Background
Enchondroma (EC), a benign cartilage forming tumor in the medulla of the bone, is thought to be a precursor of secondary central chondrosarcoma (CS). EC develops either as a single, solitary lesion or as multiple lesions in the context of Ollier disease [
1]. Ollier disease is the most common subtype of enchondromatosis and shows multiple ECs with marked unilateral predominance [
1,
2]. The risk of malignant transformation towards central CS is up to 35% in Ollier disease [
1,
3]. There is no marker that would indicate progression towards malignancy, thus there is a vital need to understand the genetics of these tumors which may help to develop markers for early diagnosis. A comprehensive understanding of the molecular events in ECs and central CS also enables the identification of possible targets for treatment [
4].
While the genetics of enchondroma is poorly understood, the involvement of the EXT genes is well established in the development of solitary as well as hereditary multiple osteochondromas (MO) (OMIM 133700), benign cartilage tumors at the surface of bone [
5]. The lack of EXT function seems to disturb hedgehog signalling in MO, while activated hedgehog signalling in mice seems to underlie the development of the Ollier related phenotype [
4]. Heterozygous mutations in
PTH1R are found in a small subset of Ollier patients [
6‐
8]. It is however unclear whether these mutations in
PTH1R are causing or modifying the disease [
7], and since ~90% of Ollier patients lack
PTH1R mutations, we aimed to study Ollier related ECs by mapping genetic changes using genomic arrays.
We hypothesized Ollier disease to be a germ-line mosaic condition due to the fact that it is non-hereditary and because of its unilateral predominance feature [
3,
9]. An early postzygotic mutation resulting in asymmetric involvement of skeletal structures can be expected, as was also shown for polyostotic fibrous dysplasia [
10]. One could speculate that an inactivating mutation in a tumor suppressor gene, similar to EXTs in osteochondroma, would have occurred in the developing mesoderm early after gastrulation. In case of a tumor suppressor gene, later on, an additional hit may be required for the formation of ECs with subsequent genetic changes causing progression towards central CS. We tested this hypothesis using high-resolution SNP array combined with expression array on DNA derived from tumor tissue and, whenever available, normal DNA from Ollier patients.
SNP arrays provide an excellent possibility for large-scale, genome-wide, high-resolution analysis of both DNA copy number alterations (CNA) and loss of heterozygosity (LOH) in cancer cells. It also provides a feasible means of detecting genotyping alterations in the tumors of individual patients and, in principle enables the identification of new areas with common allelic imbalance that could harbor potential tumor suppressor genes which helps in the identification of novel candidate genes affected by genomic abnormalities [
11,
12]. In the present study, we used Affymetrix Genome-Wide Human SNP Array 6.0 to obtain a comprehensive registry of genetic aberrations in 37 tumors of 28 patients with Ollier disease and correlate it with gene expression using Illumina Human-6 v3 Expression Array and qRT-PCR and protein expression using tissue microarray (TMA). Based on the obtained genomic profiles with limited and non-recurrent genetic changes in Ollier ECs, we conclude that they are genetically heterogeneous and that the reported CNA in this study are likely to be secondary random events in ECs.
Discussion
The origin of both solitary and Ollier related ECs is largely unknown. To address this, we performed genome-wide analysis of ECs occurring in non-hereditary Ollier disease. Since these tumors are polyostotic, with a unilateral predominance, manifesting early in life, we postulated that there may be a germ-line mosaicism. We attempted to find causative genes for ECs of the Ollier disease using a high-resolution SNP array containing 1.8 million markers combined with expression array and obtained comprehensive genetic profiles of 37 Ollier disease related tumors. This is the first and largest genome-wide molecular study on Ollier disease reported so far, which was possible through the collaboration of many different institutes within the EuroBoNeT Network and the European Musculo-Skeletal Oncology Society (EMSOS).
In general, the obtained genomic profiles showed absence of large genetic aberrations in Ollier ECs except loss of chromosome 6 in two ECs from two unrelated Ollier patients. Small non-recurrent genetic changes combining the SNP and expression array at
FAM86D,
PRKG1,
ANKS1B, NIPBL and
POUF51 were found in ECs. Most of these genes are not known to play an important role in cartilage formation. We found homozygous loss of
FAM86D in two ECs of the same patient. Function of
FAM86D is still unknown. We confirmed intronic gain at
PRKG1 in one EC, which is involved in fatty acid metabolism [
33]. We found loss at
ANKS1B, while overexpression of
ANKS1B is reported in pre-B cell acute lymphocytic leukaemia [
34]. Gain at
NIPBL was found in L2205 while at the protein level only 30% of Ollier ECs and CS expressed NIPBL. Inactivating mutations in
NIPBL are associated with Cornelia de Lange syndrome and one of the characteristic features of this syndrome is reduction in limb growth (OMIM 122470). Loss of
POU5F1 was found in two ECs with monosomy at chromosome 6 and its mRNA and protein expression was absent in all Ollier and solitary ECs and CS. The transcription factor
POU5F1 (OCT3/4) is involved in regulating pluripotency and is normally expressed during early embryogenesis in embryonic stem and germ cells [
35].
