High-risk screening for FD in selected patient cohorts has been reported. Doheny et al.  reanalyzed studies related to hemodialysis (27 reports, 23,954 men, and 12,866 women), left ventricular hypertrophy (LVH) and/or hypertrophic cardiomyopathy (17 reports, 4,054 men, and 1,437 women), and ischemic or cryptogenic strokes (16 studies, 3,904 men, and 2,074 women). The revised prevalence was estimated as 0.21% for male and 0.15% for female hemodialysis patients, 0.94% for male and 0.90% for female cardiac patients, and 0.13% for male and 0.14% for female stroke patients. In the current study, the prevalence was estimated as 0.42% (male: 0.45%, female: 0.38%) in the renal manifestation group (a), 0.94% (male: 0.77%, female: 1.33%) in the cardiac manifestation group (b), and 0.22% (male: 0.20%, female: 0.27%) in the central neurological manifestation group (c), which are comparable to those of previous reports.
The prevalence of FD in the peripheral neurological manifestation group (d) was second highest at 4.37% (male: 4.98%, female: 3.50%), whereas the patients’ age (male: median 13 [IQR: 11-18.5] years old, female: median 13 [IQR: 9-25] years old) was lesser than that of the other groups. Therefore, manifestations such as limb pain, acroparesthesia, clustered angiokeratoma, cornea verticillata, and hypo- or anhidrosis, could help identify FD patients. Politei et al.  has recommended that the cause of pain should be diagnosed early in unrecognized or newly diagnosed FD patients to improve treatment possibilities. FD experts consider that, regardless of sex or age, pain related to FD could be an early indication to commence ERT before potentially irreversible organ damage, to the kidneys, heart, or brain, prevails. However, a study conducted in Russia by Namazova-Baranova et al.  reported that no FD patients were identified from among 214 individuals (110 males and 104 females) with chronic limb pain. Moreover, the genetic, epidemiological, and ethnical information related to Russian FD patients are insufficient and future studies and information related to FD in Russia are required.
The prevalence of FD in individuals with a family history (e) was the highest at 23.40% (male: 12.49%, female: 33.02%). GLA sequencing for individuals with a family history of FD was useful in identifying undiagnosed or pre-symptomatic FD patients. Therefore, when patients experience FD-related symptoms, clinicians should confirm the presence of a family history of FD and, if applicable, whether similar symptoms developed.
The variant spectra of GLA in Japanese patients have been reported [26, 27]. GLA gene analysis was previously performed for 207 FD patients . The most common variant was c.888G>A/p.M296I (allele frequency: 5.8%, 12/207). The second-most common variants were c.639+919G>A (4.3%, 9/207) and c.679C>T/p.R227* (4.3%, 9/207), followed by c.334C>T/p.R112C (3.9%, 8/207), c.335G>A/p.R112H (3.9%, 8/207), and c.902G>A/p.R301Q (3.9%, 8/207). In another study, 73 pathogenic variants were detected in 176 patients from 115 families  and the most common variant was c.334C>T/p.R112C (allele frequency: 2.65%). The second-most common variant was c.888G>A/p.M296I (1.89%), followed by c.658C>T/p.R220* (1.52%), c.718_719delAA/p.K240Efs*8 (1.52%), and c.1025G>A/p.R342Q (1.52 %). The common variants identified in these studies, as well as those of the current study, overlapped.
We previously reported the first large-scale NBS program for FD in the western region of Japan . A total of 599,711 newborns were screened and 26 GLA variants, including 8 novel variants, were detected in 57 neonates from 54 families. Of the 26 variants, 10 were also detected in the current study and most of them were detected in patients from western Japan (Figure S1).
In the current study, 4 pedigrees (4/68, 5.9%) were perceived as de novo mutations (Table S3). The frequency might be comparable with those of previous studies performed in Japan (6.8% (5/74); Kobayashi et al. ), Italy (2.8% (3/108); Romani et al. ; and 14.3% (2/14); Morrone et al. ), Spain (4.5% (1/22); Rodriguez-Mari et al. ), and the United Kingdom (6.3 % (1/16); Davies et al. ). A high frequency of de novo mutations has been reported in X-linked disorders, such as Duchenne muscular dystrophy (DMD) and hemophilia A (F8), and de novo mutations account for approximately one-third of the mutations in these two disorders [33, 34]. This is owing to lack of fitness to reproduce in the X chromosome. Hemizygous GLA mutations may have sufficient fitness to reproduce, reducing the frequency of de novo mutations . Moreovrer, the presence of CpG dinucleotides also may increase mutational frequency . GLA contains 19 CpG sites (1/68 bases) in the coding region compared to F8 which has 68 CpG sites (1/104 bases). Of the 19 potential mutation sites in GLA, 13 variants were identified in the current study, namely, c.146G>C, c.334C>T, c.335G>A, c.427G>A, c.658C>T, c.659G>C, c.679C>T, c.901C>T, c.902G>A, c.902G>C, c.1024C>T, c.1025G>A, and c.1066C>T. Twenty-three variants were reported as de novo mutational hotspots (Table S3). However, no particular sites responsible for these de novo mutations were identified.
A few cases of homozygous or compound heterozygous female FD patients have been previously reported [30, 37]. However, homozygous or compound heterozygous female FD patients were not identified in the current study or in our previous NBS study . In our NBS study, the frequency of male FD (the allele frequency of FD variants) was estimated to be 1:6,212 (0.016%), and the probability of homozygous female was extremely low as 1:38,588,944. Therefore, among female FD patients, only those with heterozygous GLA mutations are generally identified. Interestingly, a male patient harboring two GLA variants, c.70T>A /p.W24R and c.1255A>G/p.N419D, was identified in the current study (Table S1). Unfortunately, genetic information regarding his family and his chromosome information were not available. It was unclear whether the two variants were in cis on a single X chromosome. Of 100 GLA variants, 70 were detected only in single pedigrees, whereas 20 were identified in two pedigrees. Because a bias was introduced in the distribution of variants in these pedigrees, it was difficult to discuss the correlation between genotype and phenotype, especially organ specific pathogenicity.
