Increased Rate of Fractures and Severe Kyphosis: Sequelae of Living into Adulthood with Cystic Fibrosis

Methods

Patients

The study sample consisted of 70 adults (older than 18 years of age) with advanced cystic fibrosis who were referred for lung transplantation at the University of North Carolina between January 1994 and December 1996. The Committee on Human Research (IRB #96-Med-336) approved this retrospective cohort study and required verbal consent from participating patients. Cystic fibrosis was diagnosed by elevated sweat chloride concentrations and an appropriate clinical picture. All patients had end-stage lung disease and an anticipated survival of less than 2 to 3 years [21].

Fracture History

The history, date, and mechanism of fracture were determined by personal interviews done using methods similar to those of the National Health Interview Study [22]. Fractures were required to have patient report of radiographic confirmation at the time of the fracture, but we did not review these radiographs. Confirmation of long-bone fractures required patient report of treatment with casting. Fractures occurring between birth and 6 years of age were not assessed because most patients were unable to provide accurate histories for those years. The number of years that each patient was at risk for fracture was summed by age interval to determine the total number of patient-years for this cohort. All patients were asked to give their date of puberty; to give a detailed history of corticosteroid use (expressed as cumulative dose of prednisone in grams); to state whether they had received therapy for osteoporosis; and, if they were female, to state whether and when they had had oligomenorrhea.

Bone Densitometry

Bone mineral density was measured in all patients by a single, registered radiologic technologist using dual-energy x-ray absorptiometry (Hologic QDR 1000/W, Waltham, Massachusetts) [23]. Lumbar spine (L1-L4), nondominant femoral neck, and total-body bone mineral densities were measured; measurements were expressed in grams of bone mineral per cm² of bone. Quality control was maintained by daily scanning of an anthropomorphic spine phantom. The coefficient of variation for our Hologic QDR 1000/W densitometer is 0.3%, and the reference limits for variation are ± 1.5%. Results were expressed as T-scores, which are the number of SDs that the bone mineral density measurement is above (positive value) or below (negative value) expected peak bone mass. The age at which peak bone mass is achieved differs slightly for each site but is usually between 20 and 30 years. Using World Health Organization guidelines, we defined osteopenia as a T-score greater than −2.5 and −1.0 or less. Osteoporosis was defined as a T-score of −2.5 or less [24]. Normal bone mineral density was defined as a T-score of 0 ± 1.

Laboratory Measurements

Serum calcium, phosphorus, alkaline phosphatase, and creatinine concentrations were measured with an automatic analyzer (Hitachi 911, Boehringer Mannheim, Indianapolis, Indiana). Vitamin D metabolites were extracted from serum specimens (obtained while patients were fasting) with column chromatography and were measured with radiobinding assays (Nichols Institute, Raleigh, North Carolina). The 1,25-dihydroxyvitamin D and 25-hydroxyvitamin D assays had sensitivities of 5 pg/mL and 5 ng/mL, respectively. Testosterone levels were measured by using a competitive radioimmunoassay (Diagnostics Product Corp., Los Angeles, California). Genotype analysis for the 15 most common CFTR mutations was done by using published methods [25].

Radiographic Analysis

Screening of the most recent posteroanterior and lateral chest radiographs for the extent of thoracic kyphosis as well as rib and vertebral body fractures was done by a single musculoskeletal radiologist on 65 patients (5 patients did not undergo radiography at the University of North Carolina). Thoracic kyphosis was measured from the second or third to the twelfth thoracic vertebral body by using a modification of the method of Cobb [26]. The number of vertebral compression fractures was also determined from the lateral radiograph by measuring anterior and posterior vertebral body height and expressing the difference as a percentage [27]. The posteroanterior chest radiograph was examined for the presence of rib fractures. Nonacute fractures appeared as focal deformities in the rib contour with varying degrees of reparative bone formation.

