The study was conducted according to the statement of the Declaration of Helsinki and approved by the Ethical Board Committee of Valduce Hospital - Villa Beretta Rehabilitation Centre, Lecco, Italy.
Patients
Patients were firstly enrolled among the members of As.It.O.I., the Italian Association of Osteogenesis Imperfecta. As.It.O.I., by posting an announcement of the association Facebook page.
Inclusion criteria were: confirmed diagnosis of OI type III and IV, stable condition, absence of severe cardio-respiratory pathologies, willingness to participate to the study and to travel to Milan for the tests. Informed consent was obtained from all study subjects or parents.
Quality of Life Questionnaire
All patients filled in an Osteogenesis Imperfecta specific Quality of Life Questionnaire (OIQoL) [24,25] comprising 33 questions grouped in six main themes (being safe and careful, reduced function, pain, fear, fatigue, independence). Each question was given a score ranging from 0 (worst scenario) to 4 (best scenario). For each theme, a global score was assigned as the mean of the corresponding questions.
Spinal and ribcage deformity
Thoracic scoliosis was assessed through standard antero-posterior and lateral radiographic views of the entire spine and quantified using the Cobb method[26,27].
The angle subtended at the sternal level on the transversal plane was computed to quantitatively describe the geometrical deformity of the ribcage, as previously described[8].
The spinal and ribcage deformity were combined by computing the mean value among the transversal and sagittal sternal angles and the Cobb angles of the thoracic and lumbar scoliosis.
Nocturnal oxygen saturation
Median and nadir nocturnal oxygen saturation (SpO2) was measured using a digital pulse oximeter (Nonin, 8500 digital pulse oximeter Quitman, TX). The number of apnoea and hypopnea events per hour of sleep (AHI index) as well as the number of times per hour of sleep of oxygen desaturation (ODI index) were calculated. OSA was diagnosed as mild when 5≤AHI<15; moderate when 15≤AHI<30; severe when AHI≥30[28].
Lung function and breathing pattern at rest
Measurement of forced vital capacity (FVC) and forced expiratory volume in 1 second (FEV1) were assessed through spirometry; while total lung capacity (TLC) through the nitrogen washout technique (Vmax series 22, SensorMedics, Yorba Linda,CA).
Accurate assessment of ventilatory and thoraco-abdominal pattern during spontaneous quiet breathing was assessed by opto-electronic plethysmography (OEP System; BTS, Milan, Italy) in both seated and supine position (figure 1). Breathing frequency, tidal volume, minute ventilation (i.e. the product of the two), rapid and shallow breathing index (i.e. the ratio between breathing frequency and tidal volume), ribcage contribution to tidal volume (DVRC) and thoraco-abdominal phase shift angle were computed on an average breath of at least 45 seconds of spontaneous breathing. The abdominal volume at end expiration was also computed and expressed as percentage of total chest wall volume.
Because the output of OEP is the position of passive reflective markers put according to anatomical points, the trunk height was calculated as the distance between the marker at the sternoclavicular joint and the marker below the naval at the level of the iliac crest.
Nutritional status
Anthropometric measurements
Anthropometric measurements were taken by the same operator, according to conventional criteria and measuring procedures[29]. Body weight (BW, kg) and body height (BH, cm) were measured to the nearest 0.1 kg and 0.5 cm, respectively. In subjects able to maintain the upright position, BW was measured on a beam balance scale and BH was measured with a stadiometer; in non-ambulant subjects unable to maintain the upright position, BW was measured using a digital wheelchair scale and BH was obtained measuring the supine length. BMI was calculated as BW (kg)/BH2 (m2).
Waist circumferences were measured midway between the lower rib margin and the superior anterior iliac spine using a horizontally applied non-stretch tape, and it was measured to the nearest 0.5 cm.
Dual-energy X-ray absorptiometry
DEXA scans (iDXA; General Electric, formerly Lunar Corp.,Madison, WI) were used to obtain the body composition. DEXA provides measurements of soft tissue and bone for the total body and the sub regions (namely arm, trunk, leg) including fat mass (FM), lean body mass (LM) and bone mineral content (BMC). The fat mass percentage (FM%) was obtained as the ratio between total body fat mass and total body mass (fat + lean mass + bone mass of total body) multiplied for 100. The fat free mass (FFM) was calculated by adding BMC to LM. Fat mass (FMI) and lean mass indexes (LMI) were calculated by dividing FM and LM by the squared height, respectively. To investigate the proportion between LM and FM their ratio was also calculated.
The trunk region starts at the inferior edge of the chin and the lower borders intersect the middle of the femoral necks without touching the brim of the pelvis. It includes the neck, chest, abdominal and pelvic areas[30].
The scanning was done with patients lying supine on the table, their feet in a neutral position and arms resting along their sides, palms facing upwards. The DEXA scans, performed by well-trained and certified research staff, were all done using one device and the same software (enCORE, 2010), for an average measuring time of 10 min. The exposure to radiation was <7 mSv. Daily quality-assurance was tested according to manufacturer directions. The DEXA scans were analysed using a custom-made software that allows body composition measurements in close relation to metal orthopaedic implants, by exclusion of no-osseous pixels.
Adherence to the Mediterranean Diet and eating behaviour
Adherence to the traditional Mediterranean diet was assessed using a validated 14-item questionnaire[31]. The guidelines set out by the Prevención con Dieta Mediterránea (PREDIMED) study group (www.predimed.es) with some adaptation already employed in previous studies were used to obtain the Mediterranean score (MED score)[32,33]. One point was attributed for each of the following: (1) olive oil as the main cooking fat; (2) olive oil ≥ 4 tablespoons/day; (3) vegetables > 2 servings/day (or ≥ 1 portion raw or salad)[31]; (4) fruit ≥ 3 servings/day; (5) red or processed meat < 1 serving/day; (6) butter or cream or margarine < 1/day; (7) sugar-sweetened beverages < 1/day; (8) wine ≥ 3 glasses/week; (9) legumes ≥ 3 servings/week; (10) fish/seafood ≥ 3 servings/week; (11) commercial sweets and confectionery < 3/week; (12) nuts ≥ 1/week; (13) white more than red meats (yes); and (14) use of sofrito ≥ 2/week. Participants with a MED score ≥ 9 points were considered as complying with a dietary pattern in accordance with the Mediterranean diet[31].
Eating behaviour was tested through the Binge Eating Scale (BES), a 16-item self-report questionnaire designed to capture the behavioural, the cognitive and the emotional features of regular objective compulsive overeating [34].
Statistical analysis
Mann-Whitney rank sum test was used to assess differences between the two groups that resulted non-normal distributed according to Kolmogorov-Smirnov test, with OSA status as independent factor (SigmaStat version 11.0; Systat Software, San Jose, Calif., USA).
Significance was set as p<0.05. Data in the results section of the text are reported as median value (25th and 75th percentiles).
For all the parameters that resulted significantly different between the two groups of patients, the receiver operating characteristic (ROC) curve was computed to illustrate their diagnostic ability in our binary classifier system.
The area under the ROC curve (AUC) was also calculated. AUC is an evaluation metrics for checking any classification model’s performance. The higher AUC, the better the model is at distinguishing between patients with OSA and no-OSA. AUC is considered excellent for values between 0.9-1, good between 0.8-0.9, fair between 0.7-0.8, poor between 0.6-0.7 and failed between 0.5-0.6[35,36].
Finally, the threshold of each considered parameter, which corresponds to the optimal operating point of the ROC analysis, was also derived.