Introduction

Diagnosing lysosomal storage diseases (LSDs) typically involves enzyme activity measurement, genetic testing, and clinical assessment. However, the diagnosis of acid sphingomyelinase deficiency (ASMD), also called Niemann-Pick disease (NPD) type A, B, or A/B, and Gaucher disease (GD) can be challenging due to their rarity and nonspecific clinical complications, including hepatosplenomegaly and thrombocytopenia1,2. Patients with severe subtypes, such as NPD type A and GD type 2, that exhibit neurological involvement have high mortality rates1,2. Interestingly, GD types 2 and 3, the severe subtypes, are more common, accounting for more than 50% of GD patients in Japan3,4, possibly biasing incidence rates. Moreover, there may be undiagnosed patients with subtle or nonspecific clinical findings. Even with the increasing need for accurate diagnosis of these disorders, carrier frequencies in Japan remain uncharacterized despite advancements in therapeutic approaches, such as olipudase alfa for ASMD and eliglustat for GD5,6. We sought to estimate the frequencies of GD and ASMD in Japan using publicly available genome databases.

Methods

We reviewed the variant allele frequencies (VAFs) of highly pathogenic SMPD1 and GBA1 variants in Japanese population and populations with various ancestry backgrounds using two genomic control databases: the Japanese Multi-Omics Reference Panel 54KJPN (jMorp, https://jmorp.megabank.tohoku.ac.jp/, containing genomic data from 54,302 Japanese individuals) and the Genome Aggregation Database v4.0.0 (gnomAD, https://gnomad.broadinstitute.org/, containing genomic data from 807,162 individuals). In gnomAD, we used only a subset of the variant data that passed a quality control filter in the exome. For estimating the prevalence, “DM (disease-causing)” variants reported in patients affected with ASMD and GD entered into the Human Gene Mutation Database (HGMD Professional 2023.4, http://www.hgmd.org/) were defined as pathogenic variants. Additionally, we included potential novel loss-of-function variants, such as pathogenic, frameshift, nonsense, initiation codon, and canonical ± 1 or 2 spice site variants, in SMPD1 and GBA1, even those not registered in the HGMD. Variants classified as benign in the ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar/) were excluded. Small insertion and deletion SMPD1 variants on chr11:6391620_6391652 (GRCh38/hg38) were not included in this study because the quality values of variants in this region were all low in jMorp, indicating that the mapping process could fail. The VAFs of the pathogenic variants were summed for each gene. The carrier and disease frequencies were calculated as 2pq and q2, respectively, based on the Hardy‒Weinberg equilibrium (p = allele frequency of wild-type individuals, q = VAF) using the total VAF7.

Results

In the Japanese population, 24 variants in SMPD1 and 29 variants in GBA1 from jMorp were identified as pathogenic according to our criteria (Tables 1 and 2). The most common variant type was missense—16 of 24 SMPD1 variants (66.7%) and 22 of 29 GBA1 variants (75.9%). Pathogenic SMPD1 variants that are commonly reported worldwide, including p.Arg610del, p.Arg498Leu, and p.Phe333Serfs*52, were not included in jMorp8,9. Patients with mild forms of ASMD can have a natural history of morbidity extending into their 60s10. Considering that 88.1 million individuals in the Japanese population were under 65 years old in 2023, the estimated number of ASMD patients is 687, which is much larger than expected. Notably, the 5 variants with the highest VAFs have been reported in patients with NPD type B, regardless of zygosity (Table 1). p.Pro186Leu, p.Gly244Arg, and p.Arg602His were also identified in NPD type B patients, who are known to have a high residual acid sphingomyelinase activity of 13-43%11. A few pathogenic variants associated with NPD type A or A/B were identified.

Table 1 Allele frequencies of 24 pathogenic variants in SMPD1 among Japanese population from jMorp.
Table 2 Allele frequencies of 29 pathogenic variants in GBA1 among Japanese population from jMorp.

