We recruited families and individuals to screen for BHDS by mailing four cycles of letters over a 3-year period seeking referrals from the 11,000 members of the American Academy of Dermatology for patients with cutaneous signs of BHDS. Patients were evaluated in consecutive order in a protocol approved by the National Cancer Institute's Institutional Review Board. All members of families screened for BHDS who participated in this study gave written, informed consent.

Families with BHDS were evaluated at the Clinical Center of the National Institutes of Health. Medical histories (fibrofolliculomas, spontaneous pneumothorax, and renal tumors) were obtained and physical examinations were performed. A detailed dermatologic examination was conducted and skin biopsies were obtained of selected lesions suspicious for fibrofolliculoma. The presence of fibrofolliculomas was designated as the sole criterion for the diagnosis of BHDS because fibrofolliculomas are rare and specific for BHDS (2). We defined a family as affected with BHDS when it contained one or more members with BHDS cutaneous lesions and a minimum of one lesion histologically confirmed as a fibrofolliculoma. Helical computed tomography (CT) scans of the chest were used to screen for pulmonary abnormalities in our patient population. The chest was scanned before and after intravenous administration of 120 cm3 of Ioxilan 300 (Cook Imaging Corp., Bloomington, IN). High-resolution 1-mm sections were obtained through the chest at 10-mm intervals. Pulmonary cysts were diagnosed on the basis of CT scans, and the numbers, location, and size of cysts were recorded. Pneumothorax was documented by history and a review of medical records. We defined recurrence of a pneumothorax as one occurring on the ipsilateral side more than 7 days after the most recent prior pneumothorax has resolved. α₁-Antitrypsin serum levels were obtained in 50 family members evaluated in the National Institutes of Health (NIH) Clinical Center. In addition, capillary oxygen saturation was routinely measured, together with vital signs for each person screened.

The genomic sequence analysis of the BHD gene was performed as previously reported (5). The first reported BHDS family (1), in whom a BHD germline mutation was not identified (6), although haplotype analysis showed linkage to chromosome 17p11.2 (4), was included in the present study.

Data collected from the 198 patients with BHDS consisted of pack-years of cigarette smoking, number of fibrofolliculomas, sex, number of pneumothoraces, number of cysts in each lung compartment, and the total volume of cysts in the left lung and right lung. Volumes were derived using an estimated formula under an assumption that the cysts were approximately the shape of prolate spheroids: volume = 4/3 π ab², where a is one-half the length of the longest semi-axis and b is one-half the length of the shorter semi-axis. The sum of the individual cyst volumes was used to calculate the total cystic space volume in each lung.

Analyses of dichotomized parameters and their relationship to presence or absence of a pneumothorax were done using a chi-square test. Associations between categorical parameters and presence or absence of a pneumothorax were done using an exact Cochran-Armitage test (28), as were associations between the number of pneumothoraces and dichotomous parameters. Associations between the number of pneumothoraces and ordered categorical parameters were performed using an exact Jonckheere-Terpstra test (29). Continuously measured parameters were compared between subjects with and without a pneumothorax using the Wilcoxon rank sum test. The Jonckheere-Terpstra test was used to assess the statistical significance of the trend in continuously measured parameters across increasing numbers of pneumothoraces. A Kruskal-Wallis test was used to determine whether continuously measured parameters differed according to mutation exon location, and to determine the association between exon location of the mutation and the number of pneumothoraces. Mehta and Patel's version of Fisher's exact test was used to evaluate the significance of the association between continuous parameters and the presence of one or more pneumothoraces (30).

Logistic regression analysis was used to identify whether a combination of factors could be jointly associated with either the presence or absence of a pneumothorax or with increasing numbers of pneumothoraces. Many of the variables related to the numbers of cysts were derived from one another (e.g., the number of lung cysts on the right lung is the sum of the cysts on the right lower and upper lobes). Therefore, the variables included in the model only consisted of parameters that were not directly calculated from one another. In the final regression modeling for classification of “pneumothorax or not,” all possible classifications were examined, and a threshold for classification was selected that simultaneously provided high sensitivity and specificity. The Kaplan-Meier method was used to describe the association between age and the probability of development of a first spontaneous pneumothorax (31). All p values are two-sided and were not adjusted for multiple comparisons.

