Introduction

The term “Jaffe–Campanacci syndrome” (JCS) was coined in 1982 to describe the complex of multiple nonossifying fibromas (NOFs) of the long bones, giant cell granulomas of the jaw, and café-au-lait macules (CALMs) in individuals without neurofibromas; additional variable features included intellectual disability, precocious puberty, congenital blindness, and kyphoscoliosis.1,2,3 The possibility that JCS is a form of neurofibromatosis type 1 (NF1), an autosomal-dominant disorder with CALMs and deregulated tissue growth, has been suggested4 but never resolved. Nonossifying fibromas are common benign bone lesions typically observed in the second decade of life and are predominantly found in the lower extremities. Most are asymptomatic and regress spontaneously. Giant cell granulomas of the jaw are benign bone lesions observed in the second and third decade of life and are predominantly located in the mandible. They are not inflammatory in nature and are more appropriately termed aggressive (or nonaggressive) “giant cell lesions” based on established clinical and radiographic criteria.5,6 The lesions are most commonly solitary but may be multiple.7 Only 18 cases of JCS have been published in the English literature, including four cases in which an initial diagnosis of NF1 was overturned in favor of JCS (see Supplementary Table S1 online). Multiple bilateral NOFs of the long bones are a consistent feature in JCS, whereas CALMs have been present in most, but not all, reported individuals. Relatively few jaw lesions have been identified. Extraskeletal abnormalities, when present, are not consistent. Reports of JCS have appeared primarily in the orthopedic, radiology, and pathology literature. In most reports of JCS, a diagnosis of NF1 has been considered but was usually dismissed because of the absence of neurofibromas, a hallmark of NF1. However, 16 out of 18 individuals reported with JCS were younger than 18 years of age, and hence, they may not yet have developed neurofibromas (see Supplementary Table S1 online), which typically appear in early adolescence. It is not uncommon to see adolescents and young adults harboring a known germline NF1 mutation without any neurofibromas.8 There has even been one specific in-frame deletion found (c.2970_2972delAAT) that is associated with a milder phenotype lacking dermal neurofibromas.9 Many of the JCS accounts lack documentation of slit-lamp (Lisch nodules) and skin (café-au-lait macules, freckling) examinations, evaluation of family history, or consideration of the consensus criteria for the diagnosis of NF1, even for reports published after 1988 when the National Institutes of Health (NIH) criteria were established.10 To date, no consensus criteria exist to diagnose JCS, and NF1 genetic testing was pursued in only one4 of the previously published case reports. All cases were sporadic.

There are many accounts in the English literature that describe an association between NF1 and multiple NOFs, the first dating to 1948 (see Supplementary Table S2 online). To our knowledge, only one study has reported a germline mutation in NF1 in patients with NOFs.4 In the English literature, there are eight case reports in seven publications of NF1 patients with either solitary or multifocal giant cell lesions of the jaw (see Supplementary Table S3 online); in many of these reports, no imaging of the long bones was performed or reported (see Supplementary Table S3 online). Three of these studies found an NF1 mutation in blood and/or in the giant cell lesion or in the giant cell lesion–like lesion: (i) a case study of an 11-year-old girl carrying the NF1 splice mutation c.4268A>G with a giant cell lesion of the nasal bridge (but no evidence of biallelic inactivation of NF1 in the lesion, although only exon 24 of the NF1 gene was sequenced)11; (ii) a case study of a 10-year-old boy with cherubism secondary to multiple mandibular cysts with osteoclast-like giant cells and a splice mutation (IVS37+1G>A, exon 37) in NF1 in DNA from peripheral blood lymphocytes (jaw lesion DNA not sequenced),12 and (iii) a case study of a 7-year-old girl with NF1 and a recurrent giant cell lesion of her right mandible that demonstrated loss of heterozygosity for NF1 in the lesion.13 This latter case was the first with evidence of biallelic inactivation of NF1 in a giant cell lesion. Finally, there are three case reports of individuals diagnosed with NF1 (but who lack NF1 genetic testing) with both long bone and jaw lesions (see Supplementary Table S4 online).