Statistical Analysis

Analyses were done with SAS software (version 6.12, SAS Institute, Cary, North Carolina) [28], and significance was based on two-tailed tests with a P value less than 0.05. Discrete variables are summarized as percentages, and continuous variables are summarized as means ±SD. Nonparametric methods were used when the distribution of the continuous variable was skewed. Student t-tests were used to compare differences in all clinical and anthropomorphic variables except fracture rates [29]. Relationships between spine and femur T-scores and seven continuous variables-age, age at puberty, cumulative steroid dose, body mass index, FEV₁, FVC, and vitamin D levels-were assessed with either the Pearson or Kendall correlation coefficients [30]. Backward stepwise regression analysis was used to determine the multivariate relation between spine and femur T-scores and the aforementioned predictors. Clinical predictors and T-scores were compared, using the Kruskal-Wallis test, by dividing the patients into three groups: those without fractures, those with one fracture, and those with more than one fracture. This was done to determine whether differences existed between groups. Using spine or femur z-scores (the number of SDs that a bone mineral density measurement was above or below that of age- and sex-matched normal controls) rather than T-scores gave similar results (data not shown) because most patients were in the age range at which peak bone mass is expected.

For the analysis of fractures and kyphosis angles, patients were grouped by age to conform with published databases (normal databases are based on age) [22, 31]. Two separate fracture rate analyses were done: In one, all fractures were used as the numerator; in the other, all persons with any fracture were used as the numerator. The latter analysis was done to eliminate the possibility that fracture clustering had affected the results. We used person-years as the denominator for both analyses. The first and second analyses used Poisson regression and large-sample binomial approximation to the normal distribution, respectively, to calculate P values to compare fracture rates between patients with cystic fibrosis and controls. The Student t-test was used to compare kyphosis angles in patients with cystic fibrosis and controls.

Results

Patient Characteristics

Patient characteristics are shown in Table 1. Men and women with cystic fibrosis were 5.8 ± 7.1 shorter and 5.9 ± 6.8 cm shorter, respectively, than age-matched controls [32]. Nine of 35 women (26%) reported menstrual changes, most often with exacerbations of lung disease. None had had amenorrhea for more than 3 months outside of pregnancy. Thirty of 70 patients had received corticosteroids at some time in the past. More than 90% of patients had been prescribed ergocalciferol in the form of ADEK vitamins (Scandipharm, Inc., Birmingham, Alabama), but only 1 patient had received additional antiosteoporosis medication; therapy with this medication was initiated because of multiple fractures. The effect of these medications on T-scores in this cohort could not be determined because of the lack of baseline (pretreatment) data.

Laboratory Measurements

The mean (±SD) levels of calcium (8.9 ± 0.6 mg/dL [2.22 ± 0.15 mmol/L]), phosphorus (3.6 ± 0.5 mg/dL [1.16 ± 0.16 mmol/L]), alkaline phosphatase (167.2 ± 131.5 U/L [2.79 ± 2.19 µkat/L]), 25-hydroxyvitamin D (20.9 ± 11.4 ng/mL [52.17 ± 28.45 nmol/L]) and 1,25-dihydroxyvitamin D (36.1 ± 12.4 pg/mL [86.6 ± 29.76 pmol/L]) for the group were in the normal range. However, 20% of patients had subnormal 25-hydroxyvitamin D levels, and 5% had subnormal 1,25-dihydroxyvitamin D levels. No differences between the sexes were noted for these variables. Testosterone levels were below normal in 20% of the men, but the mean level for the group was within the normal range (385 ± 136 ng/dL [1334.8 ± 471.5 nmol/L]). Twenty of the 35 patients who were tested for CFTR mutations were homozygous for DeltaF508, 12 were DeltaF508 compound heterozygotes, and 3 had other CFTR mutations.