Patients' clinical characteristics and the frequencies are listed in Table 1. Our cohort included 101 men and 97 women with a median age of 49 years. The age range was 22 to 77 years, with only eight patients younger than 30 years. Eighty-nine percent (177/198) of patients with BHDS had lung cysts on CT scans of the chest. Approximately 24% (48/198) of patients and 35% (31/89) of BHDS families screened for lung cysts had a history of spontaneous pneumothorax. In our cohort of patients affected with BHDS, we found a relatively equal distribution of pneumothorax for men and women. Of the men affected with BHDS, 20% (20/101) had a history of a pneumothorax and, of the women with BHDS, 29% (28/97) had a history of a pneumothorax. All patients with a history of pneumothorax had multiple lung cysts identified by chest CT imaging (Figure 1). The earliest reported age of initial pneumothorax was 22 years, the median age of occurrence was 38 years (range, 22–71 yr), and the median age of the last pneumothorax was 42 years (range, 22–75 yr). Seventy-five percent (36/48) of patients had a second pneumothorax, the average number of pneumothoraces per patient was two, and four patients experienced five separate pneumothoraces. Differences in recurrence among patients may reflect efficacy of treatment used.

Figure 1. Pulmonary manifestations of patients with Birt-Hogg-Dubé syndrome on conventional helical (5 mm) chest computed tomography imaging. (1) Loculated pnemothorax on the right lung as well as multiple bilateral smaller cysts in the lung bases. (2) Cluster of small cysts in the right lung base. (3) Bilateral basilar lung cysts. (4) Bilateral, multiple, peripheral lung cysts more pronounced on the right side. (5) Loculated pneumothorax in right upper lobe. (6) Pneumothorax on the left with collapse of the left upper lobe and multiple right-sided cysts.

Download Figure | Download Powerpoint

[More] [Minimize]

Ninety-three percent of patients with a diagnosis of BHDS had fibrofolliculomas with a distinctive clinical presentation characterized by multiple white or skin-colored papules distributed over the face, neck, and upper trunk. However, 14 patients in our cohort had fewer than 10 skin lesions suspicious for fibrofolliculoma, but all 14 patients had at least one biopsy-confirmed lesion. Forty-five patients with BHDS had kidney tumors. Of these, 27% (14/45) had a history of pneumothorax and 93% (42/45) had lung cysts.

Table 2 shows the pneumothorax history and treatment of the 48 patients affected by spontaneous pneumothorax. Of these patients, 58% (28/48) were women and 41% (20/48) were men. Sixty-seven percent (32/48) of patients were nonsmokers. There were 101 episodes of pneumothoraces and no patient had simultaneous bilateral pneumothoraces. The right lung had the highest frequency of pneumothorax. Approximately 48% (23/48) of patients had a pneumothorax in the right lung only, 29% (14/48) had a pneumothorax in the left lung only, and 23% (11/48) of patients had a pneumothorax in both the right and the left lungs at different times.

Twenty-three percent of pneumothoraces were managed with observation alone. These patients were in stable condition and only minimally compromised by the pneumothorax. Of the pneumothoraces managed by observation alone, only 39% (9/23) completely resolved, and the remaining 61% (14/23) recurred, requiring medical treatment. Of the 101 pneumothoraces, 77% required medical intervention and were treated by various methods. Thirty-five percent (35/101) were treated with tube thoracostomy (chest tube) only. Six pneumothoraces were managed with a combination of tube thoracostomy and chemical pleurodesis (quinacrine, tetracycline, silver nitrate, or talc). Approximately 14% (15/101) of pneumothoraces were treated by open thoracotomy and a second treatment, including mechanical pleurodesis (abrasion to produce adhesion between parietal and visceral pleura) in 10% (10/101), chemical pleurodesis in 2% (2/101), and lung resection in 3% (3/101). Approximately 13% (13/101) of pneumothoraces were treated with combined tube thoracostomy, thoracotomy, and a third treatment, including mechanical pleurodesis in 7% (7/101), lung resection in 6% (6/101), and chemical pleurodesis in 2% (2/101), respectively. In addition, seven pneumothoraces were treated with a combination of tube and open thoracostomies, lung resection, and mechanical pleurodesis. Another pneumothorax was treated with both a combination of tube and open thoracostomy, lung resection, and chemical pleurodesis. Three patients with BHDS treated at the NIH underwent video-assisted thoracoscopic surgery. Pneumothoraces were not related to a particular calendar period. After serum α₁-antitrypsin levels in 50 patients were within normal limits, we no longer performed the test. Capillary O₂ saturation showed no significant abnormality in the measurements performed during routine screening for BHDS.

We examined the association between the history of pneumothorax and the following categorical parameters: sex, smoking history, severity of fibrofolliculomas, and lung cysts. The results of the univariate analysis of categorical parameters are summarized in Table 3. The only categorical parameter that was significantly associated with pneumothorax was the presence of lung cysts (p = 0.006).

