Re. Review Article entitled ‘The neoplastic pathogenesis of solitary and multiple osteochondromas’

In a recent Review Article, Porter and Simpson summarize the many theories proposed over the years to explain the pathogenesis of osteochondroma and speculate about the underlying genetic changes [1]. Based on progress that we have recently made in understanding the genetics of the neoplastic pathogenesis of osteochondroma, and the multistep model towards secondary chondrosarcoma, we would like to supplement their article with some very recent data. Based on the available literature, the authors correctly conclude that a neoplastic pathogenesis for osteochondroma should be suspected. In our opinion, the most important support for a clonal origin of osteochondroma lies in the clonal karyotypic abnormalities described [2,3]. In support of this, we have recently obtained data demonstrating loss of heterozygosity (LOH) in the cartilaginous cap of 6 of 14 osteochondromas and DNA aneuploidy in 4 of 10 osteochondromas [4], strongly pointing towards a clonal, and thus neoplastic, origin for the cartilaginous tissue of osteochondroma. Although we understand the temptation of Porter and Simpson to speculate that the osteal part of osteochondroma may be considered reactive tumour stroma, there are no data from molecular or immunohistochemical studies available from the literature to support this view. We feel that further molecular genetic studies are required to con®rm their statement. To date, it is unclear whether complete inactivation of an EXT gene (according to the Knudson tumour suppressor model) is required for osteochondroma development, or whether a single EXT germline mutation acts in a dominant negative way, resulting in multiple benign osteochondromas. In the latter, inactivation of the remaining allele would be a prerequisite for malignant transformation. The authors propose ®ve speculative and, to us, unconvincing arguments for a `dominant negative' effect of the EXT tumour suppressor genes in hereditary multiple exostoses. We describe two patients with multiple osteochondromas demonstrating a germline EXT1 mutation [4]. This is combined with loss of the remaining wild-type allele in three osteochondromas, indicating that, in cartilaginous cells of the growth plate, inactivation of both copies of the EXT1 gene is required for osteochondroma formation in hereditary cases [4]. This was also suggested by others [5] and excludes a dominant negative effect of the EXT1 gene in hereditary osteochondroma. Their comparison of the `stepwise' process of oncogenesis for the osteochondroma±chondrosarcoma sequence to the colorectal adenoma±carcinoma sequence (Table II [1]) is interesting, but requires further comment. We do not believe that it is justi®ed to subdivide osteochondroma into pre-neoplastic and neoplastic categories, since there is now suf®cient evidence that at least the cartilaginous cap of osteochondroma is a true neoplasm. They correctly speculate that both copies of an EXT gene are inactivated, which indicates neoplastic growth. Furthermore, there is no indication that additional genetic alterations in osteochondroma occur, since both cytogenetic studies [2,3] and our own study [4] show alterations mainly restricted to the EXT loci, especially EXT1. We agree with the authors that in sporadic osteochondroma, somatic inactivation of both copies of an EXT gene could be expected. So far, only one somatic mutation in the EXT1 gene has been described in a sporadic chondrosarcoma [6]. We did not ®nd any somatic EXT1 cDNA alterations in eight sporadic osteochondromas and 14 sporadic chondrosarcomas. Future studies on a larger panel of tumours should reveal whether the situation is similar to the BRCA1 gene, for which only germline mutations in hereditary breast cancer are described [7]. The authors' suggestion that progression to a lowgrade chondrosarcoma is accompanied by chromosome 3q and 10q deletion may be an oversimpli®cation. We compared LOH patterns of peripheral chondrosarcomas secondary to osteochondromas with those of primary central chondrosarcomas [8,9]. Nineteen of 20 peripheral chondrosarcomas showed LOH at all loci tested (EXT genes, EXT-like genes, and at 9p21, 13q14, 17p13, and chromosome 10), while only 3 of 12 central chondrosarcomas exhibited LOH, restricted to 9p21, 10, 13q14, and 17p13. DNA ̄ow cytometry demonstrated a wide variation in the ploidy status in peripheral chondrosarcomas (DNA indices 0.56±2.01), whereas central chondrosarcomas were predominantly peridiploid. Remarkably, near-haploidy was found in peripheral chondrosarcomas, which could explain some of the high LOH percentages. Also, polyploidization of a near-haploid clone had occurred in two high-grade peripheral chondrosarcomas. In all studies reported in the literature so far, no separation has been made between central and peripheral chondrosarcoma; LOH data presented in this way should therefore be interpreted with caution [8,9]. In conclusion, it seems to be more appropriate to propose a genetic progression model for peripheral cartilaginous tumourigenesis based on recently available data. First, inactivation of both copies of the EXT1 gene in cartilaginous cells of the growth plate is required for osteochondroma formation, as demonJournal of Pathology J Pathol 2000; 190: 516±517.