Revisit of multiple epiphyseal dysplasia: Ethnic difference in genotypes and comparison of radiographic features linked to the COMP and MATN3 genes^†

INTRODUCTION

The recognition of genetic heterogeneity in several skeletal dysplasias has substantially increased the number of genes which may need to be screened in order to confirm the clinical-radiographic diagnosis. Multiple epiphyseal dysplasia (MED) is the prime example of this as, to date, mutations in six different genes have been shown to cause MED; the genes encoding cartilage oligomeric matrix protein (COMP), matrilin-3 (MATN3), and the alpha 1–3 chains of type IX collagen (COL9A1, COL9A2, COL9A3), have all been shown to result in autosomal dominant MED, and a specific mutation in the diastrophic dysplasia sulfate transporter (DTDST) has been shown to be associated with an autosomal recessive form of MED [Rossi and Superti-Furga, 2001; Briggs and Chapman, 2002].

The frequencies of mutations in these genes in MED patients have yet to be clearly determined. Previous studies in European MED patients have shown a frequency of: 7–35% for COMP, 14% for DTDST, 5–10% for MATN3, and 5–15% for the type IX collagen genes [Briggs and Chapman, 2002; Jackson et al., 2004; Jakkula et al., 2005]. A comprehensive screening in a Japanese population indicated that frequencies differ in East Asian patients: 23% in MATN3, 20% in COMP, 6% in COL9A2, and none in DTDST [Itoh et al., 2006].

The purpose of this study was to determine molecular genetic characteristics in a cohort of clinically and radiographically defined MED patients. In addition, we also sought to identify radiological predictors of genotype, because mutation analysis of all six known genes in MED is a sizeable task and can be costly. We considered that this time-consuming procedure could be shortened by performing a preliminary assessment to target the molecular analysis. Thus, we hypothesized that knowledge of radiological phenotypes might be used to predict the causative gene, and that genotype-radiological phenotype correlation in MED might provide information that could guide gene analysis.

MATERIALS AND METHODS

RESULTS

COMP Mutation

Regardless of age at radiographic examination, all patients showed abnormally shaped femoral heads. Patterns of abnormally shaped capital femoral epiphyses were small and round in seven patients (30%; Fig. 1A–C), dysplastic and fragmented in three (13%; Fig. 1D), and flat and crescent shaped in 13 (57%; Fig. 1E). A round shape was seen in patients between the ages of 4 and 9 years (inclusive), whereas a flat, crescent, and/or dysplastic shape was observed in patients aged 10–28 years. The femoral neck was broad in most patients and tended to be in valgus deformity. In three adult patients, the femoral neck appeared to be short and in varus deformity (Fig. 1F). The acetabular contour was irregular and showed loss of normal concavity with a steep acetabular angle and marginal spurs (Fig. 1A,B,D). One interesting finding was of an irregular contour of the iliac crest and ischium, which was noted in six patients (26%) between the ages of 6–12 years (Fig. 1D).

All patients had flat and irregularly shaped epiphyses of the distal femur and proximal tibia (Fig. 2). Because distal femoral epiphyses were flat, femoral intercondylar notches became shallow, which was the major finding in subjects aged between 11 and 28 years (Fig. 2B,D). Fragmented and cracked epiphyses were observed in the distal femur in eight patients (35%) and proximal tibia in four (17%). At the knees, thin, irregular contours of the medial and lateral borders of the proximal tibial epiphysis and a medial beak at the metaphysis were the most consistent findings (61%; Fig. 2F). Lateral knee radiographs were available in 16 patients, and of these, 11 (69%) had small and thin patella. One patient had a double-layered patella (Fig. 2G).

Hand radiographs were available for 23 patients. Carpal bones were small and had ragged contours in 19 (83%), 14 (61%) had an irregular and flat distal radial/ulnar epiphyses, 13 (57%) had brachydactyly, and 12 (52%) had a delayed carpal bone age (Fig. 3). Ankle radiographs, including feet or not, were available for 22 patients. The most common finding was a thin and irregularly shaped distal tibial epiphysis in 20 (91%). A ragged contour of tarsal bones was also observed in 15 patients (68%; Fig. 3B).

Twenty-two of the 23 patients had lateral spine radiographs. Vertebral body heights were usually normal though mild flattening was present in two patients. However, the end plates of vertebral bodies were irregular in six (27%) patients. Patients aged between 4 and 6 years showed convex end plates and a rounded vertebral body appearance.

MATN3 Mutation

The capital femoral epiphyses were flat and crescent-shaped in 23 patients (77%; Fig. 4A,D,E) with a MATN3 mutation, asymmetrically or asymmetrically flattened with a fragmented appearance in six (20%; Fig. 4B,C), and non-ossified in one (3%; Fig. 4F). In contrast to the COMP mutation, no patient aged <9 years had a small, round femoral head. Most of the patients had preserved normal concavity of the acetabular roof. An irregular acetabular contour was present in nine (29%; Fig. 4E).

