Osteogenesis imperfecta (OI) is a generalized connective tissue disorder, apart from liability to slight trauma fractures, the phenotypes usually extend to non-skeletal symptoms, like blue sclera, dentinogenesis imperfecta, hearing loss, and ligamentous laxity (1). Based on the clinical and radiographic manifestations, Sillence et al. (2) in 1979 proposed the classic classification of OI: type I-IV: type I is the mildest kind characterized by nondeforming with persistently blue sclerae; type II is the perinatal lethal form; type III is the severest among surviving patients and presents with progressively deforming; type IV is characterized by white sclerae, whose severity is between type I and type III.

Although an increasing number of genes have been identified in OI involving different modes of inheritance (3), about 85%-90% of OI cases are caused by autosome dominant mutations of COL1A1 and COL1A2 genes. They encode the α1(I) and α2(I) chains of type I collagen, which is the main component of the extracellular matrix of bone and skin and is responsible for their elastic properties (4). Mutations in COL1A1 and COL1A2 of collagen-related OI can be divided into two groups: quantitative defect (haploinsufficiency) with synthesizing about half the amount of structurally normal type I collagen, primarily resulting from nonsense, frameshift, and some splice-site variants; another is the structural defect, in which over 80% mutations are the results of glycine substitutions in the helical domain (4). The majority of the type II-IV are the consequences of a structural defect because it has more severe impacts on the extracellular matrix than haploinsufficiency does.

Parental genetic characteristics will be passed on to the next generation. At the same time, de novo mutations, a kind of novel genetic change in a family, will occur during the formation of the gametes or postzygotic in each of us (5, 6). Only mutations present in the germ cells can be passed on to the next generation (7). Distinct from inherited variants, de novo mutations haven’t undergone purifying selection during the reproductive phase, thus making them more harmful (8). Approximately 80% of all de novo germline point mutations occur on the paternal allele because oocytes experience only one additional round of DNA replication during their evolution into a mature ovum, whereas spermatogonia will encounter hundreds of DNA replication and cell fission before becoming sperm cells. What’s more, it has been demonstrated that most types of de novo mutations are much more correlated with increased paternal age at birth (9–11). Rare sporadic diseases in the population are associated with de novo mutations (12). In OI, nearly half of the cases are derived from de novo mutations (13–15).

Previous studies have elucidated the genotype-phenotype correlation and mutation spectrum of collagen-related OI through large cohorts (15–18), which contributes to achieving a more accurate and economical diagnosis of OI. However, in addition to some reports of de novo mutation cases in OI (16, 19), few large cohort studies have compared de novo with inherited mutations in collagen-related OI (14). In this study, we enrolled 135 Chinese probands of OI with de novo or inherited COL1A1/COL1A2 mutations in the department of Osteoporosis and Bone Disease of Shanghai Jiao Tong University Affiliated Sixth People’s Hospital, and then compared the gene mutation spectrum and phenotypic characteristics to investigate differences between de novo and inherited mutations among OI with COL1A1/COL1A2 mutations.

Altogether, 135 probands from 135 unrelated and non-consanguineous pedigrees, including 118 patients who had been published in our previous studies (15, 27, 28), were enrolled in the study, among which 51 probands (37.78%) belonged to the de novo mutation group, and the remaining 84 probands (62.22%) were from the inherited mutation group. The details of clinical characteristics of each proband, COL1A1 and COL1A2 mutation loci identified in the present study and the parental gene verification are listed in Supplementary Tables 1 – 4 .

Previous studies had illustrated the proportion of de novo mutations in OI with COL1A1/COL1A2 mutations: 41.03% in South Korea (13), 56.16% in Estonia, Ukraine, and Vietnam (14), and 54.6% in our previous report (15). The percentage of de novo mutations in our study (37.78%) was significantly lower than our last report (P = 0.0029), which may be due to we excluded cases without performing parental gene verification. OI is a phenotypically heterogeneous disease that familial cases with the same mutation can present remarkably phenotypic diversity, ranging from mild to severe (31, 32). In our study, the rate of nonpenetrance of hereditary probands’ parents was up to 25.64%. Therefore, some affected parents of hereditary probands exhibited very mild or even without symptoms, which may result in missed diagnosis and increasing the false-positive rate of de novo mutations if not performing parental gene verification.