On follow-up evaluation for each patient, 21 of the 26 novel variants were indicated as pathogenic, namely c.97G>C/p.D33H, c.157A>T/p.N53Y, c.184dupT/p.S62Ffs*18, c.205T>C/p.F69L, c.207del/p.F69Lfs*52, c.264C>G/p.Y88*, c.329del/p.P110Lfs*11, c.386_389dupTGAA/p.L131Efs*9, c.440G>T/p.G147V, c.563delC/p.Y188Sfs*4, c.610T>G/p.W204G, c.691_693GAC>TAT/p.D231Y, c.825delC/p.S276Afs*6, c.827G>C/p.S276T, c.848A>G/p.Q283R, c.908T>C/p.I303T, c.987C>A/p.Y329*, c.1019delG/p.E341Nfs*57, c.1054G>C/p.A352P, c.1085_1088dupCTCG/p.Y365Lfs*11, and c.1100dupT/p.A368Rfs*7. Furthermore, five variants were identified herein, which were not registered in the ClinVar or Fabry-database.org, specifically c.725T>C/p.I242T, c.801+1G>A/p.L268Ifs*3, c.908_923del21/p.S304_310Ldel, c.1124G>A/p.G375E, and c.1165C>G/p.P389A. Patients harboring the aforementioned variants developed FD-related symptoms, and some had even died of stroke or cardiac failure.
Even during high-risk FD screening, individuals are assigned an uncertain diagnosis in the absence of classical FD symptoms and when variants of unknown significance (VOUS) in the GLA gene are identified. This leads to a risk of misdiagnosis, inappropriate counseling, and extremely costly treatment. Therefore, numerous studies have attempted to generate a diagnostic algorithm for FD, which maximally excludes these risks . In our high-risk screening, individuals presenting decreased activity (<cutoff levels) with known pathogenic variants, classical signs or symptoms of FD, or a family history of FD were definitely diagnosed with FD. However, among individuals presenting decreased activity (<cutoff levels) with VOUS and late onset signs or symptoms such as cryptogenic stroke, proteinuria, or LVH without classic type signs or symptoms, a definite diagnosis is difficult to achieve. Moreover, because the disease state during late-onset FD is potentially not improved through ERT , the therapeutic effect of ERT does not facilitate the diagnosis of FD. Blood Lyso-Gb3 assays and tissue diagnosis in a myocardial or renal biopsy may be sufficient for a definite diagnosis of FD . Moreover, analysis using iPSC technology, such as Gb3 accumulation in iPSC-derived vascular endothelial cells, may lead to a definite diagnosis .
High-risk FD screening has a potential for false-positive findings. Figures 2A, 2B, and 2C show the histograms of this high-risk screening for all individuals, men, and women, respectively, in Method I. The median α-Gal A activity was 24.47, 24.50, and 24.06 (AgalU) among all individuals, men, and women, respectively. A dotted line indicates the cutoff value: <12 [AgalU] for men and <20 [AgalU] for women; 50% of median α-Gal A activity for men and 80% of median α-Gal A activity for women. Heterozygous female patients have an almost normal range of α-Gal A activity, resulting in false-negative findings in screening studies. Linthorst et al.  reported that 40% (16/40) of female patients with FD are not identified, considering a cutoff <50% of the normal control. Although we used a higher cutoff <80% of median α-Gal A activity, false-negative findings may have been obtained among the female FD patients herein. Additional tests, such as blood Lyso-Gb3 assays , hotspot mutation screening , or even whole GLA gene sequencing, may improve the rate of false-negative results. Most patients with FD in Taiwan harbor variants out of a pool of only 21 pathogenic mutations . Therefore, in regions such as Taiwan, where hotspot mutations can be detected, hotspot mutation screening is effective for high-risk screening among women. In regions such as Japan, where hotspot mutations cannot be detected and several variants are found, hotspot mutation screening is not as effective. Whole GLA gene sequencing is difficult to perform among all female patients included in the high-risk screening group because of cost-balance issues. The assay of Lyso-Gb3 in dried blood spots (DBSs) is considered an effective and realistic alternative for high-risk screening among women . We will consider applying the Lyso-Gb3 assay for high-risk screening in future studies.
Figure 2D shows a histogram of α-Gal A activity in an NBS study in Method I . The median α-Gal A activity among neonates was 42.58 (AgalU), which is approximately 2-fold that of the current high-risk screening populations. FD is associated with a significantly reduced life expectancy compared to that of the general population . Although the detailed mechanism for the low α-Gal A activity in adults is unknown, it may be associated with a premature aging process through the dysfunction of blood vessels. Therefore, aging and low α-Gal A activity are closely related.
The cutoff values in the high-risk screening populations were 12 (AgalU) for men and 20 (AgalU) for women, which is representative of the cutoff values for the 0.5 percentile in the NBS population. This is because α-Gal A activity in adults is lower than that of neonates. The current high-risk screening program identified individuals who are considered suitable candidates for migalastat treatment. Some patients were already receiving migalastat treatment. Moreover, gene therapy holds promise in effectively treating various diseases, and the clinical trials for gene therapy for FD are ongoing in Canada and the USA (https://fabrydiseasenews.com/gene-therapy-for-fabry-disease/). In the future, the development of new treatment methods for FD, other than ERT, is expected.