The epiphyses of the distal femur and proximal tibia were flat and irregular, but thinning of the epiphyses in the distal femur was more severe at the lateral side of the epiphyses, whereas, the proximal tibial epiphyses thinned out at both sides, to produce a triangular shape or a shape resembling a harlequin's hat (Fig. 5A,B,E,G). However, unlike COMP mutations, no patient showed a fragmented or cracked epiphyses. Vertical metaphyseal striations in the distal femur and proximal tibia were observed in 26 of 30 patients (87%). One of these patients underwent MRI of the knee and also showed longitudinal striations (Fig. 5E,F). Lateral knee radiographs were available for 18 out of 30 patients and the patellae were found to be thin and to have irregular contours.

Hand radiographs were available for 21 patients. Delayed carpal bone age was observed in five (24%). In contrast to the COMP mutation for which carpal bones had a ragged contour, the carpal bones of the MATN3 patients were smooth. Brachydactyly was observed in two patients. The epiphyses of distal tibiae were flat and irregular in shape and concordant metaphyseal irregularity was noted (Fig. 5H). Lateral spine radiographs in 23 patients showed mild platyspondyly and end-plate irregularities in approximately one third (Fig. 6).

DISCUSSION

MED is genetically heterogeneous and this makes molecular diagnosis difficult. Overall detection rates for MED mutations in European surveys range from 25% to 50% [Briggs and Chapman, 2002; Jakkula et al., 2005; Kennedy et al., 2005]. These low detection rates may be due in part to misdiagnosis, and in this context, preselection of MED cases by expert clinical and radiological review could enhance the mutation detection rate to 81% [Zankl et al., 2007]. In the present study, preselection of MED patients for mutation analysis by a radiological review panel resulted in a detection rate of 87%. Accordingly, the results of these two studies indicate that careful subject preselection can increase mutation detection rates appreciably.

The reported frequencies of the six known genes in MED patients vary considerably, probably because of selection bias. The contribution of the COMP locus to MED is 30–50%, followed by DTDST at about 25%, and the remainder is split between MATN3 and COL9A1,2 and 3 in European populations [Jakkula et al., 2005; Unger et al., 2008], but another study reported that the DTDST mutation was most common in molecularly defined MED [Ballhausen et al., 2003]. However, frequencies found in survey conducted in East Asia differ from those found in Western surveys. Comprehensive screening for MED mutations in the Japanese population showed that mutations in MATN3 were the most frequent (23%), followed by COMP (20%), and type 9 Collagen genes (9%), whereas no DTDST mutations were found. Our study demonstrates that the mutation spectrum of MED in Koreans is similar that reported in the Japanese survey, that is, the MATN3 mutations were most prevalent, and was present in 45% of our cohort, and the frequency of the COMP mutation was 35%, which is similar to that reported in Europeans. However, the frequency of mutations in COL9A1-3 and DTDST were very low and identified only in one patient each. These findings regarding the nature of the mutation spectrum in Korean MED patients suggest that a different molecular screening strategy should be applied in the population-based survey.

This study reveals MATN3: c.361C > T is the most common mutation in Korean MED patients, and that it accounts for approximately 67% (20/30) of patients harboring MATN3 mutations. The c.361C > T (p.R121W) mutation was the most frequently detected MATN3 mutation in a Japanese study as well, but it was found in 38% (three of eight patients) [Mabuchi et al., 2004]. The high prevalence of this specific mutation may be the main reason for the observed ethnic differences between the mutation spectra of East Asians and Europeans.

Degrees of severity of radiographic manifestations of affected joints within same mutation groups were not identical because ages at radiographic examinations differed. We observed three patterns of a misshaped femoral head for COMP mutations, that is, a small and round shape, a flat and crescent-like shape, and a dysplastic and/or fragmented shape. Of these, a round shape of the capital femoral epiphysis, referred to as a “mini-epiphysis,” has been reported to be a characteristic finding of COMP mutations [Unger et al., 2001, 2008; Lachman et al., 2005]. This finding was observed in 39% of patients with COMP mutations, but was confined to patients aged 4–9 years. A small and rounded femoral head seems to become crescent-shaped or dysplastic and/or fragmented appearance with age, but this suggestion is limited by the fact that the radiographs of patients with different ages were compared, and not sequential radiographs of same patients. A larger study with follow-up radiographs in same patients is required to explore this suggestion. On the other hand, patients with MATN3 mutations, even those aged <9 years, did not have a small and round femoral head. Furthermore, the majority of patients (97%) with MATN3 mutations had either symmetrical or asymmetrical crescent shaped femoral head. This observation suggests that a small and round femoral head is observed in patients suspected to have MED, the COMP mutation should be the first gene examined.

