in the study cohort of PKU patients. 495 Supplementary Table 4. Distribution of BH4 responsiveness in different phenotype groups. 496 Those classified as “slow responders” are included in the non-responsive group are included 497 in the non-responsive group and descriptive analysis of maximal pretreatment blood Phe 498 levels and BH4 responsiveness. 499 Supplementary Table 5. Definition of countries in the world regions. 500 Access to Supplementary files: 501 https://drive.google.com/drive/folders/15DI4NQfczgjqojGizRj19colMrjvtPW0?usp=sharing 502 503 Declaration of interests 504 Authors: please enter any COI! 505 506 18 507 Acknowledgements 508 We are indebted to Suha Daas, who works at the National Newborn screening Program, 509 Ministry of Health, Tel Hashomer Israel for her help and analyses. Farrah Rajabi, Fran Rohr, 510 Ann Wessel, Leslie Martell, Steven Dobrowolski, Per Guldberg, and the late Flemming Guttler 511 were greatly responsible for the data from New England. This work was funded in part by 512 the Fundación Isabel Gemio-Fundación La Caixa (LCF/PR/PR16/11110018), and the Regional 513 Government of Madrid (CAM, B2017/BMD3721). The research conducted at the Murdoch 514 Children’s Research Institute was supported by the Victorian Government's Operational 515 Infrastructure Support Program. 516 517 Web Resources 518 BIOPKU: http://www.biopku.org/biopku/home.asp 519 ClinVar: https://www.ncbi.nlm.nih.gov/clinvar/ 520 HGMD: http://www.hgmd.cf.ac.uk/ac/index.php 521 LOVD: https://www.lovd.nl/ 522 HGVS nomenclature recommendations: http://www.hgvs.org/content/guidelines 523 PAHvdb: http://www.biopku.org/pah/home.asp 524 19 525 References 526 1. Blau N, Van Spronsen FJ, Levy HL. Phenylketonuria. Lancet 2010;376:1417-27. 527 2. Flydal MI, Martinez A. Phenylalanine hydroxylase: function, structure, and regulation. IUBMB 528 life 2013;65:341-9. 529 3. 530 genotype, phenylalanine hydroxylase expression and in vitro activity and metabolic phenotype in 531 phenylketonuria. Mol Genet Metab 2018;125:86-95. 532 4. 533 the science and future research needs. Mol Genet Metab 2014;112:87-122. 534 5. 535 and management of patients with phenylketonuria. Lancet Diabetes Endocrinol 2017;5:743-56. 536 6. 537 and management guideline. Genet Med 2014;16:188-200. 538 7. 539 phenylketonuria. Expert Opin Pharmacother 2015;16:1-10. 540 8. 541 45. 542 9. 543 Molecular-genetic causes for the high frequency of phenylketonuria in the population from the North 544 Caucasus. PLoS One 2018;13:e0201489. 545 10. 546 incidence population: molecular characterisation of mutations in Finland. J Med Genet 1995;32:976- 547 8. 548 11. 549 and tetrahydrobiopterin-responsive phenylalanine hydroxylase deficiency in Japan. Journal of human 550 genetics 2011;56:306-12. 551 12. 552 phenotypical diversity of phenylketonuria and related hyperphenylalaninaemias. J Inherit Metab Dis 553 1993;16:602-4. 554 13. 555 art. Expert review of molecular diagnostics 2014;14:655-71. 556 14. 557 based phenotype prediction in phenylketonuria. Genet Med 2019;21:580-90. Himmelreich N, Shen N, Okun JG, Thiel C, Hoffmann GF, Blau N. Relationship between Camp KM, Parisi MA, Acosta PB, et al. Phenylketonuria scientific review conference: State of van Spronsen FJ, van Wegberg AM, Ahring K, et al. Key European guidelines for the diagnosis Vockley J, Andersson HC, Antshel KM, et al. Phenylalanine hydroxylase deficiency: diagnosis Blau N, Longo N. Alternative therapies to address the unmet medical needs of patients with Scriver CR. The PAH gene, phenylketonuria, and a paradigm shift. Hum Mutat 2007;28:831- Gundorova P, Zinchenko RA, Kuznetsova IA, Bliznetz EA, Stepanova AA, Polyakov AV. Guldberg P, Henriksen KF, Sipila I, Guttler F, de la Chapelle A. Phenylketonuria in a low Okano Y, Kudo S, Nishi Y, Sakaguchi T, Aso K. Molecular characterization of phenylketonuria Güttler F, Guldberg P, Henriksen KF, Mikkelsen I, Olsen B, Lou H. Molecular basis for the Blau N, Shen N, Carducci C. Molecular genetics and diagnosis of phenylketonuria: state of the Garbade SF, Shen N, Himmelreich N, et al. Allelic phenotype values: a model for genotype- 20 558 15. Zschocke J. Dominant versus recessive: molecular mechanisms in metabolic disease. J Inherit 559 Metab Dis 2008;31:599-618. 