Low bone mass is always presented in patients with OI, and in our study, the de novo mutation group exhibited a lower BMD Z/T-score at the lumbar spine, however, not significantly lower than the inherited group. Fractures of long bones and vertebral compression are the most notable features of OI, while there was no difference in the fracture frequency between de novo and inherited mutation groups in our study. Skeletal deformities like bowing of long bones and scoliosis are usually associated with severe type III. Apart from congenital causes, recurrent fractures of extremities can also result in deformities. As displayed in Figure 2 , lower limb deformities would seriously damage the ability to move, and most of them had no alternative but to be in a wheelchair for the rest of their lives. Most of our patients’ severe lower limb deformities were accompanied by severe scoliosis. Activity limitations and participation restrictions are linked to lower quality of life (QOL) (33), and physical QOL is worse in children with more severe OI than in mildly affected children. In an aspect of non-skeletal manifestation, blue sclera was the most prevalent feature of patients with OI (28), and consistent with the previous studies (34, 35), dentinogenesis imperfecta covered a higher proportion in type III and IV OI than type I in our study. Biochemical indexes including bone turnover markers showed no statistical differences in our study. Several papers have revealed that there were no significant differences in levels of the bone turnover markers between different OI clinical types or different OI mutation types (36, 37), therefore, the relationship between severity and bone turnover rates in OI may be more complicated.

Generally, haploinsufficiency of COL1A1 results in the mildest form of OI. In contrast, haploinsufficiency of COL1A2 results in no apparent OI-related phenotype, and homozygous null mutations of COL1A2 bring about phenotypes of varying severity (38). However, in our study, one case (p61, c.1557+3A>G) with haploinsufficiency of COL1A2 mutation showed severe phenotypes and was classified as type III. Most of the structural defects are missense mutations, in which over 80% of mutations are the results of glycine substitutions in the Gly-Xaa-Yaa repeat (4). The most frequent cause of OI with COL1A2 mutation is the structural defect of type I collagen, especially glycine substitutions in the helical domain (16, 39). Consistently, both de novo and inherited COL1A2 mutations in our study were predominantly composed of missense mutations (100% and 92.59%, respectively), especially glycine substitutions. It has been widely demonstrated that structural mutations have more severe clinical consequences than quantitative mutations, and clinical type II-IV are usually caused by mutations altering structures (1, 4). Similarly, clinical type III was mainly driven by missense mutations and type I primarily resulted from quantitative defects in our study, both in de novo and inherited mutation groups. Repeated mutations are often associated with CpG dinucleotides, described as mutational hotspots (40). The mutation of p.Gly337Ser in COL1A2 was detected in four unrelated families in our study, that is, mutational hot spots. Four different mutation loci mentioned above were shared in de novo and inherited mutation groups. Almost all of them had a consistent clinical type and clinical score between the two groups, indicating that the severity of the same mutant locus did not differ between de novo and inherited mutations.

Our results revealed that the proportion of type III and clinical scores were significantly higher in the de novo mutation group. De novo mutations are genetically different from inherited mutations and are more damaging because mutation processes in de novo mutations are happening between generations without undergoing purifying selection (41). Zhytnik et al. (14) revealed that in collagen-related OI, missense mutations and type III occupied a larger proportion in de novo mutations, and type I was lower compared with inherited mutations. In our study, the missense mutations in the de novo COL1A1 mutation group covered a higher proportion than in the inherited COL1A1 mutation group, though lacking significance, which may be a reason why there were more type III cases and higher clinical scores in the de novo mutation group. Although there was no difference in the mutation spectrum between the de novo and inherited COL1A2 mutation groups, type III cases and clinical scores were still higher in de novo mutations than in inherited COL1A2 mutations, probably due to mutation loci in de novo mutations with more serious pathogenicity since they have not undergone purifying selection. Additionally, these significant results confirmed that our clinical scoring system could quantitatively reflect the clinical severity of OI. Errors in DNA replication, exposure to mutagens, or failure to repair DNA damage can result in de novo mutations (12). Nowadays, advanced paternal age at conception has been considered the predominant factor associated with an increasing number of de novo mutations (9, 11, 42, 43). However, in our study, the average paternal age at conception was 29.2 years, significantly younger than the data of the US (29). Thus, paternal age may not be a critical factor contributing to de novo mutations in our research. In addition, Wong et al. (44) reported that medically assisted conception would increase the incidence of de novo mutations compared with natural conception. However, recently, a pilot study by Smits et al. (11) showed no impact of assisted reproductive technologies on the number of de novo mutations in the genome of the offspring. In our study, the majority of the pedigrees (98.04%) with de novo mutations were spontaneous conception, which limited us to investigate the impact of different methods of conception on de novo mutations.

