Brown Tumors Belong to the Spectrum of KRAS -driven Neoplasms

Twenty-six formalin-fixed paraffin-embedded (FFPE) tissue samples from patients with BT were included in the study, comprising 20 cases from the Bone Tumor Reference Center in Basel (CH) and 6 cases from the Royal National Orthopaedic Hospital in Stanmore (UK). The diagnosis was confirmed by experienced pathologists with a particular expertise in bone and soft tissue pathology. The study was approved at the University Hospital Basel, following the approval of the ethical committee for mutational analysis of samples (“Ethikkommission beider Basel” ref. 274/12). The Royal National Orthopaedic Hospital Biobank is approved by the National Research Ethics Committee of the Health Research Committee (reference: Integrated Research Application System [IRAS] project identifier: 272816). This study was approved by the National Research Ethics Committee approved UCL/UCLH Biobank Ethics Committee (project no: EC17.14). Gene panel sequencing (Ampliseq Solid Tumor DNA—Colon and Lung V2 Panel) and Sanger sequencing were carried out following routine protocols (for details, see Supplement, Supplemental Digital Content 1, http://links.lww.com/PAS/B426 ). Variant calling was performed on Ion Reporter Analysis Software. Polymorphisms were filtered against UCSC common SNP, ExAC, 1000 Genomes and 5000 Exomes databases. Detected sequence variants were evaluated for their pathogenicity based on previous literature, databases (COSMIC, ClinVar) and classified as pathogenic, variant of unknown significance, and benign.

The average age at diagnosis was 49 years (range: 18 to 82 y, median: 42 y), our study included 14 males and 12 females. The majority of tumors (68%) were found in the peripheral skeleton, while a minority was located in the axial skeleton (32%). In most patients, BT occurred in the context of secondary hyperparathyroidism due to renal insufficiency (79%), the remaining patients had a history of primary hyperparathyroidism (21%). Imaging was available in 6 cases and showed predominantly lytic lesions with an average size of 3 cm (range: 2 to 5 cm), located in the trunk, skull, and peripheral skeleton (Fig. 1 ). One patient presented with multiple lesions in both hands; however, the exact number of tumors could not be determined due to missing clinical and radiologic information. According to the clinical records, the other patients showed monostotic involvement only. Full-body imaging of the skeleton was not available for any of the patients. Curettage was performed in 8 cases (50%), and biopsy or resection in 6 and 2 cases, respectively (38%, 12%). Follow-up data were available in 6 cases, with an average follow-up time of 9 months (range: 1 to 38 mo). In 2 cases, tumor regression was documented following normalization of PTH levels. All clinicopathologic data are summarized in Table 1 . Representative imaging findings of BTs. A, Axial computed tomography scan of the skull shows a lytic lesion of the left zygomatic arch with a distinct extraosseous component. T2-weighted sagittal magnetic resonance imaging (B) and corresponding x-ray (dorsoplantar view; C) of the digit of the right hand show a geographic expansile lytic lesion with intermediate signal intensity of the middle phalanx. D, Axial computed tomography scan of the pelvis shows a purely lytic lesion of the right superior pubic ramus with cortical breakthrough. Summary of the Clinical and Pathologic Features Table with summary of the clinical data of all cases (n=16), including sex, age at diagnosis, localization, radiology, type of hyperparathyroidism, follow-up, and mutational status of KRAS , FGFR1 , and TRPV4. CT indicates computed tomography; F, female; FU, follow-up; M, male; MRI, magnetic resonance imaging; NA, not available; WT, wild-type. Histologically, all tumors had a similar appearance consisting of monomorphic spindle cells arranged in a patternless and partly storiform pattern and intermingled multinucleated giant cells. The osteoclast-like giant cells were small compared with those seen in giant cell tumor of bone, evenly distributed and contained less than ten nuclei per tissue section on average. Mitotic activity was low and there were no atypical mitotic figures; significant cellular atypia and/or tumor necrosis was absent. Hemosiderin-laden macrophages, considered responsible for the brown color of BT, were present to varying degrees as were extravasated erythrocytes. Most tumors showed fibrous septae with reactive new bone formation that incompletely subdivided the tumor into lobules. Adjacent preexisting bone was not present in the majority of cases and morphologic features of mineral and bone disorder in the context of chronic kidney disease was not observed. A representative selection of the histologic appearance is shown in Figure 2 . The tumor cell content, represented by the mononuclear cells of the specimens, was 70% on average (range: 20% to 90%). Representative histology of BTs. A–D, Typical morphology of BTs with osteoclast-like giant cells embedded in a fibrous stroma containing hemosiderin deposits and extravasated erythrocytes. Cellular atypia or mitosis are absent. Sequencing was successfully performed in 16/26 tissue samples. For the remaining cases, insufficient DNA quality prevented molecular analyses. The analyses revealed pathogenic hotspot mutations in the KRAS gene in 10/16 tumors (62%). The most common alterations were p.A146P/T/V and p.G12D mutations, which account for 70% of all KRAS mutations. Other activating mutations included p.G13C (n=1), p.K117R (n=1), p.Q22K (n=1). The allelic frequencies (AF) ranged from 0.02 to 0.27 (mean: 0.19, SD: 0.07). KRAS mutations were found in BT in both the axial (n=4) and the peripheral (n=3) skeleton. In the 3 remaining cases with KRAS mutation, the anatomic site was not available. FGFR1 and TRPV4 mutations previously detected in GCG and NOF were not identified. In 10/26 cases, sequencing analyses failed due to insufficient tissue and/or DNA preservation. It would have been valuable to compare the genetic profile of distinct tumors from the same patient but unfortunately, tissue samples for performing such a comparison was not available. All sequencing data are summarized in Table 1 as well as in Figures 3 and 4 . Overview of clinical and molecular features. OncoPrint with each column representing a patient with additional information about the tumor localization and each row representing specific genes. NA indicates not available. Lolliplot of the gene KRAS . Illustrative representation of the gene KRAS with mapping of the mutations observed along the gene.

