Trends in Bone Mineral Density in Young Adults with Cystic Fibrosis over a 15 Year Period

Abstract

Background

Improvements in clinical care have led to increased life expectancy in patients with cystic fibrosis (CF) over the past several decades. Whether these improvements have had significant effects on bone health in patients with CF is unclear.

Methods

This is a cross-sectional study comparing clinical characteristics and bone mineral density (BMD) measured by dual energy X-ray absorptiometry (DXA) in adults with CF evaluated in 1995–1999 to age-, race-, and gender matched patients with CF evaluated in 2011–2013 at the same center on calibrated DXA machines.

Results

The cohorts were similar in terms of age, BMI, pancreatic insufficiency, presence of F508del mutation, and reproductive history. In the most recent cohort, pulmonary function was superior, and fewer patients had vitamin D deficiency or secondary hyperparathyroidism. Areal BMD measures of the PA spine, lateral spine, and distal radius were similarly low in the two cohorts.

Conclusions

Although pulmonary function and vitamin D status were better in patients in the present-day cohort, areal BMD of the spine was reduced in a significant number of patients and was no different in patients with CF today than in the late 1990s. Further attention to optimizing bone health may be necessary to prevent CF-related bone disease.

Introduction

Over the past several decades, patients with cystic fibrosis (CF) have been living longer, largely due to improvements in the treatment of the underlying pulmonary and gastrointestinal manifestations of their disease. With this increasing life expectancy, other complications of CF have become more prevalent, including low bone density and fractures.

Multiple studies have confirmed that adolescents and young adults with CF have lower areal bone mineral density (BMD) as measured by dual energy X-ray absorptiometry (DXA) and are at higher risk for fracture (1–5). In particular, patients with CF are at risk for rib and vertebral fractures, which can lead to significant morbidity and mortality (6, 7). An unusual pattern of bone turnover has also been described in this patient population, with higher than expected markers of bone resorption and lower than expected markers of bone formation (8, 9). Although the exact pathogenesis of CF-related bone disease is unclear, patients with CF possess multiple risk factors for low bone density, including pancreatic insufficiency, malabsorption, poor nutritional status, vitamin D deficiency, delayed puberty, glucocorticoid use, inflammation, and physical inactivity (10).

Since the late 1990s, awareness of CF-related bone disease has led to increased attention to bone density screening, optimization of nutritional status, and reduction of modifiable risk factors for low bone density. The Cystic Fibrosis Foundation and the European Cystic Fibrosis Society published guidelines in 2005 and 2011, respectively, addressing the importance of bone health in this population (10, 11). At the same time, establishment of CF Centers has occurred that focus on standardizing the care of CF patients and the development of new medications and interventions to improve respiratory function (12). Whether these improvements in clinical care have led to significant changes in bone health in patients with CF over the past 15 years is unclear.

The purpose of this study was to compare DXA-derived areal BMD in patients with CF evaluated 15 years ago with age-, race- and gender-matched patients with CF in the present day. We hypothesized that patients in the present-day cohort would exhibit improved BMD compared to the historic cohort, due to advances in clinical care and increased attention to bone and mineral status in these patients.

Materials and Methods

Subjects and Eligibility Criteria

Two cohorts of ambulatory patients with CF were evaluated in this study. In the historic cohort, adults with cystic fibrosis ages 18–50 years of age were recruited from the Massachusetts General Hospital (MGH) Cystic Fibrosis Center between November 1995 and June 1999. Exclusion criteria included history of solid organ transplantation and current pregnancy. This protocol was approved by the Institutional Review Board (IRB) at MGH. In the present-day cohort, subjects were recruited from the MGH and Boston Children’s Hospital (BCH) Cystic Fibrosis Centers between December 2011 and November 2013. Exclusion criteria were the same as for the historic cohort. This protocol was approved by the MGH IRB with ceded review by the BCH IRB. Written informed consent was obtained from all participants. A total of 52 subjects were enrolled in the historic cohort and 38 in the present day cohort. Thirty-two subjects in each cohort were able to be matched by age, race, and gender and were included in this study.