Bone Mineral Density

Decreased bone mineral density was universal in this cohort of adults with late-stage cystic fibrosis (Figure 1). None of the 200 spine, femur, or total-body measurements were equal to or greater than expected peak bone mass (10 total-body measurements could not be obtained because patients were too dyspneic to lie flat for the duration of the measurement). As a group, the patients with cystic fibrosis had significantly lower mean bone mineral density than age- and sex-matched controls (P < 0.001) at all sites. Only 3 of 70 patients with cystic fibrosis (4%) had normal bone mineral density, whereas 27 (39%) had osteopenia and 40 (57%) had osteoporosis (Figure 2). The age-specific rates of osteoporosis were significantly higher in patients with cystic fibrosis than in a female control population (P < 0.001) [33]. Among patients with cystic fibrosis, no significant differences in mean spine, femur, or total-body T-scores were found between men and women.

Univariate associations between bone mineral density (that is, T-scores) and relevant clinical variables are shown in Table 2. The strongest predictors of femur T-scores were cumulative steroid dose, body mass index, and age at puberty. The strongest predictors of spine T-scores were cumulative steroid dose and age at puberty. Age, FEV₁, FVC, vitamin D levels, testosterone levels, and menstrual irregularity were not associated with spine or femur T-scores. In multivariable analyses, 37% of the variation of the T-scores at the spine could be predicted by the following equation: spine T-score = 3.66 −0.037 (cumulative steroid dose) −0.39 (age at puberty). Similarly, 40% of the variation in femur T-scores could be predicted by the following equation: femur T-score = −0.65+ 0.15 (body mass index) −0.029 (cumulative steroid dose) −0.29 (age at puberty).

Fractures

Thirty-seven patients reported 54 fractures over 1410 patient-years. Twenty patients had 1 fracture, and 17 had 2 fractures. All fractures were associated with trauma, but none were due to high impact (such as an automobile accident). Types of fracture are shown in Table 3. When compared with the normal population in data from the National Health Interview Survey, significant increases in fracture rates were found for men with cystic fibrosis aged 25 to 45 years (2.8 compared with 8.5 per 100 patient-years; P = 0.04) and women with cystic fibrosis aged 17 to 34 years (1.7 compared with 3.7 per 100 patient-years; P = 0.015) but not for male or female patients with cystic fibrosis aged 6 to 16 years [22]. The higher fracture rate in women with cystic fibrosis was predominantly due to a much higher fracture rate for women aged 25 to 34 years. When we addressed concern about clustering by using all persons with any fracture as the numerator, the difference in fracture rate between patients with cystic fibrosis and controls showed a trend toward but did not reach significance (P = 0.11 for women; P = 0.12 for men). Patients without fractures did not differ from those with 1 or more than 1 fracture on the basis of T-scores, age, age at puberty, body mass index, cumulative steroid dose, FEV₁, FVC, or vitamin D levels.

In addition, 62 unreported vertebral compression fractures (>20% anterior wedging deformities) and 14 unreported rib fractures were detected in 33 and 10 patients (of 65), respectively, solely by review of chest radiographs (Table 3). Five patients did not have chest radiographs at the University of North Carolina because they brought these films from and returned them to referring facilities. Vertebral fractures detected on chest radiographs were not included in the aforementioned assessment of fracture rate because the age at which the fractures had occurred was unknown. The mean vertebral compression fracture was 74 ± 13 degrees (range, 45 to 89 degrees). Vertebral and rib fractures are expected at the rate of 0.05 and 0.13 per 100 patient-years, respectively, among persons aged 6 to 44 years in the general population [22]. Therefore, as Figure 3 shows, patients with cystic fibrosis had spine and rib fracture rates that were approximately two orders (that is, 4.6 per 100 patient-years) and one order (that is, 1.0 per 100 patient-years) of magnitude, respectively, higher than expected (P < 0.001 for both). Significant differences (P ≤ 0.002 for all comparisons) remained for rib and spine fractures in a more conservative analysis that minimized the influence of clustering.

Kyphosis

Forty of 65 patients (62%) had abnormal kyphosis angles (>40 degrees). The mean kyphosis angle for the group as a whole was 44 ± 14 degrees, with no differences between the sexes. Significant differences (P < 0.001) were seen for kyphosis angles in men and women with cystic fibrosis and controls (Figure 4). Mean kyphosis angles for men and women with cystic fibrosis, despite their young age, were higher than those of controls who were older than 60 years of age [31]. As in the normal population, kyphosis angles increased with age in both women and men with cystic fibrosis. Kyphosis angles were associated with spine fractures (see above) and spine T-scores (r = 0.31; P = 0.02).

