Germline BHD-Mutation Spectrum and Phenotype Analysis of a Large Cohort of Families with Birt-Hogg-Dubé Syndrome

Patient Recruitment

We recruited members of 61 BHD-affected families to our study at the NCI, from 1998 to 2001, through patient recruitment letters to dermatologists (55 families) and by referrals from urologic surgeons (6 families). All of the families with BHD were invited to participate in the study regardless of the number of affected individuals in the family or the presence or absence of associated health problems. A family was considered affected with BHD if it had (1) one or more members with 10 or more skin lesions that were clinically compatible with FFs and/or (2) a minimum of one histologically proven FF. Histologically, an FF was characterized by multiple anastomosing strands of 2–4 epithelial cells extending from a central hair follicle. Phenotypic expression of BHD skin papules can be variable among affected members of a family with BHD; therefore, once a proband with clinically positive or histologically proven FFs was identified in a BHD-affected family, other family members were screened and classified as affected for genotype-phenotype evaluation on the basis of (1) the presence of a histologically proven FF, (2) inheritance of the family’s germline BHD mutation, (3) inheritance of the family’s BHD-affected haplotype, or (4) obligate carrier status. We also included family 238 as affected with BHD, because multiple members were affected with bilateral, renal oncocytic hybrid neoplasms, a rare histologic variant uniquely associated with BHD. Participants in this study provided written informed consent. The protocol was approved by the institutional review boards of the NCI and the University of Manitoba.

Patient Evaluation

All members of families with BHD who were aged >20 years were evaluated at the NIH Clinical Center and/or in the field. Blood samples were obtained for DNA extraction and mutation analysis. Each patient received a detailed dermatologic examination, and biopsies were performed for lesions suspected to be FFs.

Family members seen at the NIH were evaluated for other phenotypic manifestations associated with BHD. Occult renal malignancies were detected by CT scan of the abdomen before and after administration of ∼120 ml of Ioxilan 300 (Cook Imaging). The presence of lung cysts and evidence of spontaneous pneumothorax were evaluated by thoracic CT scan, by use of high-resolution sections (1 mm) through the chest at 10-mm intervals. Additional information was obtained from patient medical histories, outside medical records, and tissue specimens obtained for NIH pathology review.

Mutation Analysis of the BHD Gene

DNA was extracted from peripheral-blood leukocytes, in accordance with standard procedures. Exon/intron structure of the BHD gene was determined by BLAST alignment of two overlapping, uncharacterized, full-length transcripts from skin melanoma (GenBank accession numbers BC015725 and BC015687), against the BAC clone RP11-45M22 (GenBank accession number AC055811); intronic primers were designed to amplify 14 coding exons and splice junctions for sequencing, as described elsewhere (Nickerson et al. 2002). Subsequent analyses showed that this exon/intron structure matched the cloned full-length BHD transcript. PCR products were generated using standard conditions and were purified; double-stranded sequencing was performed using BigDye Terminators (Applied Biosystems), as described elsewhere (Nickerson et al. 2002). Sequence variants were examined for cosegregation with disease in their respective families by denaturing high-performance liquid chromatography or by single-stranded sequencing. Insertions and deletions were subcloned with Topo Cloning Kit (Invitrogen) and were sequenced. At least 160 unaffected individuals were examined for each disease-associated sequence variant.

BHD Haplotype Carrier Analysis

Family members were evaluated for inheritance of the BHD-affected haplotype in cases in which a BHD germline mutation was not identified in the proband. To determine if a family member inherited the BHD-affected haplotype, individuals were genotyped using seven microsatellite markers (D17S953, D17S740, [location of BHD] D17S2196, D17S620, AGAT100, D17S805, and D17S1824) that spanned an 8-cM distance flanking the BHD locus on chromosome 17p. Haplotypes were determined from the genotype order in which the least number of recombinants occurred. All members of BHD-affected families who presented with FFs inherited the family’s BHD-affected haplotype.

