Mutations in the gene encoding cartilage oligomeric matrix protein (COMP) were identified in three diseases, including pseudoachondroplasia (PSACH; OMIM 177170), multiple epiphyseal dysplasia-1 (EDM1; OMIM 132400), and carpal tunnel syndrome (CTS; OMIM 619161) (1–3). Both PSACH and EDM1 are osteochondrodysplasia inherited in an autosomal dominant manner. However, PSACH generally causes more severe short stature than EDM1, showing characteristic anterior beaking of the vertebral bodies on lateral X-rays in childhood (4, 5). Despite the differences in phenotypic severity, PSACH and EDM1 are not mutually exclusive entities but rather part of a continuous phenotypic spectrum driven by overlapping clinical manifestations that vary in frequency and severity and shared causative variants that can lead to either phenotype.

COMP is a homopentameric glycoprotein belonging to the family of thrombospondins (TSPs), also known as TSP-5, which is mainly expressed in the extracellular matrix of cartilage, ligament, tendon, and skeletal muscle (6). As shown in Figure 1A, each monomeric arm consists of an N-terminal domain (NTD), four type II (T2) epidermal growth factor-like repeats, eight type III (T3) calmodulin-like calcium-binding repeats and a globular C-terminal domain (CTD) (7). Both T3 repeats and CTD bind calcium ions which are critical to correct protein folding. T3 repeats contain five highly conserved aspartic acid residues in both N-type and C-type motifs, with two calcium-binding sites per motif. CTD is a lectin-like-β sandwich domain composed of 15 antiparallel β-strands and containing calcium binding sites. Most variants (99%) in COMP gene leading to PSACH/EDM1 occurred in T3 repeat and CTD (8). Notably, in several instances, the same pathogenic variant was reported to cause both PSACH and EDM1.

The current strategy for diagnosis of inherited diseases is to combine clinical characteristics, imaging findings and molecular genetic testing. With the increasing application of gene sequencing, the diagnostic rate of PSACH/EDM1 has been improved. Nevertheless, the overlap in both clinical phenotypic spectra and mutational profiles between PSACH and EDM1 complicates differential diagnosis. Thus, this study aimed to review all published data on PSACH/EDM1 patients up to 24 March 2025, including the COMP gene variants, clinical symptoms, radiographic features, and consequently to explore the distribution characteristics of COMP gene variants along with the potential genotype-phenotype correlation.

This study reviewed COMP gene variants related to PSACH/EDM1 from 106 previous publications, providing a comprehensive description of the types and locations of COMP gene variants in 830 patients, along with clinical information. In addition to English literature, this study incorporated literature published in Chinese (accounting for 15.3% of the included patients), thereby enabling a more comprehensive exploration of genotype-phenotype associations compared to previous study (8).

Data summarized in this review allowed the comparison of clinical data between PSACH and EDM1 with large sample size for the first time, confirming that PSACH manifested earlier age of onset and visit compared to EDM1. Consistent with the statement in previous literature (6), we confirmed that PSACH exhibited more severe clinical manifestations, including shorter stature and higher rates of lower limb deformity, gait abnormality, and joint laxity. In contrast, EDM1 showed a higher rate of joint pain/osteoarthritis. Regarding imaging features, PSACH was characterized by the presence of anterior beaking of the vertebra and a higher rate of metaphyseal involvement compared to EDM1. Additionally, EDM1 exhibited a higher prevalence of abnormal femoral head than PSACH.

