This study investigated parental mosaicism in 20 families, in which the children had been previously diagnosed with developmental disorders with apparent de novo variants identified through trio sequencing. Utilizing family history, additional biological samples including buccal mucosa and semen samples, coupled with sensitive technologies including ddPCR and BDA, we illustrated that 20% (4/20) of the families have parental mosaic variants which is previously regarded as DNVs in the patients.

The median maternal age at conception in our cohort was 34.0 years, ranging from 25–41 years of age. Excluding the three families without an affected first child, the median maternal age at first childbirth for our cohort was 34.4 years old (ranging from 26–40 years old), slightly higher than the data reported by the Census and Statistics Department Hong Kong in 2021 [25], where the maternal median age at first childbirth was 32.6 years. The median paternal age at first childbirth for our cohort is 36.0 years old (ranging from 29–44 years old). There was a lack of Census data for paternal median age at first childbirth for a similar comparison, highlighting that discussions regarding parental age at childbirth tend to focus predominantly on females, despite the significant role of paternal age [22]. According to Census data [25], both the median age at first marriage for female and male has been steadily increasing, from 27.5 and 30.2, respectively, in 2001 to 30.6 and 32.2, respectively, in 2021, which can be inferred that maternal and paternal age at first childbirth may also be on a steady increase. While our current societal focus revolves around maternal age as a public health concern, it is equally important to educate the public about the potential risk of advanced paternal age and the burdens it may bring.

Despite the significant differences in inclusion criteria, detection methods and sample types used, the diagnostic yield of 4/20 (20%) disease-causing parental mosaicism found in our cohort with diverse Mendelian diseases was comparable with previous studies, where parental mosaicism were found between 0.3%-26.5% [4, 26, 27] in cohorts that focused on similar diverse Mendelian disease. Among the families with detected parental mosaicism, half had two affected siblings. This suggests that parental mosaicism should be strongly considered in families where the disease recurs. Moreover, our data suggest that the empirical VAF obtained from laboratory tests may further help predict recurrence risk. A higher VAF in multiple tissues indicates an earlier occurrence during embryonic development leading to a higher recurrence risk. This is exemplified in family15 with paternal gonosomal mosaicism, in which both affected offspring inherited the FLNC variant from the father. We identified the paternal mosaic variant in tissues representing ectoderm (buccal), intermediate mesoderm (sperm) and mesoderm (blood). The average VAF in sperm was higher (29.3%) compared to blood and buccal samples (average VAF of 14.5% and 15.0%, respectively). These results imply that low level parental mosaicism is prone to be missed if only blood or buccal samples are tested.

The mutational event in family 14 with the paternal gonadal CHD7 variant would have occurred later after differentiation of PGCs since the mosaic variant is confined in the semen at 13.9% and not found in blood and buccal. The recurrence risk in this family appears to be smaller than that of family 15, as demonstrated by the presence of two affected and one healthy offspring. However, as CHD7 is a candidate PAE gene associated with proliferation advantages; the recurrence risk may increase with the father’s age. Although Pauli et al. [28] did not find any PAE in their cohort of affected children carrying CHD7 variants of paternal origin (n = 12), earlier studies by Tellier et al [29] (n = 41) and Blake et al [30] (n = 39) suggested an association between CHD7 mutation and PAE.

Although PAE may increase the prevalence of paternal mosaicism, and previous studies report a ratio of paternal to maternal DNVs is at 4:1 [8, 31, 32], our observations confirm that maternal mosaicism still exist, as demonstrated by two families in our study. To increase the sample size, we reviewed parental mosaicism studies with more than ten families since 2009 (Table 2). Among those with a known positive parental mosaic variant affecting non-sex chromosome, the incidence of paternal mosaicism and maternal mosaicism were comparable at 59% (47/80) and 41% (33/80), respectively. On the other hand, X-linked recessive diseases (e.g., DMD) have a higher maternal mosaicism rate and it is not surprising because males are more likely to have an X-linked recessive diseases as their X-chromosome can only be inherited from their mother. Besides, X-linked dominant diseases seems to show a higher rate of paternal mosaicism (e.g., MECP2 and PCDH19 in Rett Syndrome and Developmental and epileptic encephalopathy 9, respectively, as shown in Table 2), however, a more thorough literature search that includes all cases using an unbiased approach is required to confirm these preliminary findings.

Indeed, the most accurate method for detecting gonadal mosaicism should involve direct observation of germ cells. While sperm can be sued to detect paternal gonadal mosaicism, maternal gonadal mosaicism would require an invasive biopsy of ovaries [33], which is impractical in most cases. Therefore, known cases of maternal gonadal mosaicism are likely to be underestimated. Notably for the same reason, the absence of detectable mosaicism in paternal semen does not necessarily stratify low recurrence risk unless the DNV of interest is known to phase to paternal allele, which usually requires long-read sequencing that is not commonly accessible in routine clinical laboratories [14]. Caution is required to completely rule out maternal mosaicism. Nonetheless, the use of blood and saliva analysis could possibly pick up both maternal and paternal gonosomal mosaicism [34].