Both Jaffe and Campanacci considered the likelihood that JCS is a form of neurofibromatosis.2,3 Solitary NOFs are common, and the incidence of multiple NOFs may be underestimated,14 raising the possibility of a simple coincidental presence of NOF in NF1 patients. Colby and Saul (2003)4 reported a case series of four individuals meeting consensus criteria for the diagnosis of NF110 who could, however, equally legitimately have been diagnosed with JCS. A partial deletion, not further specified, of NF1 was identified in one subject (patient CS4); the other three subjects had not undergone genetic testing at the time of the paper’s publication.4 To clarify the relationship of NOF, giant cell lesions, and JCS with NF1, we performed an extensive literature review, pursued further NF1 mutation analysis in previously published cases,4 reviewed all patients with solitary or multiple NOFs and/or giant cell lesions referred to the Center for Human Genetics at University Hospitals Leuven, and reviewed all patients referred for comprehensive NF1 testing to the Medical Genomics Laboratory at the University of Alabama Birmingham (UAB), due to the presence of at least one NOF and/or a suspected diagnosis of JCS. Given some similarities in pigment and bony abnormalities in NF1 and McCune–Albright syndrome, in individuals with negative NF1 mutation testing with JCS and/or NOF, we also performed sequencing of exon 8 of GNAS.15 Since Legius syndrome is only recently described and prominently features pigment abnormalities, we also performed SPRED1 sequencing.16

Materials and Methods

NF1 mutation analysis in patients described by Colby and Saul

For patients 1–3, comprehensive cDNA-based sequencing of NF1 was performed as previously described.17

Cell culture of NOFs and giant cell lesions and NF1 mutation detection

Surgically excised NOFs or giant cell lesions were digested at 37 °C overnight with collagenase (160 U/ml) and dispase (0.8 U/ml). Cells were grown to confluency in Dulbecco’s Modified Eagle Medium (Gibco, catalog no. 11995-065): Nutrient Mixture F-12 (Gibco, catalog no. 11765-054) + 10% fetal bovine serum (Atlanta Biologicals, catalog no. S11150) + penicillin + streptomycin (Gibco, catalog no. 15070-063) and harvested. NF1 mutation detection was performed essentially as previously reported.17

Retrospective review of patient samples with suspected JCS and/or NOF referred for NF1 testing to UAB Medical Genomics Laboratory

A phenotypic checklist (http://www.genetics.uab.edu/medgenomics), completed by the referring physician, accompanies each sample referred for comprehensive clinical NF1 testing to the UAB Medical Genomics Laboratory. Search for genotype–phenotype correlations using a Microsoft Access database containing deidentified phenotypic and mutation data was performed using JCS-related descriptors, including “nonossifying fibroma” (solitary or multiple), “giant cell granuloma” of the mandible (solitary or multiple), and the label “Jaffe–Campanacci syndrome.” Referring physicians were recontacted to confirm the accuracy of the previously submitted information as well as to obtain updated phenotypic information about the tested individuals, if available since the time genetic testing was performed. All NF1 mutation–negative patients (except for KUL-22) with JCS and/or NOF also underwent comprehensive SPRED1 genetic testing, as previously described.8,18 In addition, direct sequencing of GNAS1 exon 8 was performed after amplification using two pairs of primers: 8-1f: 5′-ggactctgagccctctttcc-3′, 8-1r: 5′-ggactggggtgaatgtcaag-3′ and 8-2f: 5′-gagcgatcaggtgtgcaaaa-3′, 8-2r: 5′-cagagggactggggtgaatg-3′. In addition, clinical and genetic data on five individuals diagnosed with JCS at the Center for Human Genetics from the University Hospitals Leuven (KUL) was reviewed and included. Descriptions of patients UAB/NIH-919 and KUL-1420 have been previously reported. This work was performed under the auspices of the institutional review boards of the respective participating institutions.