Clinical Assessments

In both the historic and present-day cohorts, study procedures were performed in the MGH Clinical Research Center. Subjects were surveyed regarding medical history, current medication use, history of fractures, pubertal and reproductive history, and lifetime glucocorticoid exposure. A registered bionutritionist assessed calcium and vitamin D intake using the validated NDS-R (13). The NDS-R version was updated in the years between the recruitment of the two study cohorts to reflect the most updated nutrient data available at the time. Medical records were reviewed to obtain pulmonary function tests and CFTR genotype. In all subjects, height was measured on a wall-mounted standiometer and weight on a standardized scale.

Laboratory Assessment

After fasting overnight, morning serum and urine samples were obtained in all subjects. In the historic cohort, the labs were processed by the MGH General Clinical Research Center, and in the present-day cohort, labs were processed through the MGH Clinical Laboratory Research Core. Urine N-telopeptide (NTX) was measured by competitive-inhibition enzyme-linked immunosorbent assay (Osteomark) in both cohorts. Serum 25-hydroxyvitamin D [25(OH)D] levels were measured by radioimmunoassay (DiaSorin in the historic cohort; Abbott Diagnostics in the present-day cohort). Bone specific alkaline phosphatase (BSAP) was measured using enzyme immunoassay (Metra Biosystems in the historic cohort; Quidel Corporation in the present-day cohort). Intact parathyroid hormone (PTH) was measured by immunoradiometric assay (Nichols Institute) in the historic cohort and by electrochemiluminescence immunoassay (Roche Diagnostics) in the present-day cohort. The reference ranges established by the manufacturers were the same for both cohorts.

Assessment of areal BMD

In both cohorts, DXA areal BMD of the lumbar spine in the posterior-anterior (PA) and lateral projections and the distal radius were measured in the MGH Bone Density Center. Scans were performed on a QDR4500A model in the historic cohort and a Discovery A model in the current cohort (Hologic Inc, Bedford, MA). When the QDR4500A model was replaced by the more recent model, calibration studies were performed to allow for direct comparisons.

Statistical Analysis

Statistical analysis was performed using SAS 9.3 software (SAS Institute Inc., Cary, NC). Subjects in the historic and present-day cohorts were matched by age (within 5 years), race, and gender to minimize effects of these covariates on bone outcomes. Characteristics between subjects in the two cohorts were compared using independent samples two-sided t-tests for normally distributed data or Wilcoxon rank sum for non-normally distributed data. Categorical variables were compared using Chi Square or Fisher Exact tests.

Because the laboratory manufacturers differed between the historic cohort and the present cohort for all assays with the exception urine NTX, we did not perform direct comparisons between lab values between the two cohorts. Instead, using manufacturer references ranges, BSAP and PTH results in each cohort were categorized normal or abnormal. Subjects were also categorized as vitamin D sufficient or deficient, as defined by a 25(OH)D level below 20 ng/mL (14). The vitamin D assays (DiaSorin and Abbott) have good correlation at 25(OH)D levels of less than 50 ng/mL, but the Abbott assay has systemic positive bias compared to DiaSorin at levels above than 50 ng/mL (15). Therefore, this bias was avoided by introducing the above categories. These categorical variables were then compared between cohorts using Chi Square or Fisher Exact test. The urine NTX method and manufacturer was the same in both cohorts, and these results were compared between cohorts using Wilcoxon rank sum for non-normally distributed data.

The primary outcome was PA spine areal BMD. General linear regression was used to compare BMD between groups before and after multivariable adjustment for pulmonary function as measured by percent predicted forced expiratory volume in one second (FEV1), presence of vitamin D deficiency, presence of secondary hyperparathyroidism, and prior steroid use. Covariates in the multivariable model were chosen based on clinical relevance as well as significance on the univariate screen.