Statistical Analysis

The percentage of BHD-affected individuals with renal tumors was calculated in two ways: (1) by dividing the number of BHD-affected individuals with renal tumors by the number of BHD-affected individuals seen at the NIH and in the field, and (2) by dividing the number of BHD-affected individuals who had renal tumors by the number of BHD-affected individuals who were screened by abdominal CT scan. To overcome selection bias, we assessed the frequency of renal tumors found in patients with BHD two ways. Generally, patients who were affected with BHD skin lesions and/or were symptomatic for renal tumor came to the NIH to be seen and screened. Some patients who were less severely affected did not come to the NIH and were seen only in the field and therefore did not undergo abdominal/pulmonary CT screening. Differences in frequency of renal tumors, pneumothorax, or lung cysts between the groups with exon 11 C-insertion and C-deletion mutations were tested with Fisher’s exact test (two-sided) (see the Calculation of Fisher’s Exact Test Web site). P<.05 was considered statistically significant.

Description of Patients

For the present study, 219 individuals (110 females and 109 males) from 53 families (whose members inherited either a germline mutation or the affected haplotype) were classified as affected with BHD. Of those individuals, 187 were evaluated in the field or at the NIH Clinical Center. Data were available for 7 deceased individuals with obligate carrier status, 12 individuals were evaluated by outside dermatologists, and 10 individuals provided a blood sample for mutation analysis but did not undergo further medical evaluation at the NIH.

The number of affected individuals in a family ranged from 1 to 25. In the present study, the largest number (25) of affected family members in a family with BHD was found in the Canadian family with BHD, family 172, originally studied by Drs. Birt, Hogg, and Dubé (Birt et al. 1977). Twenty-four families had only one member affected with BHD.

Dermatologic Manifestations

The hallmark BHD cutaneous manifestation is a benign tumor (hamartoma) of the hair follicle, an FF. Of the 175 BHD-affected individuals whose skin papules were biopsied and evaluated histologically, 147 (84%) were diagnosed with a histologically proven FF. All families that carried a BHD mutation or inherited the affected haplotype contained individuals who developed BHD skin lesions, with the exception of one family, family 238, referred to the NCI by the proband’s urologist, in which three sisters developed bilateral renal oncocytic hybrid neoplasms and lung cysts without evidence of FFs. In the present study, the cutaneous lesions were not found in patients aged <20 years.

Pulmonary Manifestations

A high percentage of patients with BHD develop air-filled lung cysts (Zbar et al. 2002) and have a 50-fold higher risk of developing spontaneous pneumothorax. In the present study, 64 (32%) of 198 BHD-affected individuals for whom medical history was available had a history of spontaneous pneumothorax. Of the 129 BHD-affected individuals who were screened by pulmonary CT scan, 110 (85%) were found to have one or more lung cysts consistent with the BHD phenotype. Forty-five (85%) of 53 families with BHD germline mutations or affected-haplotype carriers had members who developed lung cysts or spontaneous pneumothorax.

Renal Neoplasia in BHD

BHD is unique among inherited renal cancer syndromes because patients develop not one, but a variety of histologic types of renal tumors, including chromophobe RCC, renal oncocytoma, clear-cell RCC, and papillary RCC (Pavlovich et al. 2002, 2005). Most often, BHD-associated renal neoplasms are oncocytic hybrid tumors, comprising elements of both chromophobe and renal oncocytoma. Thirty-eight BHD-affected individuals representing 24 BHD-affected families developed renal tumors. Males developed renal tumors 2.5-fold more often than did females (27 males; 11 females). The median age at diagnosis was 48 years (range 31–71 years). Tumors developed both bilaterally with multiple foci and unilaterally with a single focus. The number of patients with renal neoplasia in a BHD-affected family ranged from one to three, most frequently one case per family. The frequency of renal tumor among BHD-affected individuals seen at the NIH or in the field was 20% (38/187). The frequency of renal tumor among BHD-affected individuals who were evaluated by CT scan was 29% (38/132). Overall, 45% (24/53) of BHD-affected families with germline BHD mutations or affected-haplotype carriers contained individuals who were affected with renal tumors.

Other Phenotypic Features Found in BHD-Affected Individuals

There are reports that describe a number of additional phenotypic manifestations in BHD-affected individuals, including colon polyps and colon carcinoma (Hornstein and Knickenberg 1975; Hornstein 1976; Rongioletti et al. 1989; Le Guyadec et al. 1998), lipomas (Chung et al. 1996; Toro et al. 1999), angiolipomas (Chung et al. 1996), parathyroid adenomas (Chung et al. 1996), and parotid oncocytomas (Liu et al. 2000). Among the 219 BHD-affected individuals in the present study, four parotid gland tumors were reported in 2 men (diagnosed at ages 20 and 39 years) and 2 women (diagnosed at ages 62 and 72 years). Three of the parotid tumors were classified as oncocytomas. Additionally, three individuals with lipomas and nine individuals with colon polyps/carcinoma were identified through medical history.