The spectrum of GBA1 variants included in jMorp was similar to that identified in Japanese GD patients reported by Eto et al. (Table 2)12. The VAFs of the common variants p.Leu483Pro, p.Phe252Ile, and p.Asp448His, which have been observed in homozygous GD type 3 patients, were high in jMorp, reflecting the high prevalence of GD type 3 in Japan13,14. Our data predict that there are 929 GD patients among the Japanese population under 65 years of age, which is 7.5 times greater than the 124 patients clinically diagnosed in Japan as of 200915. However, this estimation has limitations because it does not account for the mortality rates of all GD types.

Discussion

ASMD and GD have diverse incidence rates worldwide, and our data confirmed this diversity (Supplementary Table S1). ASMD is more prevalent in Ashkenazi Jewish and Middle Eastern populations, and the incidence rate of GD is also much greater (1 in 1,124, as estimated previously) in Ashkenazi Jewish people due to a founder effect of the p.Asn409Ser variant (Table 3 and Supplementary Tables S2 and S3)7. We found that the spectrum of the SMPD1 and GBA1 variants in Japan was distinct from that in populations with high frequencies except for East Asians, suggesting that the Japanese population has a unique variant distribution in ASMD and GD. The variant spectra in the East Asian population were similar to those in Japan. Specifically, 41.7% (10 out of 24 variants) of the pathogenic variants in SMPD1 and 34.5% (10 out of 29) of those in GBA1 found in Japanese people were also present in East Asian population in the gnomAD database, which may include some contributions from individuals of Japanese descent (Supplementary Tables S2 and S3).

Table 3 Estimated frequencies of ASMD and GD in Ashkenazi Jewish and Middle Eastern populations from gnomAD.

A nationwide survey revealed that only three ASMD patients were living in Japan as of 20214. Our study revealed a gap between the number of identified patients and the calculated disease frequencies, which could be explained by patients who have yet to be identified. One important observation was the high frequencies of the variants associated with the milder form of nonneuronopathic NPD type B. Compared to hepatosplenomegaly, other accompanying symptoms, such as lung infiltration and thrombocytopenia, may develop later or are not significant enough to be recognized until hepatosplenomegaly is detected8. It is speculated that unidentified patients may have hepatosplenomegaly with other symptoms that are not severe enough to receive medical attention or to consider ASMD as a part of a differential diagnosis. Variants such as p.Ser438Arg, p.Ser233Pro, and p.Ala601Thr have been previously identified in Japanese patients with NPD type B16,17. However, these variants were extremely rare or not listed in jMorp, suggesting that they are sporadic private variants.

Many patients with GD were found to have known recombinant alleles harboring the p.Asp448His, p.Leu483Pro, and p.Ala495Pro variants in cis by homologous recombination between the GBA1 gene and the GBAP1 pseudogene18. This event could have resulted in a falsely higher estimated frequency of GD in this study. However, a trial of newborn screening for GD in a local prefecture in Japan revealed that GD occurred in 1 out of 77,720 births, which is close to our result of 1 in 94,79119.

This study has limitations, including functionally unverified novel presumable loss-of-function variants and variants annotated as “uncertain significance” or not registered in ClinVar. Because of this limitation, some benign variants unrelated to the disease may have been included in our calculations, even if they were identified in ASMD or GD patients. Additionally, copy number variations and novel missense variants of unknown significance were excluded from the calculations. Despite these constraints, the infrequent occurrence of these variants suggests that our results remain largely unaffected. The prevalence of autosomal recessive inheritance disorders can be influenced by genetic factors, including carrier frequency, founder effects, and consanguinity. However, our findings suggested that the perceived prevalence based on the clinical diagnosis rate may be influenced by other factors, such as lack of awareness of the diseases. The use of genome databases is crucial in modern medicine for determining true disease prevalence. Our findings could increase disease awareness, aiding in identifying more patients who may benefit from emerging therapies. Moreover, this knowledge may prompt physicians to consider ASMD and GD in patients with nonspecific clinical symptoms, such as hepatosplenomegaly and thrombocytopenia.