When a femoral head has a symmetrical or asymmetrical crescent shape with fragmentation, radiographic abnormalities may overlap with bilateral Legg-Calve-Perthes disease (LCPD). In fact, some of our patients were initially suspected to have bilateral LCPD or Meyer disease (Fig. 4B,C). However, careful clinical examination and radiographic evaluations of knee and ankle joints can exclude bilateral LCPD. In our skeletal survey of 55 patients, non-ossified capital femoral epiphysis was observed in only one patient who had a MATN3 mutation (Fig. 4F); however, the radiographic examination was performed at 2 years of age, and absence of ossification may be due to delayed in appearance for ages.

Acetabular changes were a distinctive feature in the hip joints of patients with COMP mutations. In a previous study, an irregular acetabular contour was described as radiographic change associated with the presence of COMP mutation [Lachman et al., 2005]. We analyzed acetabular contours in more detail. Straight and shallow acetabular roofs, indicating loss of normal concavity, were observed in the 19 out of 23 patients (83%) with COMP mutations. In addition, marginal spikes of the acetabulum were noted in 65%, which may be the consequence of acetabular irregularities. Acetabular contour irregularities and shallowness were not observed in three adult patients, and thus acetabular changes seem to be age-dependent alteration. In contrast to the COMP mutations, acetabular irregularities were less frequent (29%) in patients with MATN3 mutations. Acetabular contours also should be carefully evaluated in the hip radiographs of MED patients, as this could provide additional indications of genotype.

Knee joints consistently showed dysplastic epiphyses and metaphyses of the distal femur and proximal tibia, regardless of genotype. A fragmented or cracked distal femoral epiphysis was observed in 35% of patients with COMP mutations, and this was not observed in any patient with MATN3 mutations. Unger et al. [2008] described a cracked distal femoral epiphysis as the “glacier crevice” sign, and found this helpful for diagnosing the presence of COMP mutation. In the present study, a cracked epiphysis was primarily seen in the distal femur, but the proximal tibial epiphysis was also affected in four patients. A flat distal femoral epiphysis tended to result in a flat intercondylar eminence and shallow intercondylar notch with age. This finding was exclusively observed in three adult patients and frequently seen in COMP mutations around the age of physeal closure, but also observed in patients with MATN3 mutations.

Makitie et al. [2004] described vertical striations in their report on MATN3 mutations in 12 patients. In the present study, this finding was observed in 26 of 30 patients (87%) with MATN3 mutation, and in 13% (three of 23) patients with COMP mutation. Our study shows that when vertical striations are present in the distal femoral and/or proximal tibial metaphyses with distinct epiphyseal changes, the probability of MATN3 mutation increases.

Previous studies have reported that a double-layered patella is found exclusively in patients with the DTDST mutation, and thus, suggested that this feature be considered pathognomonic [Makitie et al., 2003; Superti-Furga et al., 1999]. However, other studies have reported a double-layered patella in a patient with COL9A2 mutation and in a pseudoachondroplasia patient with COMP mutation [Nakashima et al., 2005; Vatanavicharn et al., 2008]. In the present study, one patient with COMP mutation had a double-layered patella. Thus, when a double-layered patella is found, diagnostic sensitivity for the DTDST mutation should be considered high, but not exclusive.

Radiographic evaluations of the hands in MED patients should focus on bone age and the contours of carpal bones. A delayed carpal bone age was evident in around half of the patients with COMP mutation and in 23% of patients with MATN3 mutation. Ragged and irregular contours of carpal bones are predominated (83%) in patients with COMP mutations, whereas this finding was never observed in MATN3 mutations. The presence of brachydactyly and small, round metacarpal epiphyses are frequently associated with COMP mutations [Unger et al., 2001; Lachman et al., 2005]. Over half of the patients (57%) with COMP mutation in our series had brachydactyly, whereas two patients (9%) with MATN3 mutation had radiographic evidence of brachydactyly.

The term “multiple epiphyseal dysplasia” was coined at a time when spinal involvement was not recognized. However, experience shows that some patients have accompanying mild spinal dysplasia, such as, endplate irregularities and/or a slight decrease in vertebral body height [Mabuchi et al., 2004]. In the present study, endplate irregularities with normal vertebral body heights or mild platyspondyly were observed in 55% of patients with the COMP mutation and in 35% with the MATN3 mutation. However, oval-shaped vertebral bodies (superior and inferior endplate rounding) were also observed in patients aged between 4 and 7 years with the COMP mutation, and thus, it is evident that some spinal alteration may accompany in patients with COMP and MATN3 mutations.

With next generation sequencing techniques, multi-gene sequencing will become less cumbersome but molecular interpretation will become more complicated. Thus, detailed evaluation of radiological findings will enhance suspicion of the presence of particular gene mutations in MED patients and facilitate the planning of genetic studies.