Here we studied extensively candidate genes obtained from paired analysis of enchondromas. Normal DNA was available from 4 Ollier patients enabling paired analysis. Despite the low number of paired samples our data suggest that no common CNA are associated with EC development. Extending the analysis with the unpaired samples we could not see any common CNA in all ECs. All aberrations we obtained in at least 5 out of 14 ECs are reported as common copy number variants in DGV database (
http://projects.tcag.ca/variation/). Loss of chromosome 6 was the only recurrent change in ECs of two unrelated Ollier patients. Therefore, relatively small numbers of copy number alterations that we found per ECs are more likely to be secondary random genetic changes. Although SNP array technology is a powerful analysis tool, it can not detect balanced chromosomal translocations, inversions and whole-genome ploidy changes. However, balanced translocations and inversions have not been reported for the Ollier tumors in the literature so far [
2].
Previously,
PTH1R was reported to be the gene causing Ollier disease [
6]. However, in subsequent studies it was shown that
PTH1R is only mutated in a sub group of patients (~10%) decreasing receptor function to ~70%, suggesting that it may contribute to the disease but is probably not causative [
7]. PTH1R is a key player within the IHH pathway which is crucial for endochondral ossification. The presence of known point mutations (R150C, R255H, G121E and A122T) in
PTH1R was excluded in the present series (unpublished data). Also, we could not find a deletion or LOH at the 3p21.31 region harboring the
PTH1R gene. Recently, inactivating mutations in
PTPN11 are reported in another enchondromatosis subtype called metachondromatosis in which multiple ECs are combined with osteochondroma-like lesions [
2,
36]. In our series we could not detect any CNA or LOH at
PTPN11 by SNP array. Also, expression of
PTPN11 was not decreased in ECs as compared to controls in expression array.
Large gains, losses and LOH were seen more often in Ollier CS as compared to ECs, which is in line with increased genetic instability and aneuploidy seen in solitary central chondrosarcoma progression. Most common CNA involve 3p, 5q, 6q and 9p in Ollier CS. However, we could not detect recurrent CNA in all Ollier CS that could have been responsible for malignant transformation of ECs. For Ollier CS, a deletion at the short arm of chromosome 1 [
37], LOH at
RB1 at chromosome 13 and p15/p16 loci at the short arm of chromosome 9 [
38] were reported in single cases each. Our results show very few copy number alterations in ECs and an increased number of variable genomic alterations in CS. This is in support of the multistep model for CS development [
39].
In conclusion, we present genome-wide copy number and expression profiles of the largest international series of Ollier ECs and CS reported so far. We show absence of recurrent CNA and LOH in majority of ECs suggesting that instead point mutations or other copy number neutral structural changes (inversions, insertions, balanced translocations) or deletions below the resolution of this platform involving a single or a few exons only [
40] might have an important role in EC pathogenesis. This opens the possibility to study these tumors further using a next generation sequencing approach. An increased number of genetic alterations are found in Ollier CS, supporting the multistep genetic progression model.
Acknowledgements
Our work on Ollier disease is supported by The Netherlands Organization for Scientific Research (917-76-315) and is performed within EuroBoNet, a European Commission granted Network of Excellence for studying the pathology and genetics of bone tumors (018814). The authors thank B. van den Akker, D. de Jong, H. Baelde, R. Duim, M. van Ruler, I. H. Briaire-de Bruijn, M. Winter and M. L. Kuijjer for their expert technical assistance. We are thankful to Dr. S. Daugaard, University of Copenhagen, Denmark, Dr. S. Boeuf, Heidelberg University, Germany, Prof. F. Mertens, Lund University Hospital, Sweden, Dr. L. Kindblom, Royal Orthopedic Hospital, United Kingdom, Dr. R. Forsyth, Ghent University, Belgium, Dr. P. Jutte, UMCG, The Netherlands, S. Bos, LUMC, The Netherlands, Dr. P. Mainil-Varlet, Bern University, Switzerland, Dr. B. Toker, Istanbul University Medical School, Turkey, Dr. B. Liegl-Atzwanger, University Clinic of Orthopaedic Surgery and Medical University of Graz, Dr. M. San-Julian, University of Navarra, Spain for providing samples. The continuous support of the Netherlands Committee on Bone Tumors is greatly acknowledged. Antoinet C.J. Gijsbers, LUMC and the team of LGTC (
http://www.lgtc.nl), Leiden, The Netherlands for support of the array experiments.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
The study was designed by TCP, JVMGB, PCWH, JO and KS. Data analysis was done by TCP and JO. Tissue microarray was constructed by TCP and TK. RS, LS, AHMT, SHMV were responsible for acquisition of patient material and patient data. The manuscript was written and approved by all the coauthors.