Mosaicism means an individual develops from a single fertilized egg but has two or more genetically distinct cell populations (45). Postzygotic mutations, generated in the first few cell divisions after fertilization, can result in mosaicism and exist in many different tissues (43). It has been estimated that approximately 7% of de novo mutations are early postzygotic events with high-level mosaicism in the blood (46–48). The timing of the postzygotic mutations is of vital importance as it determines the proportion of affected tissues and phenotypes (45). The recurrence risk is very low, the same as the population risk, in parents having a child with a postzygotic mutation (45). Some de novo mutations, but actually because of the parental germline mosaicism, can be shared by more than one sibling. The higher the proportion of mosaicism in parental blood cells, the higher the recurrence risk among siblings (43). Furthermore, some mutations that arise in the germ cell lineage during early embryonic development are undetectable in somatic cells like blood or buccal mucosa, but can also be passed on to the next generation (7). Halvorsen et al. (49) revealed that germline mosaic mutations, which can be transmitted, have not had to undergo purifying selection as de novo mutations. Pyott et al. (50) found that the recurrence of lethal OI usually results from parental mosaicism. Therefore, de novo mutations actually resulting from paternal germline mosaicism have meaningful necessities for genetic counseling about recurrence risks. The detection limit of Sanger sequencing is about 15-20% of somatic mosaic variants. Therefore, deep sequencing, like SNaPshot minisequencing assay or whole-exome sequencing, is recommended to determine the percentage of mosaicism.

Although our comparative study has carried out a meticulous comparison between de novo and inherited COL1A1/COL1A2 mutations of OI, there are still some limitations to the study that must be considered: (1) This study was a retrospective review and suffers from the limitations inherent in such a design. (2) By adhering strictly to the inclusion criteria, we removed over 36% of probands in our institution who were confirmed with COL1A1/COL1A2 pathogenic mutations, which may have resulted in differences in the prevalence of de novo mutations we report when compared with other studies, and influence the clinical and genetical differences in two groups. However, our primary purpose was to compare the difference between two mutational patterns in OI, thus we must be 100% confident with the mode of inheritance of each proband in our study cohort. Therefore, we consider this as less of a limitation and more of a strength of the study. (3) We have not adopted deep sequencing in parental samples to make it clear whether some de novo mutations were actually caused by parental germinal mosaicism. (4) The research about non-skeletal phenotypes is not comprehensive in our study, for instance, cardiovascular abnormalities.

Our study showed that the proportion of clinical type III and clinical scores were significantly higher in de novo mutations compared with inherited COL1A1 mutations, which may demonstrate that the pathogenicity of de novo mutations is, on the whole, more severe than inherited mutations because de novo variants have not undergone purifying selection. These findings highlight the further need to investigate the differences between de novo and inherited mutations in OI and to research the occurrence of germline mosaicism in the parents of de novo mutations, which can provide accurate genetic counseling for both early prenatal and preimplantation diagnosis.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by The Ethics Committee of Shanghai Jiao Tong University Affiliated Sixth People’s Hospital. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin. Written informed consent was obtained from the individual(s), and minor(s)’ legal guardian/next of kin, for the publication of any potentially identifiable images or data included in this article.

ZZ, and HZ designed the research, and revised the manuscript. YM summarized the clinical data, analyzed the sequencing data, and drafted the manuscript. HZ contributed to funding acquisition. All authors contributed to the article and approved the submitted version.

This research was funded by the National Natural Science Foundation of China (81870618).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.