Variable Expression of Phenotypic Manifestations

BHD is characterized by FFs, lung cysts, spontaneous pneumothorax, and an increased risk of developing renal neoplasia. These manifestations may occur singly or in various combinations in a patient, and variable phenotypic expression may be observed among family members from the same BHD-affected family. Among the 53 families with BHD whose affected members had a BHD germline mutation or carried the BHD-affected haplotype, 22 (41.5%) of the families had skin, lung, and renal manifestations, 22 (41.5%) of the families had members with skin and lung phenotypic features, 1 (2%) of the families had renal and lung phenotypic features, 7 (13%) of the families had only skin papules, and 1 (2%) of the families had renal tumors and skin papules (table 1).

BHD-Mutation Spectrum and Mutational Hotspot

Elsewhere, we reported the identification of BHD germline mutations, by direct sequence analysis, in 30 of 61 families with BHD recruited to the NCI (Nickerson et al. 2002). In the present study, we have completed mutation analysis of the remaining 30 families and, to date, have identified germline BHD mutations in 51 (84%) of 61 families with BHD in the NCI cohort. Each family’s mutation cosegregated with disease and was not found in at least 160 unaffected individuals. Altogether, 22 novel mutations have been identified (5 were reported elsewhere [Nickerson et al. 2002]), including 16 insertions or deletions that cause frameshifts leading to premature truncation of the protein, 3 nonsense mutations that encode early termination codons, and 3 putative splice-site mutations (table 1 and fig. 1). Of note, no germline missense BHD mutations were detected in these BHD-affected families.

A total of 53% (27/51) of families with a germline BHD mutation had a cytosine insertion (c.1733insC [18 families]) or deletion (c.1733delC [9 families]) in the mononucleotide tract of eight cytosines (c.1733–1740) in exon 11. This mutational hotspot has been identified by several laboratories (Khoo et al. 2002; Nickerson et al. 2002). Mononucleotide tracts are thought to be hypermutable due to slippage of the DNA polymerase during replication (Streisinger et al. 1966), which results in frameshift mutations leading to protein truncation. Although most of the mutations identified in the remaining 24 BHD-affected families were unique, 2 families had the same 28-bp duplication in exon 9 (c.1378–1405 dup), 3 families shared a nonsense mutation in exon 12 (c.1844C→G), and 2 families shared a putative splice-site mutation in intron 9 (IVS9+2T→G) (table 1 and fig. 1).

Genotype-Phenotype Correlation: Location within the BHD Gene

Mutations were identified in all translated exons (exons 4–14), with the exception of exons 8 and 10 (fig. 2). Exon 11 was the most frequent site of mutation in the 51 BHD-affected families (27 families had the hotspot mutation), exons 9 and 12 were each the site of mutations in 7 families, and exons 5 and 7 were each the site of mutations in 3 families. Exons 4, 6, 13, and 14 were each the site of one unique mutation in a BHD-affected family. Eight (16%) of the families had mutations in the 5′ half, and 43 (84%) of the families had mutations in the 3′ half of the BHD gene.

The most 5′ mutation occurred at nt 513 (c.513delT) in exon 4 (20 aa from the initiation codon) in family 7306 (fig. 2). This frameshift mutation was predicted to truncate the protein 34 aa (missense) downstream. The affected individual in this family developed only BHD skin manifestations, without the other phenotypic features. The most 3′ mutation occurred at nt 2034 (c.2034 C→T) in exon 14 (family 168) and was predicted to produce a premature termination codon at codon 527, eliminating only 52 aa from the full-length protein (fig. 1b). Nonetheless, full expression of the BHD phenotype was observed in family 168; three affected members developed renal neoplasia and lung cysts, and one was diagnosed with FF. Interestingly, the other nonsense mutations predicted to produce premature termination codons were located in exon 12 (c.1844C→G and c.1884C→T), also near the 3′ end of the BHD gene.

Renal tumors were found in 5 (14%) of the mutation carriers with mutations in the 5′ half of the BHD gene and in 30 (86%) of the mutation carriers with mutations in the 3′ half of the BHD gene. Renal tumors were present in patients with frameshift mutations (insertion or deletion), nonsense mutations, and splice-site mutations in BHD. Of patients with spontaneous pneumothorax, 97% (55/57) inherited mutations in the 3′ half of the BHD gene (table 1), and frameshift, nonsense, and splice-junction mutations were represented among those affected individuals.