COMP plays an important role in stimulating proliferation of chondrocyte, promoting collagen secretion, and enhancing mechanical strength of extracellular matrix (ECM) tissues (9–11). Thus, extensive studies have been conducted to elucidate the pathogenic mechanisms underlying COMP gene variants. Prior experimental studies have demonstrated that variants in the T3 repeat compromise calcium binding, and then prevent the protein folding correctly, resulting in endoplasmic reticulum (ER) stress by upregulating CHOP (CCAAT-enhancer-binding protein homologous protein) due to the retention of mutant COMP in the ER of chondrocyte (12, 13). Subsequent studies have further inferred that ER stress then drives oxidative stress and inflammation responses, forming a self-perpetuating stress loop (14). ER stress combined with inflammation is hypothesized to activate mTORC1 signaling, inhibiting autophagy which prevents clearance of mutant COMP (15). In addition, oxidative stress leads to necroptosis of chondrocytes by DNA damage (16). These mechanistic hypotheses suggest that premature loss of chondrocytes, impaired proliferation of chondrocytes and decreased synthesis of ECM in growth plate could be potential contributors to impaired linear growth of bone. Moreover, decreased COMP-mediated collagen secretion into ECM, insufficient ECM replenishment of the articular cartilage, and senescence-driven matrix degradation may be associated with early-onset osteoarthritis. Different from the effect of variants in T3 repeat, the variant in CTD (p.Thr583Met) induced unfolded protein and cell stress response without retention of the mutant COMP in ER, leading to reduced chondrocyte proliferation and increased apoptosis (17). Based on the pathology of PSACH, therapeutic interventions showed potential efficacy in the mouse model of PSACH (MT-COMP mice) including preventing/reducing COMP synthesis, eliminating CHOP expression, reducing inflammation and oxidative stress (18–20). Unfortunately, the only clinical trial exploring the efficacy of resveratrol in PSACH was terminated due to inability to recruit target number (NCT03866200). Thus, there remains a critical unmet clinical need for effective therapies in PSACH/EDM1.

In this review, mutational spectra of COMP gene from 471 probands demonstrated that exon 13 exhibited the highest frequency of both missense and in-frame variants. Exon 13 is 182 bp in length, second only to exon 19, which could partially account for the concentration of missense variants in exon 13. The high occurrence rate of in-frame variants in exon 13 was due to the prevalent 3 bp deletion of GAC (c.1417_1419del, p.Asp473del) located in this exon.

The most common type of variant in COMP gene in this review was missense variant, mainly located within T3 repeat region. More than half of the probands (278/471, 59.0%) harbored missense variants located in this region, providing the possibility to explore genotype-phenotype associations. In addition to the missense variants in T3-6, T3-7, and T3–8 previously summarized to be more strongly associated with PSACH (8), we found that missense variants in T3–1 were also more likely to cause PSACH due to the accumulation of newly reported PSACH cases with missense variants in T3-1. In contrast, missense variants in T3–4 and T3–5 were more strongly associated with EDM1. This genotype-phenotype statistical correlation indicated that variants in T3–4 and T3–5 may be associated with a less severe phenotype, with the underlying regulatory mechanism, requiring further functional validation. The height trend observed across T3 repeat region aligned with this inference, as patients with variants in T3–4 and T3–5 exhibiting greater height compared to those with variants in other regions in both PSACH and EDM1 shown in Figure 5.

Second only to missense variants, the next most frequent type of variants was in-frame variant, all of which located in T3 repeat region. Majority of the in-frame variants resulted in PSACH (114/133, 85.7%), indicating that in-frame variants tend to cause greater disruption to the structure of COMP protein even though without frame shift. Among them, deletion of one aspartic acid residue in the GAC repeat region (c.1405-1419) was the most common variant in COMP gene which exclusively found in PSACH, along with deletions of two or three aspartic acid residues. In contrast, duplication of one aspartic acid residue was more likely to cause EDM1, while duplication of two aspartic acid residues only caused PSACH. These statistical association findings suggested that the deletion of one aspartic acid residue in the GAC repeat region may have a profounder effect on COMP protein folding than the duplication of one aspartic acid residue, though this inference requires further confirmation through in vitro functional experiments or animal models. Furthermore, duplication of two aspartic acid residues may exert a stronger effect on COMP protein folding than duplication of one aspartic acid residue, and this hypothesis remains to be supported by direct molecular functional validation (Figure 7).

Frameshift variants, which usually result in premature stop codons, are expected to cause more severe dysfunction of protein than missense variants and in-frame variants. To date, only three frameshift variants of COMP gene have been identified in PSACH/EDM1. Interestingly, two of the three frameshift variants were reported in patients with EDM1. A novel frameshift variant in exon 13 (T3-6), c.1359del (p.Asn453Lysfs*62), was recently identified in a EDM1 family (21). The affected family members presented with early-onset osteoarthritis and low-normal height (-1.9SD ~ 0.04SD), even though some of them also had Noonan syndrome caused by variants of LZTR1 gene. Another frameshift variant, c.2223_2224insC (p.Asn742Glnfs*2) in exon 18 (CTD), was also identified in a 16-year-old girl with EDM1 and her affected mother without detailed clinical information reported (22, 23). The third frameshift variant is c.846_867 + 3del (p.Cys282Trpfs*5) in exon 8 (T3-1) reported in 6-year-old girl with PSACH presented with short stature and windswept deformity of the lower limbs, whose unaffected mother presented low-level somatic mosaicism (5%) (24). Notably, the PSACH-associated frameshift variant in exon 8 yields a severely truncated protein, whereas the EDM1-linked frameshift variants in exons 13 and 20 produce proteins that are only mildly truncated and retain most T3 repeats. This structural distinction explains their phenotypic divergence. The fact that these truncations cause EDM1, a stark contrast to the multitude of PSACH-causing missense variants, collectively suggests that the key factor dictating PSACH versus EDM1 is not variant type per se, but rather the variant’s propensity to cause core domain misfolding, secretory incapacity, and the resultant gain-of-toxicity (dominant-negative effect) within the chondrocytes.