Discussion

Our study is the first to document significantly increased fracture rates and kyphosis angles in adults with cystic fibrosis. Because both the mechanical strength of bone and the force of injury play roles in the occurrence of fracture, the compromised activity level of patients with advanced cystic fibrosis does not seem to be sufficient to protect them against fractures. In fact, the composite fracture rate (130 fractures over 1410 patient-years) of 92 per 1000 patient-years is approximately equal to the fracture rate of postmenopausal women with severe osteoporosis who have already sustained a fracture (the heretofore worst-case scenario of osteoporosis) [24]. In our study, increases in fracture rate and kyphosis angle generally affected aging patients with cystic fibrosis in a pattern similar to that in the general population [22] but occurring approximately three decades earlier. These results indicate that osteoporosis in cystic fibrosis is not a subclinical problem. Higher fracture rates and worsening kyphosis-the latter further compromising lung function-add to the growing list of problems plaguing patients with cystic fibrosis as they survive into adulthood.

The results reported here are useful to clinicians who manage patients with cystic fibrosis. In addition to contributing to pain and debilitation, osteoporosis may accelerate the decline of respiratory function through structural chest wall changes and may compromise quality of life after transplantation for patients who choose this option. The knowledge that osteoporosis causes illness in patients with cystic fibrosis as they age calls for the study of effective interventions. As with osteoporosis in other populations, preventive therapies done in a multidisciplinary manner may be necessary to prevent the serious sequelae associated with this disorder in patients with cystic fibrosis.

Our study has several potential limitations. First, because we studied an older, very ill cohort of patients with cystic fibrosis, our results may not pertain to the typical patient with cystic fibrosis. This group was studied primarily to allow for a large number of years beyond childhood for the analysis of fractures. In addition, the number of patient-years in the older age groups, in which fracture rates began increasing, was relatively small. However, the data from review of chest radiographs strongly support the fracture rate data computed from patient interviews to confirm that fracture rates were significantly higher in patients with cystic fibrosis than in healthy controls. Because patients with cystic fibrosis now live longer, data to verify this finding will probably become increasingly available over the next few years. The third potential limitation of our work is that some of our patients with cystic fibrosis had been treated with corticosteroids; this may have been a partial cause of the osteoporosis and its sequelae noted in this group. However, our steroid-treated patients with cystic fibrosis, despite their lower T-scores, had fracture rates that were not significantly different from those of patients who had not received steroids. This finding suggests that cystic fibrosis, alone, is a significant risk factor for fracture.

Our data support anecdotal reports that suggest a prevalence as high as 10% to 20% for fractures in selected populations of adults with cystic fibrosis. Bhudhikanok and Bachrach and their coworkers found 12 fractures, including 1 of the femur, in 9 patients from a combined cohort of 71 children and adults at the Stanford Cystic Fibrosis Center [9, 14]. Our results are also consistent with observational data from Henderson and Specter [17], who found increased rates of fracture in 6- to 16-year-old girls with cystic fibrosis, but not boys, in a survey of 143 pediatric patients [17]. In contrast, the increased fracture rates in our study were found in older age groups (>25 years) for both male and female patients and were statistically significant. Surveying adults with late-stage cystic fibrosis and severe declines in bone density provided more power to detect significant differences between patients with cystic fibrosis and healthy persons.