To evaluate associations between clinical characteristics and areal BMD among all CF subjects (n=64), correlations were performed using Pearson correlation coefficient, or Spearman correlation coefficient in the case of non-normally distributed data. Clinical covariates included FEV1, BMI, glucocorticoid use, age at menarche in females, age at pubarche in males, calcium and vitamin D intake, urine NTX, and other laboratory values recoded into normal or abnormal categories. In addition, ANOVA was used to determine if areal BMD differed significantly between patients with CFTR genotypes containing zero, one, or two copies of the F508del mutation.

Based on the standard deviation of the primary outcome in the current cohort, this sample size has 80% power to detect at least a 7% difference in absolute BMD between cohorts. A smaller difference was not considered clinically meaningful. Data are reported as mean ± standard deviation (SD) unless otherwise noted, and p values ≤0.05 areconsidered statistically significant.

Results

Cohort Characteristics

The mean time difference between measurements in the historic cohort (evaluated between 1995 to 1999) and the present-day cohort (evaluated between 2011 to 2013) was 15.0 ± 1.3 years. Clinical characteristics of subjects in both cohorts are presented in Table 1. Age, height, weight, and BMI were similar between cohorts. Nine patients (28%) in the historic cohort and seven (22%) in the present-day cohort were underweight with a BMI <18.5 (p=0.56). A similar, small number of patients, 10 (31%) in the historic cohort and 8 (25%) in the current cohort (p=0.58), had a BMI at or above that recommended by the CF Foundation (16).

Table 1.

Clinical and Demographic features of Subjects with CF in the historic and present-day cohorts.

	Mean (SD), Median (range), or n (%)
	Historic Cohort (n=32)	Present-Day Cohort (n=32)	p-value
Age (yr)	27.6 (7.0)	27.5 (7.7)	0.94
Female	19 (59%)	19 (59%)	-
Caucasian Race	32 (100%)	32 (100%)	-
Height (cm)	167.0 (6.2)	166.3 (8.3)	0.71
Weight (kg)	57.7 (10.1)	58.5 (11.3)	0.76
BMI (kg/m2)	20.6 (2.7)	21.1 (3.7)	0.52
Percent predicted FEV1 (%)	53.9 (21.8)	67.5 (28.2)	0.039
Calcium intake (mg/day)	1927 (426–4762)	1836 (539–9795)	0.81
Vitamin D intake (IU/day)	910 (77–1501)	1397 (36–8279)	0.09
Pancreatic insufficiency	26 (81%)	27 (84%)	0.74
CFTR genotype
F508del/F508del	15 (47%)	11 (34%)	0.51
F508del/Other	12 (37.5%)	13 (41%)
Other/Other	5 (15.5%)	8 (25%)
Previous fracture	14 (44%)	13 (41%)	0.80
Age menarche (y)	13.6 (2.6)	12.7 (1.4)	0.16
Amenorrhea >6 months	5 (26%)	1 (3%)	0.18
Pregnancy	7 (22%)	5 (16%)	0.56
Age pubarche (males)	12.5 (11–19)	13 (11–15)	0.75
Use of oral glucorticoids
Current use	2 (6%)	1 (3%)	1.0
Previous use	14 (44%)	22 (69%)	0.029
Lifetime exposure (mo)	4.0 (8.7)	8.4 (10.4)	0.012
Current use of inhaled glucocorticoids	8 (25%)	22 (69%)	<0.001

BMI, body mass index; FEV1, % predicted forced expiration in 1 second.

Pulmonary function was moderately impaired in both cohorts, and the historic cohort had a significantly lower percent predicted FEV1 than the present-day cohort (p=0.03). There were no differences between cohorts in reported fractures, pubertal and reproductive history, the prevalence of pancreatic insufficiency, or the presence of the F508del mutation.