Both PSACH and EDM1 were inherited in an autosomal dominant manner, as heterozygous variants in the COMP gene are well known causes. To date, only one study reported a biallelic COMP variant (c.1423G>A, p.Asp475Asn) in two individuals with severe PSACH in a four-generation consanguineous family, while heterozygosity for the variant in other family members (25). The phenotypes of heterozygous and homozygous individuals presented in this study comply with a gene dose dependent phenotype, and an additive effect of two mutant alleles.

A critical insight from this review is that PSACH and EDM1 constitute a continuous phenotypic spectrum rather than strictly dichotomous diseases. This spectrum is supported by two lines of evidence: first, the overlapping clinical features, such as lower limb deformity and short stature, are present in both conditions but differ in frequency and severity; second, the shared causative variants, such as c.1021_1026del, can manifest as either PSACH or EDM1. Mechanistically, this spectrum is likely driven by the degree of COMP protein dysfunction induced by different variants: variants that severely impair calcium binding, protein folding, and extracellular secretion tend to cluster at the PSACH end of the spectrum, while variants with milder effects on protein function are more commonly associated with EDM1. The phenotypic heterogeneity among patients with same variants suggests that factors beyond the variant itself, such as genetic modifiers, environmental factors, or stochastic events, may modulate the severity of the phenotype, reinforcing the concept that PSACH and EDM1 are not dichotomous but interconnected entities.

There are some limitations in this study. First, this study is a comprehensive review focused on case aggregation rather than a strict systematic review, thus formal risk of bias assessment with standard tools was not performed. This is primarily due to the heterogeneous nature of included literature (predominantly case reports/case series with incomplete clinical data and inconsistent reporting standards) and the study’s core goal of summarizing variant distribution and preliminary genotype-phenotype correlations. To mitigate potential biases, we implemented strict quality control (e.g., Sanger/NGS-verified COMP variants, exclusion of ambiguous/duplicate data, stratified data usage), but residual biases (e.g., publication bias favoring more severe phenotypes) may still exist and should be considered when interpreting results. Second, some cases diagnosed based solely on clinical manifestation and radiographic findings were excluded, which may miss a part of cases. Thirdly, the differences of severity between EDM1 and PSACH should be interpreted with caution due to the predominantly cross-sectional nature of the included studies and the age-related longitudinal progression of osteoarthritis and functional disability in EDM1. Fourthly, future studies should focus on identifying genetic modifiers or environmental factors that influence the position of individuals along the PSACH-EDM1 spectrum, as well as exploring the functional gradient of COMP variants to better explain the continuum of phenotypic severity. Finally, only published articles in English and Chinese in PubMed, CNKI, and Wanfang were searched, so the analysis might not be comprehensive.

In conclusion, PSACH and EDM1 represent a continuous phenotypic spectrum rather than strictly dichotomous entities, characterized by distinct but overlapping clinical and imaging features, that pose challenges for differential diagnosis. This review enrolling 830 patients confirmed that PSACH exhibits more severe phenotypic manifestations compared with EDM1. Genotype-phenotype correlation analysis revealed that in-frame variants in T3 repeat region are more strongly associated with PSACH, with the most common variant (p.Asp473del) being PSACH specific. Missense variants in T3–1 and T3–6 to T3–8 showed stronger associations with PSACH, whereas those in T3–4 and T3–5 were more related with EDM1. Recognizing the continuous spectrum of PSACH and EDM1, combined with the genotype-phenotype correlations identified in this study, will assist clinicians in recognizing the clinical and imaging characteristics of these rare conditions and facilitating accurate diagnosis.