Genotype-Phenotype Correlation: Mutational Hotspot

The most frequently mutated site in BHD was a cytosine insertion or deletion in the C₈ tract of exon 11 (c.1733insC or c.1733delC), identified in 27 BHD-affected families. We estimated the frequency of developing FFs, lung cysts, spontaneous pneumothorax, or renal tumors in BHD-affected family members with the C-insertion mutation compared with those individuals with the C-deletion mutation (fig. 3). A total of 64 individuals carried the C-insertion mutation, and 27 individuals carried the C-deletion mutation. The frequency of histologically proven FFs among those individuals whose lesions were biopsied was similar among C-insertion–mutation carriers (83% [44/53]) and C-deletion–mutation carriers (87% [20/23]). The frequency of pneumothorax episodes was also similar between carriers of C-insertion (37% [23/62]) and C-deletion (35% [9/26]) mutations. Lung cysts were detected on thoracic CT scans in 88% (35/40) of C-insertion–mutation carriers, compared with 77% (13/17) of C-deletion–mutation carriers, which is not significantly different. However, we noted that renal tumors developed in significantly fewer C-deletion–mutation carriers than C-insertion–mutation carriers. Among C-deletion carriers, only 6% (1/16) of those who were evaluated by CT scan (4% [1/26] of those seen at the NIH or in the field) developed renal tumors. By comparison, 33% (13/40) of C-insertion–mutation carriers who were evaluated by CT scan (23% [13/56] of those seen at the NIH or in the field) developed renal tumors (P=.03) (fig. 3). The median age (46.5–50 years) of patients with BHD seen at the NIH was similar for men and women with the C-insertion and the C-deletion mutations, and the number of males and females in each mutation group was approximately equal.

Genotype-Phenotype Correlation: Putative Splice-Site Mutation in IVS9

Two apparently unrelated families were found to have a putative splice-site mutation in the BHD gene within intron 9, IVS9+2T→G (table 1 and fig. 1a). Renal-tumor frequency in the affected members of families 229 and 238 was 3/5 and 3/3, respectively, which raises the possibility that this splice-site mutation might be associated with a higher frequency of renal tumors. A third family (family 242) inherited a different nucleotide change in the same splice-donor site (IVS9+1G→A) and had a renal-tumor frequency of 1/9 (table 1).

Linkage to 17p11.2 without Detectable BHD Mutation in Two Large Families with BHD

We screened 61 families with BHD that were affected with histologically proven FFs, and we found germline BHD mutations in 51 families. Two additional families, however, showed linkage to chromosome 17p11.2 by haplotype analysis with polymorphic microsatellite markers, but we were unable to detect a sequence variation in the BHD gene by direct sequence analysis. Evidence of heterozygosity at a number of informative SNP loci reduced the possibility of intragenic deletion in these families, although this deletion cannot be absolutely ruled out.

Family 172 is the original Canadian family described by Drs. Birt, Hogg, and Dubé (Birt et al. 1977) and was used in linkage analysis studies that identified the BHD-disease locus on 17p11.2 (LOD score 4.98 at marker D17S740; θ=0) (Schmidt et al. 2001). Twenty-five family members had the same BHD-affected haplotype over an 8.5-cM region of chromosome 17p11-12 (Schmidt et al. 2001). However, direct sequencing of the 14 coding exons revealed no BHD sequence variation in peripheral-blood DNA from affected members of family 172. All family members whose lesions were biopsied had at least one histologically proven FF; one affected family member developed a renal tumor, and six members had a history of pneumothorax (table 1).

In family 240, a large family with BHD recruited to the NCI study, eight family members were identified with histologically proven FFs (Lindor et al. 2001). Upon genotyping with six polymorphic microsatellite markers flanking the BHD locus, from D17S953 to D17S1824, 17 members were found to inherit the BHD-affected haplotype on chromosome 17p11.2 (data not shown). One BHD-haplotype carrier had a history of pneumothorax, two developed renal tumors, and four were found, on thoracic CT scan, to have lung cysts (table 1). Complete sequencing of BHD in DNA from affected-haplotype carriers in family 240 did not reveal any sequence variation.