Our observations extend those of Erkkila and colleagues [34], who first reported an increased prevalence of kyphosis in patients with cystic fibrosis, and others [17, 35, 36]. These previous reports, which primarily surveyed children and adolescents with cystic fibrosis, showed an unexpectedly high prevalence (9.1% to 40%) of abnormal kyphosis angles (>35 degrees or >40 degrees), worsening kyphosis with age, and more severe kyphosis in women than in men. Our data show a prevalence of kyphosis (defined as a Cobb angle > 40 degrees) that is twofold higher than the rates seen in younger patients with cystic fibrosis and an order of magnitude higher (60% compared with approximately 6%) than that expected for age- and sex-matched controls [31]. The increased kyphosis angles were due, in part, to a high prevalence of vertebral compression fractures; vertebral fractures, albeit less severe ones, have been reported previously [13, 18]. Because lateral chest radiographs are less sensitive than lateral spine series for the identification of vertebral deformities in the setting of dense pulmonary parenchymal infiltration, we may have underestimated the rates of vertebral fractures. The diminished stature of the patients in the study, averaging 6 cm less than normal, was probably partly due to both kyphosis and vertebral compression fractures.

The pathogenesis of osteoporosis and its sequelae in cystic fibrosis are not well defined. The multifactorial nature of osteoporosis in patients with cystic fibrosis has been discussed elsewhere [4]. Our results agree with those of others that suggest that poor nutrition is a harbinger of poor bone mineralization in cystic fibrosis [6, 12, 14, 19, 37]. The association between cumulative steroid dose (both before and after adjustment for confounders) and bone mineral density strengthens previously reported data [9] and suggests that corticosteroids play a significant (and potentially controllable) role in the pathogenesis of osteoporosis in cystic fibrosis, as they do in other lung diseases. However, because only approximately 20% of the variation in bone mineral density could be explained by corticosteroid use, osteoporosis related to cystic fibrosis is unlikely to be solely a corticosteroid-induced problem. We found little evidence of a significant association of vitamin D or sex hormone levels with low bone mineral density. Overall, less than half of the variation in bone mineral density in our patients could be explained by the clinical variables studied. Our data therefore suggest that a significant portion of the variation in bone mineral density in cystic fibrosis is due to other factors (for example, cytokines that alter the cellular activity of bone) or to the same factors but in other ways. Single measurements of serum biochemicals, sex hormones, or other potentially relevant end points will probably not accurately assess their association with a problem (such as low bone mineral density) that evolves slowly over time. Osteoporosis in cystic fibrosis is probably rooted in subnormal bone growth during childhood and manifests in adulthood as additional factors contribute to accelerated bone loss.

In summary, we have shown that osteoporosis in patients with cystic fibrosis results in significant increases in rates of fracture and kyphosis. The fracture rate increases and excessive kyphosis occurs at least three decades before it would normally be expected; this is consistent with severely depressed bone mineral density, which estimates chronologic bone age to be, on average, approximately 70 years in this group. As life expectancy continues to increase in patients with cystic fibrosis, severe osteoporosis and its sequelae will become more common and will add to the debilitation that results from progressive respiratory insufficiency. Efforts should be made to maximize bone growth during childhood and to increase and maintain bone density during adulthood to prevent fractures and kyphosis. The best approach to achieving these goals is not currently known, but an aggressive stance on maintaining good nutrition and minimizing long-term corticosteroid use are reasonable first steps. As more attention is focused on osteoporosis in cystic fibrosis, further clues to pathogenesis and useful therapies will appear. Studies at our institution are now being done to determine the roles of the gastrointestinal (mal)absorption of calcium and pulmonary infection (with the attendant increase in systemic cytokines that may promote bone resorption) in the pathogenesis of osteoporosis in cystic fibrosis. Trials of both oral and intravenous bisphosphonates are also under way; these may shed light on the potential benefit of these agents to patients with cystic fibrosis.

Dr. Renner: CB# 7510, 724 Burnett-Womack Building, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7524.

Ms. Buell: Duke Research Institute, Box 3865, Duke University Medical Center, Durham, NC 27710.

Ms. Riggs: Diagnostic Radiology, CB# 7600 University of North Carolina Hospitals, Durham, NC 27599.

Dr. Lester: Department of Orthopaedics, CB# 7055, 724 Burnett-Womack Building, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7524.

Dr. Ontjes: Division of Endocrinology, CB# 7527, 724 Burnett-Womack Building, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-7524.