In this study, we performed targeted sequencing in a cohort of patients with BT and found pathogenic KRAS mutations in almost two third of the cases. The RAS-MAPK pathway is a highly conserved signaling pathway involved in a wide range of cellular processes, including cell proliferation and survival. 11 Mutations in the genes involved are present in around 30% of malignant neoplasms and activating KRAS mutations have been described in a broad range of tumors, including colorectal and non–small cell lung carcinoma. 12 , 13 Interestingly, the same mutations have increasingly been identified also in non-neoplastic conditions such as endometriosis, arteriovenous malformations and even histologically normal endometrium, indicating that the sole presence of a driver mutation is not sufficient to cause cancer. 14 – 16 Indeed, inverse correlations have been observed for BRAF mutations: for instance, 80% of benign melanocytic nevi harbor this mutation, whereas it is present in only 40% of malignant melanomas. 17 The context in which the RAS-MAPK signaling pathway is activated, including the cell affected and the interaction with the microenvironment among others, therefore seems essential for its pathogenic impact. 9 Our data in BT along with similar findings in gnathic tumors described by Guimaraes et al 10 constitute strong evidence that BT should be considered KRAS -driven neoplasms and add another lesion to the broad family of tumors that are caused, or at least influenced, by constitutive RAS-MAPK signaling activation. Furthermore, our data suggest that the site of origin of BT does not impact mutational status because peripheral and axial BT show identical KRAS mutation rates. Long-standing hyperparathyroidism is an increasingly rare condition in patients with access to high standards of care. It is, therefore, difficult to assemble large sets of these rare lesions that uniformly follow a benign clinical course. 2 In addition, only 2% of patients with hyperparathyroidism develop BT. Thus, an elevated serum PTH level appears to be a mandatory yet not sufficient factor for the development of BT. In contrast, BT tend to regress spontaneously following normalization of PTH secretion. In this regard, BT share similarities with NOF as both lesions are usually self-limited and seem to be influenced by endocrine stimulation. 9 In NOF, increased estrogen levels during puberty have been hypothesized to interfere with RAS-MAPK overactivation and to terminate tumor growth. Similar observations have been made also in osteochondromas, which cease growing after skeletal maturation, most probably due to a decrease in hGH and IGF1 levels. 9 In this regard, endometriosis, another hormonal-dependent lesion with underlying driver mutations, including KRAS mutations, among others, shares similarities with BT. Indeed, in both lesions a hormonal stimulus seems to act as a “second hit” and unlock the growth potential of the mutated cells (Fig. 5 ). Such hormonal triggers might serve as the underlying mechanism also in other self-limiting tumors. In addition, since neoplasms might best be defined as a cellular proliferation with an underlying and driving genetic or epigenetic alteration, lesions traditionally considered non-neoplastic such as endometriosis might undergo a change in perception. Fibrous dysplasia (FD), a classic example of a benign tumor-like lesion, has also been revised as neoplastic disease according to the current WHO classification. 19 Schematic illustration of the proposed model for the pathogenesis of BTs. Sequential, 2-hit model leading to the development of BTs with the first hit being an activation of the MAPK pathway—insufficient on its own for neoplastic growth—followed by a second hit, a hormonal stimulus (elevation of PTH) leading to oncogenic growth. Illustration modified from Molecular Biology of the Cell by Alberts et al. 18 Adaptations are themselves works protected by copyright. So in order to publish this adaptation, authorization must be obtained both from the owner of the copyright in the original work and from the owner of copyright in the translation or adaptation. The cause of KRAS mutations in BT is unclear, but as mentioned above, KRAS mutations have also been found in normal and non-neoplastic, tumor-like tissues as well as being associated with age and smoking. 20 In 6/16 of our sequenced cases, neither KRAS , nor FGFR1 or TRPV4 mutations were found. In these, an alternative, non-mutational based activation of the RAS-MAPK might be speculated. In the mutated cases, we noticed a low AF of KRAS mutation despite high tumor cell content in the specimens. This might be explained by the fact that the neoplastic cells in BT do not represent the predominant cell population or that only a subpopulation of cells harbors the mutation. FD, a fibro-osseous lesion of bone caused by mosaic GNAS mutations, behaves in a similar manner to BT and NOF. Indeed, as the skeleton matures, FD lesions often (but not always) cease to grow and mature / develop regressive changes. As the underlying mechanism of this phenomenon, progressive apoptosis of the mutant cells has been proposed which is also reflected by a decrease in AF of the GNAS mutation over time that can ultimately drop below the detection threshold of the individual analysis method used. 21 Similarly and depending on the degree of stimulation by PTH, the number of KRAS mutated cells in BT could diminish over time, making the mutation increasingly difficult to detect. To the best of our knowledge, our study is the first to describe pathogenic hotspot KRAS mutations in BT of the peripheral and axial skeleton, suggesting that these tumors are indeed true neoplasms that require both a mutation in the KRAS gene and (obviously) a hormonal stimulus as a second hit. Particularly in tumors with self-limited growth potential, endocrine stimulation might be a so far underestimated oncogenic driver.

Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are provided in the HTML and PDF versions of this article on the journal's website, www.ajsp.com .