In a second analysis, we investigated the association of continuous parameters (the total number of cysts per lung lobe, total number of intraparenchymal cysts, total number of subpleural cysts, number of lung lobes with cysts, smoking history, age when scanned, and lung cyst total volume) with pneumothorax. The results of tests for significance from the univariate analysis of continuous parameters are summarized in Table 4. We found that every parameter related to the number of lung cysts was significantly associated with history of pneumothorax. No association was found between age at scan or smoking history and presence or frequency of pneumothoraces. In addition, total lung cyst volume and largest cyst diameter and volume were significantly associated (p < 0.0001) with history of pneumothorax (Table 5). In addition, we found an association between the number of pneumothoraces and the total cyst volume (p < 0.0001) (Table 5). There was an increase in median total cyst volume (0.8, 5.1, 8.5, and 10.3 cm³) with increasing number of pneumothoraces (0, 1, 2, +3, respectively). Similarly, we found that the largest cyst diameter and volume were significantly associated with the number of pneumothoraces (Table 5).

The relationship between history of pneumothorax and different parameters related to lung volume and cysts was investigated using a logistic regression model. The logistic regression model determined that only the total number of cysts in the right parenchymal lower lobe (p = 0.050) and the total number of cysts in the right pleura-based middle lobe (p = 0.002) were needed to classify a patient as to whether he or she was likely to have a pneumothorax or not. These same parameters were also associated with increasing numbers of pneumothoraces.

We used the Kaplan-Meier method to describe the association between age and the first spontaneous pneumothorax among all patients in the BHDS cohort (Figure 2). No pneumothoraces were diagnosed before age 22 years in this cohort, with the oldest patient experiencing her first pneumothorax at 71 years. By age 30 years, the probability of having the first pneumothorax is 6% (95% confidence interval [CI], 3–10%), 14% (95% CI, 10–20%) by 40 years, and 75% (95% CI, 19–32%) by age 50. Of the 48 patients who had a pneumothorax, 88% (42/48) of them had his or her first pneumothorax before age 50, 10% (5/48) had one between the ages of 51 and 60 years, whereas only 1 patient had a pneumothorax that occurred after age 60, at 71 years.

Figure 2. Kaplan-Meier curve showing the association between age and the first spontaneous pneumothorax among patients with Birt-Hogg-Dubé syndrome.

Download Figure | Download Powerpoint

[More] [Minimize]

BHD mutations were present in 81% (154/190) of patients and 85% (68/80) of BHDS families tested. The cytosine mononucleotide tract in exon 11 of BHD was most frequently mutated, accounting for 48% (70/154) of patients and 47% (32/68) of BHDS families who had a mutation detected. Thirteen patients from one large BHDS kindred that previously showed linkage to chromosome 17p11.2 by haplotype (4) had no sequence variation in the BHD gene (6). Nine families were not screened for BHD mutations by direct sequencing.

Patients with a history of pneumothorax share a similar BHDS mutation spectrum with all the patients with BHDS in this study, but with some exceptions. BHD mutations were present in 87% (41/47) of individuals and 90% (27/30) of BHDS families with a history of pneumothorax tested. Similar to previous studies, the “hot spot” mutation, c.1733ins/delC, in exon 11 was the most frequent site of mutation, accounting for 38% (18/47) of individuals with a history of a pneumothorax and 44% (18/41) of all the mutations detected. We were unable to detect a sequence variation in BHD by direct sequence in six individuals, but four of these individuals were part of a family that showed linkage to chromosome 17p11.2 by haplotype analysis (4). In addition, one individual was not screened for BHD mutations.

There was no association between BHD mutation status (no mutation vs. mutation and/or linkage to chromosome 17p11.2), mutation types (insertion, deletion, nonsense, and splice site, and frameshift vs. nonsense and splice site; splice site vs. exon mutation, or hot-spot mutation vs. all other BHD mutations, or splice site vs. all other BHD mutations), or location, and lung cyst parameters or pneumothorax. Our analysis showed a trend for differences in pneumothoraces according to exon location (p = 0.01), with individuals with BHD mutations in exon 9 and exon 12 having more pnemothoraces than individuals with BHD mutations in other exons (Table 6). In addition, our analysis revealed that mutation exon location was associated with the numbers of cysts (p = 0.0002), with individuals with BHD mutations in exon 9 having more cysts (median = 32) than individuals with mutation in other exons (Table 7). Similarly, we found that the BHD mutation exon location was associated with the size (p < 0.005) and volume (p < 0.01) of the largest cysts. Individuals with mutations in exons 9 and 12 had the largest cyst diameters (1.9 and 2.9 cm) and volumes (2.7 cm³) (Table 7).