The average daily dietary and supplemental calcium and vitamin D intake was above the national Recommended Daily Allowance (RDA) (14) in both cohorts. Although not significant, there was a trend toward higher vitamin D intake in the present-day cohort than the historic cohort (Table 1). Although a similar small number of patients in both cohorts reported current use of oral glucocorticoids, fewer patients in the historic cohort reported ever using oral glucocorticoids (p=0.029), and the median duration of prior oral glucocorticoid use was significantly lower in the historic cohort (p=0.012). Current use of inhaled glucocorticoids was also significantly greater in patients in the present-day cohort (p=<0.001).

Laboratory Results

Laboratory results are presented in Table 2. Urine NTX levels were similar in both cohorts (p=0.83). A greater number of patients in the historic cohort had 25(OH)D levels less than 20 ng/ml than in the current cohort (48% vs. 22%, respectively, p=0.039). The seasons during which the 25(OH)D levels were drawn did not differ between the two cohorts (p=0.18). Serum PTH levels were above the normal range in 19% of patients in the historic cohort, whereas no patients in the present-day cohort had an elevated PTH level (p=0.026). No differences in normal or abnormal BSAP levels between cohorts were noted. No patients in either cohort had renal failure (data not shown).

Table 2.

Laboratory Results.

	Historic Cohort	Present-Day Cohort	Reference Range
PTH (pg/mL)	29 (11–95)	33 (16–60)	10–60
25(OH)D (ng/mL)	20 (1–68)	27 (12–91)	>20
Urine NTX (nM BCE/mM Cre)	67 (14–473)	54 (19–236)	3–65
BSAP (U/L)	21.3 (11–110)	27.9 (13–56)	Female 11.6–30.6 Male 15.0 - 41.3

Data expressed as median (range).

PTH, parathyroid hormone; 25(OH)D, 25-hydroxyvitamin D; NTX, N-telopeptide; BSAP, bone specific alkaline phosphatase

Areal BMD Results

There were no significant differences in unadjusted areal BMD measurements of the PA spine, lateral spine, or distal radius in the historic cohort compared to the present-day cohort. Multivariable adjustment accounting for percent predicted FEV1, vitamin D sufficiency, secondary hyperparathyroidism, and glucocorticoid use did not change the significance of these findings (Table 3).

Table 3.

DXA Areal BMD Results.

	Unadjusted			Multivariable-Adjusted
DXA areal BMD	Historic Cohort	Present-Day Cohort	p value	Historic Cohort	Present-Day Cohort	p value
PA Spine (g/cm2)	0.983 ± 0.034	0.981 ± 0.034	0.94	0.987 ± 0.022	0.968 ± 0.023	0.59
Lateral Spine (g/cm2)	0.780 ± 0.017	0.803 ± 0.019	0.37	0.771 ± 0.019	0.809 ± 0.024	0.26
1/3 Distal Radius (g/cm2)	0.723 ± 0.016	0.730 ± 0.012	0.68	0.731 ± 0.019	0.731 ± 0.014	0.99

Data are expressed as mean ± SE. Multivariable adjusted model includes the following covariates: FEV1, vitamin D sufficiency (defined as 25[OH]D>20ng/mL), secondary hyperparathyroidism (defined as elevated PTH level), and glucocorticoid use.

FEV1, % predicted forced expiration in 1 second; PTH, parathyroid hormone; 25(OH)D, 25-hydroxyvitamin D; PA, posterior-anterior; BMD, bone mineral density.

Z-scores were reported using the Hologic reference databases available at the time that each DXA scan was obtained. Twelve patients in the historic cohort and 11 patients in the present-day cohort had a Z-score of at least 1.0 standard deviation below the mean, and five and three patients in the historic and present-day cohorts, respectively, had a Z-score at least 2.0 standard deviations below the mean (Figure 1). These proportions were similar between cohorts (p=0.74).

Figure 1. Percent of patients with normal or reduced PA spine BMD Z-scores in the historic cohort and present-day cohort (p=0.74).