Although the first few reported parental mosaicism cases in the late 1980s and early 1990s mainly focused on isolated families with genetic disorders such as Duchenne muscular dystrophy and osteogenesis imperfecta [35, 36] with limited molecular evidence available. However, the rapid advancement of molecular technologies, particularly NGS, has accelerated the identification ofparental mosaicism in a broader range of disorders, for example epilepsy [37,38,39], ASD [14, 40, 41] and developmental delay (DD) [3], as well as a wide spectrum of genetic disorders including Marfan [42], Noonan [43], polycystic kidney [44], primary immunodeficiency diseases [45] and congenital heart diseases [46] (Table 2).Earlier studies, such as Myers et al. [37] showed that among 120 children having epilepsy with an apparently DNV, approximately 10% of them had a parent with mosaicism. Krupp et al. [41] also showed that parental mosaicism was found to range from 7–11% in a large ASD cohort with 2300 families using blood as the sample type. Using semen as the sample type, Breuss et al. [14] showed that 29% (4/14) fathers were mosaic for the causative DNVs transmitted to their ASD-affected children. For intellectual disability (ID) caused by DNVs, paternal mosaicism was found in 4.7–6.5% of the families in cohorts of around 50 patients [7, 15]. In a cohort of 237 patients with a DNVs among a wide spectrum of developmental disorders, Shu et al. [27] also found 3% parental mosaicism using ES as the detection method at read depth of 2000X.

Based on the above studies, it can be concluded that parental mosaicism can be found in 3–29% of various developmental disease cohorts. Diagnostic yield for parental mosaicism is much lower at 0.3–0.5% if a non- targeted approach was used, as demonstrated in two large cohort studies using ES at a read depth of 50-130X as the initial detection method. Wright et al. [3] examined trio ES data of 4,293 probands at ~ 50 X average depths from the Deciphering Developmental Disorders (DDD) Study and only 0.5% of parental mosaicism was found. Similar study performed by Cao et al. [4] based on ES data of 11,992 probands at 130X average depth from Baylor Genetics identified only 0.3% parental mosaicism in the analyzed families. While only seven mosaic variants were identified directly by trio ES, 33 mosaic variants could be found during Sanger confirmation. These larger cohort studies with > 4000 patients indicated that parental mosaicism may have been underestimated unless a more targeted approached is implemented in clinical settings. Despite its increasingly strong association with many disorders and great clinical impact, detection of mosaicism is still mostly passive and technically challenging. This underscores the need for continued research and development of more sensitive and targeted detection methods.

Our study, in line with previous studies, had demonstrated that the detection of parental mosaicism is challenging as mosaicism maybe tissue-specific or tissue-limited [2, 3, 12, 14, 20, 26, 47]. Using Sanger sequencing as a golden standard and blood as the most clinically accessible sample type may miss the detection of low- level parental mosaicism. While isolated study demonstrated that LOD of Sanger sequencing could go as low as 0.25% [48], such LOD require thorough optimization of the whole Sanger sequencing procedure, and depends on the particular location of the mosaic variants. BDA-qPCR is a valuable tool in determining low-level mosaicism although the experimental conditions need to be extensively optimized and technically more demanding than Sanger sequencing and ddPCR. The robustness coupled with high sensitivity and precision of ddPCR provides a reliable alternative for the detection of low-level parental mosaicism. Based on the EBF3 assay, in-house ddPCR LOD is determined to be 0.08%-0.16%, which is comparable to manufacturer’s guideline (Figure S2). Therefore, our data demonstrated the feasibility of routine clinical analysis for parental mosaicism evaluation in families in need, using buccal and/or semen samples. With the increased accessibility and affordability of NGS technology, high depth sequencing may also help reveal more cases of parental mosaicism [3, 14]. While GS is not yet offered as the first-tier genetic testing in routine clinical setting, with increased throughput and lower costs, universal GS-based clinical genetic testing is within reach. Once this becomes routine, the chances of detecting these low-level mosaic variants through massive parallel sequencing at the current 30X read depth might diminish. Recent advancements in bioinformatics such as DeepMosaic [49] may have the potential to reveal mosaic SNVs with GS at 50X; however, an orthogonal method may still be needed to validate these potential mosaic findings [50]. Consequently, we recommend careful examination of the potential for parental mosaicism in instances where an apparent de novo variant has been identified through either routine exome sequencing or genome sequencing, particularly for families contemplating a second child. It is crucial for diagnostic laboratories to thoroughly validate their methodologies, including assessment of sensitivity, precision, and accuracy, to ensure reliable findings that can effectively inform both healthcare professionals and parents."

In conclusion, the approach to diagnosing low-level parental mosaicism should start at the genetic clinic. Attention should be given to cautious observation of any mild symptom in parents, an increased paternal age at conception or a previously affected child with PAE/candidate PAE or X-linked dominant disorders. Although paternal mosaicism is known to be common, possibility of maternal gonadal mosaicism cannot be excluded. Technically, the detection of parental mosaicism relies on using appropriate sample types, such as buccal and/or semen samples, in conjunction with sensitive methods that exceed those routinely applied in clinical diagnostics. The robust detection of parental mosaicism is crucial as it permits accurate assessment of recurrence risk during genetic counseling. This information allows families to make early, informed decisions about future pregnancies. Additionally, it enables prenatal medical teams to formulate appropriate plans for pregnancy, prenatal testing, and delivery.