A retrospective analysis of longitudinal changes in bone mineral content in cystic fibrosis

What’s known on this subject

Older children and adolescents with cystic fibrosis may have poorer bone mineral content than younger patients.

What this study adds

• Bone mineral content decreases with time in children with CF

• The reduction is associated to low FEV1% and low body mass index standard deviation scores (BMI SDS)

• Low vitamin D may also play an important role when associated with a high parathyroid hormone (PTH)

Description of subjects

In a retrospective analysis of 289 children with CF, receiving either full care or shared care at the Royal Hospital for Sick Children Glasgow (RHSC) and one of its affiliated hospitals, we selected children that had at least one DXA scan performed over a period of 10 years (January 2003–June 2013). Children were included if they had the diagnosis of CF by molecular diagnosis, were aged between 8 and 18 years and had at least one DXA assessment performed in the RHSC. Children were excluded if clinical data were not available. Supplemental feeding was defined as the child receiving enteral feeding at a minimum frequency of once a week in the month prior to the scan. Oral/inhaled corticosteroid was defined as corticosteroid treatment during the year previous to the scan. Forced expiratory volume in 1 s percent (FEV1%) was used as a marker of pulmonary function and was assessed using a spirometer (Jaeger Masterscreen Body, CareFusion, San Diego, CA, USA). Impaired glucose tolerance was defined, according to the American Diabetes Association guidelines, as a fasting plasma glucose level of less than 7 mmol/L and a 2-h plasma glucose level of 7.8 to 11 mmol/L. Diabetes was diagnosed when a fasting glucose level was 7 mmol/L or higher or a 2-h plasma glucose level of more than 11 mmol/L [11].

Auxology

Anthropometric data were collected from the clinical records and height and BMI were expressed as SDS using the UK growth reference [12]. Given the lack of availability of puberty data, age at peak height velocity (PHV) was used as a marker of pubertal progress [13]. The mean age increment was 1.2 years (range 0.42, 1.86) and the height gain (cm) was divided by this age increment for each year of age from 9 to 16 years, to obtain height gain per month and estimate the annualised height velocity.

Segregation status

The segregation status of children was assessed by categorisation into three classes: chronically colonised with Pseudomonas aeruginosa (PSA) or Burkholderia cepacia (CA) intermittently colonised with PSA and PSA naïve. Chronic colonisation with PSA or BC was defined as PSA or BC in sputum in the past 6 months, with three or more isolates in the past 12 months post-eradication. Intermittent colonisation with PSA was defined as no PSA in current sputum cultures and clear of PSA for more than 6 months, but intermittent isolates in past; or currently has PSA in sputum or in the past 6 months but cannot yet be classed as chronically colonised.

Bone biochemistry

Biochemical data at the time of the DXA scan were collected from local records. PTH was analysed by chemiluminescent microparticle immunoassay (CMIA) on an Abbott Architect ci8000 (ABBOTT, Wiesbaden, Germany). Serum PTH levels≥7.5 pmol/L (70 pg/mL) were considered to be significantly elevated. Vitamin D (25-hydroxyvitamin D2 and 25-hydroxyvitamin D3) status was measured by liquid chromatography tandem mass spectrometry (LC-MS/MS) using commercial standard in human serum (Chromsystems, München, Germany). Using the British Paediatric and Adolescent Bone Group’s position statement, vitamin D deficiency was defined when serum vitamin D (25OHD) levels were<25 nmol/L, insufficiency when levels were 25–50 nmol/L and sufficiency at a concentration greater than 50 nmol/L [14].

DXA

DXA of the lumbar spine (LS, L2–L4) and total body (TB) using Lunar Prodigy (GE Medical Systems, Waukesha, WI, USA) was performed as part of routine clinical practice. As outlined in previous studies [15], [16], using the manufacturer’s software, absorptiometry was related to bone mineral content (BMC) expressed in grams of ashed calcium hydroxyapatite equivalents and to grams per centimetre squared of BMD corresponding to BMC divided by the bone area scanned. Lunar software calculated an ethnicity; age- and gender-matched BMD SDS using a reference population of more than 2,000 American children between the ages of 5 and 19 years evaluated by Lunar Prodigy. The reference data are comparable to recently published data on Dutch children by van der Sluis et al. [17]. They provide age- and gender-matched BMC and bone area values for each age and gender band between 5 and 19 years. By linear interpolation, these data allowed calculation of a predicted and a percentage predicted bone area for age and gender (pp bone area). A child with short stature should have small bones and would have a low pp bone area. The reference data also allows a comparison of the actual BMC of the patient with the predicted BMC for a subject of the patient’s gender and bone area. This allows the calculation of a BMC SDS. A child with osteopenia should have thin bones, irrespective of whether the bones are small or large, and would have a low BMC SDS.

Ethics

As the study simply analysed data that was routinely collected as part of the clinical care of the patient, the study did not require informed consent and was approved as an evaluation of health care delivery.