Results

NF1 mutation analysis in patients described by Colby and Saul

An NF1 mutation was found in each of the four individuals reported by Colby and Saul4 ( Table 1 ): a minor lesion mutation in patients CS1–CS3 and a deletion in patient CS4. The deletion in patient CS4 spanned 1.4 Mb indicative of a type 1 microdeletion.21

Table 1 Individuals with JCS (multiple cafe au lait spots and multiple nonossifying fibromas or giant cell lesions). Only patients CS1, CS2, and UAB-19 lack neurofibromas and thus have JCS as described in the original clinical reports (refs. 1 and 2). All these patients except UAB-19 meet the clinical diagnostic criteria for NF1, and all the patients except UAB-19 were found to have pathogenic NF1 mutations
Full size table

Thirteen of fourteen individuals with “classical” (multiple CALMs and multiple NOFs or giant cell lesions) JCS harbored pathogenic mutations in NF1

Table 1 lists, in addition to the four patients described by Colby and Saul4 (2003), 10 individuals with suspected JCS referred to UAB or KUL. Only individual UAB-19 did not harbor a mutation in NF1. If the definition of JCS is restricted to include patients with CALMs and NOFs or giant cell lesions and no neurofibromas (CS1, CS2, and UAB-19), as defined in the original reports,1,2 two of the three patients (CS1 and CS2) harbor a pathogenic mutation in NF1. We were not able to evaluate UAB-19 for NF1 mosaicism, which seems a likely diagnosis.

A second group of patients (UAB-12, KUL-14, and KUL-15 in Table 2 ) who did not have classical JCS but had CALMs and a single NOF of a long bone or a giant cell lesion of the jaw also met diagnostic criteria for NF1. In this group, all three (100%) had a pathogenic mutation in NF1. In summary, all patients (16/16) who met NF1 diagnostic criteria also harbored a pathogenic mutation in NF1. Five of the 16 patients (31%) with an NF1 mutation and at least one NOF or giant cell lesion had a history of pathologic fracture, in some cases requiring multiple orthopedic procedures. In the group with an NF1 mutation, we observed a male:female ratio of 11:5 with an age range of 2 months to 43 years at the time of NF1 genetic testing. Seven of the 16 individuals (42.7%) carried truncating NF1 mutations, five had splicing mutations (31.2%), two had a multiexon NF1 deletion (12.5%), one a total NF1 gene deletion (6.6%), and one (6.6%) a missense NF1 mutation. The distribution of mutations spanned the length of the NF1 gene.

Table 2 Other individuals with nonossifying fibromas of the long bones or giant cell lesions of the jaws. KUL-22 was not tested for mutations in SPRED1
Full size table

Table 2 lists the nine individuals with NOFs of the long bones or giant cell lesions of the jaws. As noted above, three (UAB-12, KUL-14, and KUL-15) harbored pathogenic mutations in NF1. The remainder had one or two NOFs (patients UAB-16, UAB-17, UAB-18, and KUL-22), a single giant cell lesion and one NOF (patient UAB-21) or multiple lytic lesions but no CALMs (patient UAB-20). Of the six patients ( Table 2 ) without an NF1 mutation, we observed a male:female ratio of 2:4 with an age range of 9–15 years at last follow-up. Two of the six patients (33%) had a history of a pathological fracture. Short clinical descriptions of these six individuals are provided in the Supplementary Data online.

Cell culture of NOFs and giant cell lesions and NF1 mutation detection

Somatic NF1 mutations were detected in two giant cell lesions (patients KUL-14 and KUL-15; Table 1 ) but not in two NOFs (patient KUL-13; Table 1 and patient KUL-22; Table 2 ). However, it is possible that none of the NF1 two-hit cells in the NOFs enriched in tissue culture and were therefore undetectable.