PA, posterior-anterior; BMD, bone mineral density.

Areal BMD of the spine was positively associated with BMI and negatively associated with urine NTX in all subjects (n=64) (Figure 2). Reported duration of prior glucocorticoid use, current use of inhaled steroids, pubertal history, calcium and vitamin D intake, and FEV1 were not significantly correlated with BMD. PA spine BMD was not associated with the presence of F508del.

Figure 2. Correlations between PA spine BMD and BMI (2a) and urine NTX (2b) among all subjects with CF.

PA, posterior-anterior; BMD, bone mineral density; BMI, body mass index; NTX, N-telopeptide; CF, cystic fibrosis

Discussion

Despite improvements in clinical care and life expectancy in patients with CF over the past 15–20 years, areal BMD as measured by DXA is no better in young adults with CF today than in an age-, race-, and gender matched cohort of young adults with CF evaluated in 1995–1999. Areal BMD was similar between these cohorts even after adjusting for the significantly improved pulmonary status and lower rate of vitamin D deficiency and secondary hyperparathyroidism of the present-day patients. These data suggest that further attention to optimizing bone health in CF patients, possibly by focusing on modifiable factors such as the avoidance glucocorticoids and improvement in nutritional status and BMI, will be important to prevent skeletal consequences related to CF.

Over the past 15–20 years, treatments aimed at improving nutritional status and slowing pulmonary decline in patients with CF have led to greater life expectancy and improved quality of life (17). The development of health care teams and CF Centers dedicated to the multidisciplinary treatment of these patients has allowed for standardization of care and utilization of best practice protocols. In addition, aggressive management of respiratory infections, improvements in therapies targeted at mucociliary clearance, and a push for early identification and management of non-pulmonary complications have also contributed to improved clinical care of patients with CF (12).

Despite these clinical improvements, CF-related bone disease remains a concerning clinical problem leading to significant morbidity. In a recent meta-analysis reviewing studies published prior to 2008, the pooled prevalence of osteoporosis and osteopenia in adults with CF were 24% and 38%, respectively, and the incidence increased with age (18). Guidelines published in 2005 (10) and 2011 (11) recommend routine DXA screening, optimization of calcium and vitamin D intake, and aggressive nutritional interventions to maintain adequate body weight.

To our knowledge, this is the first study directly comparing BMD in two cohorts of age-, gender- and race matched cohorts of ambulatory patients measured over a decade and a half apart on calibrated DXA machines. Pulmonary function was significantly better in the present-day cohort, and fewer patients in the current cohort had vitamin D deficiency or secondary hyperparathyroidism. For these reasons, we expected that bone density in the current cohort would be better than in the historic cohort. However, areal BMD in young adults with CF today is identical to matched patients measured 15 years ago. A significant and similar number of patients had reduced bone density Z-scores below -1.0, and almost half of patients in each cohort reported a prior fracture.

There are several possible explanations for this lack of improvement in BMD. First, as in previous studies (19–22), we found a significant correlation between BMI and BMD in patients with CF. Despite increased attention to nutritional status over recent years, BMI was similar between the two cohorts. Although the CF Foundation recommends that women attain a BMI of at least 22 kg/m2 or greater and men at least 23 kg/m2 (16), only a minority of patients in either cohort met that goal. Difficulty achieving nutritional goals remain as significant a problem today as 15 years ago, and this issue may help to explain the similar BMD findings between cohorts.

In addition, more patients in the present-day cohort reported prior oral glucocorticoid use and current inhaled steroid use, likely related to differences in treatment practices between the two cohorts. Systemic and inhaled glucocorticoids can be detrimental to bone (23–25). Although the duration of oral glucocorticoid use and current use of inhaled steroids were not correlated with BMD, the greater exposure to glucocorticoids in the current cohort may help to explain the lack of improvement in BMD compared to the historic cohort. These data support the recommendations to minimize glucocorticoid use whenever possible (10, 11).