560 16. 561 Genet 1999;15:267-72. 562 17. 563 hydroxylase variants and effects of interallelic complementation on in vitro enzyme activity and 564 genotype-phenotype correlation. Mol Genet Metab 2015;117:328–35. 565 18. 566 Vliet, Danique ; Blau, Nenad ; van Spronsen, Francjan J. . The first European guidelines on 567 phenylketonuria: its usefulness and implications for BH4 responsiveness testing. Journal of inherited 568 metabolic disease 569 2019. 570 19. 571 evaluation of responsiveness to sapropterin dihydrochloride in patients with phenylketonuria. Mol 572 Genet Metab 2019;127:1-11. 573 20. 574 mutations at the PAH gene locus in a Turkish patient with phenylketonuria. Hum Genet 1991;87:389- 575 93. 576 21. 577 masquerading as missense mutation (Y204C). Hum Mutat 1997;9:88-90. 578 22. 579 population (both sexes combined) by region, subregion and country. 2019. 580 23. 581 Biochem 1991;24:337-42. 582 24. 583 countries, Turkey, and Iran: A systematic review. BioMed Res Internat 584 2018;doi.org/10.1155/2018/7697210. 585 25. 586 classic phenylketonuria based on neonatal screening programs: Systematic review and meta-analysis. 587 Clin Exp Pediatr 2020;63:34-43. 588 26. 589 Africa. J Inherit Metab Dis 2007;30:482-9. 590 27. Scriver CR, Waters PJ. Monogenic traits are not simple: lessons from phenylketonuria. Trends Shen N, Heintz C, Thiel C, Okun JG, Hoffmann GF, Blau N. Co-expression of phenylalanine Evers RAFvW, Annemiek M J ; Anjema, Karen ; Lubout, Charlotte M A ; van Dam, Esther ; van Muntau AC, Adams DJ, Belanger-Quintana A, et al. International best practice for the Konecki DS, Schlotter M, Trefz FK, Lichter-Konecki U. The identification of two mis-sense Ellingsen S, Knappskog PM, Eiken HG. Phenylketonuria splice mutation (EXON6nt-96A-->g) United Nations -Population division DoEaSA. World population prospects, File Total Matalon R, Michals K. Phenylketonuria: screening, treatment and maternal PKU. Clin El-Metwally A, Al-Ahaidib LY, Sunqurah AA, et al. The prevalence of phenylketonuria in Arab Shoraka HR, Haghdoost AA, Baneshi MR, Bagherinezhad Z, Zolala F. Global prevalence of Saadallah AA, Rashed MS. Newborn screening: experiences in the Middle East and North Zschocke J. Phenylketonuria mutations in Europe. Hum Mutat 2003;21:345-56. 21 591 28. Danecka MK, Woidy M, Zschocke J, Feillet F, Muntau AC, Gersting SW. Mapping the 592 functional landscape of frequent phenylalanine hydroxylase (PAH) genotypes promotes personalised 593 medicine in phenylketonuria. J Med Genet 2015;52:175-85. 594 29. 595 patients with phenylketonuria reveal a high rate of BH4 non-responders in Russia. PLoS One 596 2019;14:e0211048. 597 30. 598 Phenotype Correlation and Comparative Overview of the Patient Cohort Before and After Nation- 599 Wide Neonatal Screening. JIMD Rep 2017:10.1007/8904_2017_61. 600 31. 601 potential impact of mutations on BH4 responsiveness. Acta Biochim Pol 2013;60:613-6. 602 32. 603 analysis. Hum Genet 1995;95:112-4. 604 33. 605 Hyperphenylalaninemia in the Czech Republic: Genotype-phenotype correlations and in silico 606 analysis of novel missense mutations. Clinica chimica acta; international journal of clinical chemistry 607 2013;419:1-10. 608 34. 609 among Austrian patients with PAH deficiency: comprehensive results from molecular analysis in 147 610 patients. Journal of inherited metabolic disease 2013;36:7-13. 611 35. 612 8. 613 36. 614 Am J Hum Genet 1992;51:1355-65. 615 37. 616 Mediterranean. Genomics 1991;9:96-103. 617 38. 618 Genet 1994;2:24-34. 