Discussion

JCS is a controversial entity4 that features CALMs, NOFs, and/or giant cell lesions and a significant overlap with the NF1 phenotype. The majority of patients with a JCS diagnosis have been published in the orthopedic, radiography, or pathology literature (see Supplementary Table S1 online). NF1 genetic testing has been reported in only one case of JCS.4

We found germline NF1 mutations in 13 of 14 patients with multiple NOFs and multiple CALMs, a group that could be considered to have “classical” JCS ( Table 1 ). If the diagnosis of JCS is further constricted to include only those with multiple CALMs, multiple NOFs, and an absence of neurofibromas, as has been advocated,1,2 two of the three (patients 1 and 2) harbored pathogenic NF1 mutations. The third individual (UAB-19) may be mosaic for NF1, although this was not tested. A second group of three (patients UAB-12, KUL-14, and KUL-15) with a single NOF of a long bone or a giant cell lesion of the jaw met NF1 diagnostic criteria but were not considered to have “classical” JCS; all three patients had NF1 pathogenic mutations. The finding of multiple, bilateral NOFs in an individual with a germline NF1 mutation suggests that the NOFs arise secondary to biallelic inactivation of NF1, as predicted by the Knudson hypothesis, and as previously demonstrated in a number of NF1-related tumors and CALMs.22 However, we did not find such a somatic NF1 mutation in an NOF from patient 13, who harbored a germline NF1 mutation. The detection rate of somatic NF1 mutations in NF1-associated neurofibromas is high (~84%)23; however, the detection rate of somatic NF1 mutations in rarer NF1-associated tumors is unknown, but might be lower depending on the fraction of cells carrying a second hit in the NOF.22 As only one NOF was analyzed, no firm conclusions can be made at this point. However, we identified germline and somatic truncating NF1 mutations in both giant cell lesions from patients KUL-14 and KUL-15, establishing definitive genetic proof that these are NF1-associated tumors. This observation extends the previous finding of NF1 loss of heterozygosity in a mandibular giant cell lesion in a female NF1 patient by Friedrich et al.13

The 16 patients with at least one NOF or giant cell lesion and who met NF1 diagnostic criteria harbored a pathogenic NF1 mutation. These were distributed along the entire NF1 gene: no mutational hotspots were identified.24 Five nonsense, five splice-site, two multiexon deletions, two frameshift, one type 1 NF1 gene microdeletion, and one missense mutation were detected. All variants except the but the missense mutation are expected to result in haploinsufficiency or absence of NF1 and/or flanking gene products. One splice-site mutation (patient UAB-12) was the result of a de novo, novel AluSb1 insertion, as recently described,25 and would normally escape detection using exon-by-exon gDNA-based sequencing. There are only two well-documented examples of NF1 genotype/phenotype correlation: a three base-pair deletion of one amino acid (c.2970_2972delAAT in exon 17) that is associated with an absence of neurofibromas9 and the microdeletions of the NF1 gene (and a variable number of flanking genes) which are consistently associated with a more severe phenotype, including increased neurofibroma burden, learning disabilities, facial dysmorphism, and possible increased risk of malignant peripheral nerve sheath tumors.26,27 Given the wide spectrum of NF1 mutation type observed in this study, there was no obvious genotype–phenotype correlation, although the total number of patients is small.

The six patients without NF1, SPRED1, or exon 8 of GNAS1 mutations detected in the blood are a heterogeneous group and did not meet NF1 diagnostic criteria ( Table 2 ). Patients UAB-16, UAB-17, UAB-18, UAB-21, and KUL-22 have two or fewer NOFs and/or giant cell lesions and fewer than six CALMs. Since a single NOF is common in the general population,28 as is a single CALM,29 it is plausible that some of the patients may have a coincidental cluster of minor abnormalities. Patient UAB-17 had fibular dysplasia, a rare but well-characterized abnormality.30 The familial multiple CALMs without other features of NF1 in patient 16 is unexplained, despite mutation testing of NF1 and SPRED1.16 Patients UAB-19 and UAB-20 feature a more severe phenotype with multiple lytic lesions or NOFs in one or more legs; however, only UAB-19 can be considered as “true” JCS given her multiple CALM. She may be mosaic for an NF1 mutation. In patient UAB-20, it is unlikely that the bony lesions arose from NF1 mosaicism because no NF1 mutations were detected in cells cultured from the resected NOF of the left femur. Given the presence of so many aggressive lesions, an underlying genetic factor is presumed but unidentified. Multiple lytic foci of the mandible and femur in patients with a diagnosis of JCS have been reported previously (see Supplementary Table S1 online), but no mutational data were available from these patients.