Statistical analysis

Descriptive statistics are outlined for all variables assessed. Means and standard deviations (SDs) were calculated for all normally distributed continuous measures. Data were analysed with SPSS 20 (SPSS Inc, Chicago, IL, USA). Baseline (T0) was assigned at the time of the first DXA assessment for each child while T1 and T2 were noted as the time when each child had a second and third DXA assessment. The Student’s t-test and repeated-way ANOVA for three samples were used for comparing variance. Hierarchical regression models (backward method) were used to assess the factors that influence baseline LS BMC SDS and also the factors that influence change in bone mass. The difference between two DXA assessments was calculated as ∆ LS BMC SDS=T1−T0 and T2−T1. Further, as the time between DXA assessments was variable, this difference was annualised by dividing it with the time between assessments in decimal years. This annualised ∆ LS BMC SDS was used as the dependent variable in hierarchical regression model (backward method) to assess longitudinal change in BMC. The predictors of change in BMC were considered in variables measured at baseline, T1, T2 and were calculated as differences, (i.e. ∆ FEV1%=T1 FEV1%−T0 FEV1%), for all variables.

Anthropometry

Descriptive clinical data for the children with CF at baseline are presented, in Table 1, for all children and separately the group that had a single DXA assessment (n=52) versus those who had repeated DXA assessments (n=48). On an average, DXA assessments were performed 1.8 years apart (SD=0.7). Significant differences were found between boys and girls at baseline in mean age at PHV, mean PHV and mean height SDS as seen in the general population.

Vitamin D, PTH status

Of the 94/100 children where a vitamin D level was available at baseline, 24% were deficient, and 39% were insufficient (Figure 1). Of the 84/100 children where a PTH level was available, 14% children had a value above 7.5 pmol/L (70 pg/mL). Vitamin D and PTH values were available simultaneously for 81/100 children. Of them, 49% had low vitamin D (insufficient and deficient) but normal PTH, while 12% had low vitamin D and high PTH.

Figure 1:

Baseline bone mass, vitamin D and PTH status.

LS BMC SDS, lumbar spine bone mineral content standard deviation score; vitamin D deficiency <25 nmol/L, insufficiency 25–50 nmol/L, normal ≥50 nmol/L. High PTH levels if≥7.5 pmol/L, normal PTH if <7.5 pmol/L. Vit D, N normal vitamin D; Vit D, ↓ if vitamin D deficient and insufficient; PTH, N if <7.5 pmol/L; PTH, ↑ if ≥7.5 pmol/L. The number in figures show number of patients out of the total of 100 for each column, in a specific category.

On longitudinal analysis, 27% of the children were vitamin D deficient on one occasion, 17% on two occasions and none on all three occasions (Figure 2A, B, C). Thirty-four percent of the children were vitamin D insufficient on one occasion, 27% on two occasions and 5% on three occasions.

Figure 2:

The dynamics of vitamin D and LS BMC SDS at T0, T1 and T2.

(A–C) The dynamics of vitamin D at T0, T1 and T2 separated in (A), (B) and (C), in regards to baseline vitamin D status. T₀, baseline, first DXA assessment; T₁, second DXA assessment; T₂, third DXA assessment; vitamin D deficiency defined when vitamin D levels were<50nmol/L (20ng/mL), insufficiency if the vitamin D levels were between 52.5 and 72.5nmol/L (21–29ng/mL), and normal if vitamin D level ≥75nmol/L (30ng/mL). (D–F) The dynamics of LS BMC SDS at T0, T1 and T2 separated in (D), (E) and (F), in regards to baseline status in LS BMC SDS, lumbar spine bone mineral content standard deviation score. Three classes of LS BMC SDS: >0.5; between −0.5 and 0.5, and ≤−0.5.

Bone mineral content

In the 100 children, baseline LS BMC SDS was more than 0.5 in 13% of the children; 50% had LS BMC SDS between −0.5 and 0.5, while 37% had SDS ≤−0.5 (Figure 1). Ten children (10%) had LS BMC SDS below −1 at baseline. In longitudinal analysis (across 3 DXA assessments) the mean LS BMC SDS varied from T0 to T1 and T2 (−0.51; −0.64; −0.56; p=0.178 repeated way ANOVA). In the 37 children who had a baseline LS BMC SDS >−0.5, 29 (78%) assessments showed a fall in BMC SDS compared to previous assessment. Conversely, in the 35 children who had a poor baseline LS BMC SDS<−0.5, 12 (35%) assessments showed a decline in BMC SDS compared to previous assessment (Figure 2D, E, F).

Determinants of initial bone mineral content

Significant determinants of basal LS BMC SDS were identified using hierarchical multiple regression (backward method). The significant determinants of initial LS BMC SDS were height SDS and supplemental feeding (coded 0 and 1 – for presence of supplemental feeding). All determinants correlated inversely with LS BMC SDS (Table 2). Variables that did not reach statistical significance were: age, gender, age at peak height velocity and peak height velocity (in centimetres), initial BMI SDS, vitamin D level, PTH, FEV1%, oral corticosteroids and impaired glucose tolerance.

Determinants of change in bone mineral content

To identify predictors of change for bone mineral content, annualised ∆ LS BMC SDS was used as dependent variable in a hierarchical multiple linear regression (backward method).

The significant determinants of initial LS BMC SDS were ∆ FEV1%, second BMI SDS, and initial vitamin D level. All determinants correlated directly with annualised ∆LS BMC SDS, as shown in Table 2. Variables that did not reach statistical significance were: age, gender, peak height velocity (in centimetres), age at peak height velocity, initial BMI SDS, second vitamin D level, initial and second FEV1%, initial, second and change in PTH, oral corticosteroids, supplemental feeding, impaired glucose tolerance and segregation status.