619 39. Statistics FOo. “On the results of the All-Russia Population Census 2010". 620 40. Desviat LR, Perez B, De Lucca M, Cornejo V, Schmidt B, Ugarte M. Evidence in Latin America 621 of recurrence of V388M, a phenylketonuria mutation with high in vitro residual activity. Am J Hum 622 Genet 1995;57:337-42. 623 41. 624 deficiency in Mexico: genotype-phenotype correlations, BH4 responsiveness and evidence of a 625 founder effect. Clin Genet 2015;88:62-7. Gundorova P, Stepanova AA, Kuznetsova IA, Kutsev SI, Polyakov AV. Genotypes of 2579 Lillevali H, Reinson K, Muru K, et al. Hyperphenylalaninaemias in Estonia: Genotype- Bik-Multanowski M, Kaluzny L, Mozrzymas R, et al. Molecular genetics of PKU in Poland and Kadasi L, Polakova H, Ferakova E, et al. PKU in Slovakia: mutation screening and haplotype Reblova K, Hruba Z, Prochazkova D, Pazdirkova R, Pouchla S, Fajkusova L. Sterl E, Paul K, Paschke E, et al. Prevalence of tetrahydrobiopterine (BH4)-responsive alleles Zschocke J, Hoffmann GF. Phenylketonuria mutations in Germany. Hum Genet 1999;104:390- Eisensmith RC, Okano Y, Dasovich M, et al. Multiple origins for phenylketonuria in Europe. Okano Y, Wang T, Eisensmith RC, et al. Phenylketonuria missense mutations in the Kleiman S, Avigad S, Vanagaite L, et al. Origins of hyperphenylalaninemia in Israel. Eur J Hum Vela-Amieva M, Abreu-Gonzalez M, Gonzalez-Del Angel A, et al. Phenylalanine hydroxylase 22 626 42. Ramirez KRHNAJRR. Overview of Race and Hispanic Origin: 2010. United States Census 627 Bureau 2011. 628 43. 629 in the United States: a single center study. Mol Genet Med 2019:submitted. 630 44. 631 of the phenylalanine hydroxylase gene. American journal of human genetics 1991;48:791-8. 632 45. 633 based on data from the Chinese newborn screening information system: a descriptive study. BMJ 634 Open 2019;9:e031474. 635 46. 636 population using next-generation sequencing. Sci Rep 2015;5:15769. 637 47. 638 responsiveness in classical and mild phenylketonuria (PKU). J Inherit Metab Dis 2002;25:21. Rajabi F, Rohr F, Wessel A, et al. Phenylalanine hydroxylase genotype-phenotype associations Hofman KJ, Steel G, Kazazian HH, Valle D. Phenylketonuria in U.S. blacks: molecular analysis Xiang L, Tao J, Deng K, et al. Phenylketonuria incidence in China between 2013 and 2017 Li N, Jia H, Liu Z, et al. Molecular characterisation of phenylketonuria in a Chinese mainland Hennermann JB, Vetter B, Kulozik AE, Mönch E. Partial und total tetrahydrobiopterin- 23 639 Legends to figures 640 Figure 1. Prevalence (A: 1:<30,000; B: 1:>30,000 live births) of PKU (all phenotypes) in 64 641 countries from 6 world regions. 642 Figure 2. A: Boxplot (median, 25th-75th percentile, 1.5) of the maximal pretreatment blood 643 Phe levels for three metabolic phenotypes in 6,796 PKU patients. The circles in the cPKU bar 644 represent ordinary high blood Phe levels, since cPKU doesn’t have an upper Phe limit for its 645 classification; B: Contour plot of two-dimensional densities of pretreatment blood Phe levels 646 and corresponding genotypic phenotype values (GPV) for 6,796 PKU patients; C: Boxplot 647 (median, 25th-75th percentile, 1.5) of GPV in 2,246 BH4 non-responder and 1,755 responder 648 PKU patients; D: Boxplot (median, 25th-75th percentile, 1.5) of GPV (APVmax) for 11,584 PKU 649 patients with a known genotype, but not tested for BH4 responsiveness. Horizontal grey bar: 650 separation area between GPVs for predicted BH4 responsiveness (3.8-10) and non- 651 responsiveness (0-3.3). 652 Figure 3. The world map with the relative frequency (%) of PKU and the corresponding most 653 common variants for classic PKU (cPKU), mild PKU (mPKU) and mild hyperphenylalaninemia 654 (MHP). AF: allele frequency; APV: allelic phenotype value (cPKU=0-2.6; mPKU=2.7-6.6; 655 MHP=6.7-10). 656 Figure 4. Likely routes of transmission of common PAH variants associated with migration in 657 Europe. 24