The high percentage of pathologic fractures in our cohort (patients CS1, CS2, KUL-5, UAB/NIH-9, and KUL-13) almost certainly reflects an ascertainment bias for bone cysts among the NF1 patients, and systematic screening in NF1 patients for presence of NOFs has been suggested.4 Such screening needs to be balanced with the risk of radiation exposure in the pediatric (increased risk of leukemia)31 and NF1 populations (possible increased risk of malignant peripheral nerve sheath tumor).32 However, it seems reasonable to have a lower threshold to image sites of persistent bony pain in individuals with NF1.

The JCS diagnosis requires an absence of neurofibromas and the presence multiple CALMs, the latter a classical hallmark of NF1. This is especially problematic in children because neurofibromas typically do not appear until adolescence, or later. In many of the reports describing JCS, insufficient or inadequate phenotypic data were provided, hampering evaluation of the appropriateness of the diagnosis. In addition, there is no feature unique to JCS that must be present to distinguish it from other syndromes, including NF1. Axillary freckling, a specific finding in NF1, has also been reported in some patients with JCS.1,19 Since accurate diagnosis is essential to ensure appropriate medical management and genetic counseling, we recommend the clinical algorithm outlined in Figure 1 to be followed for patients presenting with multiple NOFs and/or giant cell lesions. The algorithm is derived from our data, which supports the hypothesis that JCS is frequently a manifestation of NF1.4 It also incorporates the hypothesis that a fraction of NF1 mutation–negative patients with multiple NOFs may be mosaic for an NF1 mutation. Formal consideration of a diagnosis of NF1 (and genetic testing, as needed) should prevent the “re-diagnosis” of JCS (from NF1) and its attendant confusion and avoid potentially missed screening opportunities (e.g., for optic pathway glioma in children). We recommend that the term “Jaffe–Campanacci syndrome” only be used for the likely small percentage of patients with multiple CALMs, multiple NOFs, and/or giant cell lesions who also lack a germline NF1 mutation or who lack evidence of an other recognized, usually distinct and rare disorder associated with NOFs and giant cell lesions ( Table 3 ). In these individuals, testing of melanocytes from multiple biopsies from CALMs for first and second hits in NF1, if available, is needed to exclude a diagnosis of mosaic NF1.33,34 It has been recognized recently that cherubism-associated giant cell lesions (and thus mutations in SH3BP2) result in loss of tankyrase-mediated destruction of adapter protein 3BP2 and thus stabilization and hyperactivation of SRC, SYK, and VAV pathways.35 The association of giant cell lesions with mutations in NF1 and PTPN11 ( Table 3 ), both members of the RAS signaling pathway, may provide additional insights to the pathogenesis of these tumors.

Figure 1
figure 1

Proposed algorithm to evaluate patients with multiple nonossifying fibromas (NOFs) and/or giant cell lesions and café-au-lait macules (CALMs). The appearance of diagnostic features of neurofibromatosis type 1 (NF1) are time dependent, and thus a child who does not meet diagnostic criteria should be periodically re-evaluated. JCS, Jaffe–Campanacci syndrome; NIH, National Institutes of Health.

Full size image
Table 3 Disorders reported to be associated with multiple NOFs and giant cell lesions
Full size table

In conclusion, the majority of patients with CALMs and NOFs or giant cell lesions harbored a pathogenic germline NF1 mutation, suggesting that many JCS cases may actually have NF1. In addition, we provide the first proof of specific somatic second-hit mutations affecting NF1 in two giant cell lesions from two unrelated patients, establishing that these are NF1-associated tumors. The diagnosis of the JCS may apply to a small number of patients with multiple CALMs, multiple NOFs, and/or giant cell lesions who have undergone meticulous clinical evaluation to exclude other rare disorders associated with NOFs and giant cell lesions. In addition, comprehensive molecular analysis to exclude a germline NF1 mutation and, when possible, testing to exclude mosaic NF1 should be performed.

Disclosure

The authors declare no conflict of interest.