Other potential factors that may affect bone health in patients with CF include delayed puberty, hypogonadism, and pancreatic insufficiency. Patients in both cohorts reported an overall normal age at menarche and pubarche, and menstrual and reproductive history in females was not significantly different. Similar numbers of patients also reported pancreatic insufficiency treated with pancreatic enzyme replacement. The similarity of these characteristics may have contributed to the equivalent BMD between cohorts.

Abnormalities in bone turnover have been described in patients with CF, with high bone resorption and low bone formation (8–10). In both the historic and current cohorts, urine NTX, a marker of bone resorption, was similar and exceeded the normal range in 42% of patients. Urine NTX was also significantly and negatively correlated with areal BMD of the spine. These results are consistent with previous studies (8, 9) and suggest that increased bone resorption plays a prominent role in CF-related bone disease, past and present.

The cause of this striking uncoupling of bone turnover has not been fully elucidated and raises the intriguing possibility that CF-related bone disease may be in part due to CFTR dysfunction itself (26–28). CFTR is expressed in human bone cells (29) and may affect osteoclast activation by altering osteoprotegerin production (27, 30). Other potential factors that may increase bone resorption could include underlying inflammation, chronic respiratory acidosis, hyperglycemia, and reduced physical activity. The fact that average BMD in patients with CF has not changed in 15 years despite significant improvements in clinical care and life expectancy could support the hypothesis that these underlying factors may be playing an independent role in CF-related bone disease. This also raises the possibility that improving CFTR function with new CFTR modulators may also improve bone health in CF patients in the future.

Strengths and limitations of this study merit consideration. Our study has several important strengths, including the detailed clinical information regarding factors affecting skeletal health in patients with CF prospectively collected in each cohort. In addition, both cohorts of patients were relatively homogenous, limited to young, ambulatory adults prior to onset of severe lung disease and transplantation. Importantly, the use of calibrated DXA machines at a single center allowed for direct comparisons of BMD between the two cohorts despite measurements occurring 15 years apart. Our study also has several important limitations. The sample size was relatively small, and it is possible that smaller, albeit less clinically meaningful, differences in BMD could be uncovered if a larger number of patients were evaluated. In addition, BMD of the proximal femur was not assessed in the historic cohort. Laboratory methods and manufacturers have also changed over the past 15 years, and this limited our ability to make direct comparisons between laboratory results between the two cohorts. In addition, serum 25(OH)D was measured by RIA rather than LCMS, which was not available in the historic cohort. Finally, the cross-sectional design of this study cannot capture the dynamic changes that occur in the skeleton nor account for the effects of clinical variables that may have occurred in the years prior to this evaluation.

In conclusion, young adults with CF evaluated in late 1990s had similar areal BMD of the spine and distal radius as age-, gender- and race-matched patients measured 15 years later. Despite improved pulmonary function and lower prevalence of vitamin D deficiency and secondary hyperparathyroidism in the present-day cohort, a similar and significant number of patients had low BMD Z-scores and reported fractures. These results suggest that attention to bone health, including aggressive nutritional interventions to optimize BMI and attempts to minimize glucocorticoid use, may be important in preventing this complication. Further studies will be required to improve our understanding of the prevention and treatment of CF-related bone disease.

Highlights.

• Osteoporosis is an increasingly prevalent and burdensome complication of CF.

• It is unclear if advances in clinical care over the past 15 years have improved bone health.

• We compare bone density in adults with CF in the late 1990s with patients evaluated today.

• Bone density in adults with CF is no better today than it was 15 years ago.

• Strategies to optimize bone health will be important to prevent this complication.

Abbreviations

DXA

dual energy X-ray absorptiometry

BMD

bone mineral density

CF

cystic fibrosis

PTH

parathyroid hormone

25(OH)D

25-hydroxyvitamin D

NTX

N-telopeptide

BSAP

bone specific alkaline phosphatase