Cardiopulmonary Status in Adults with Osteogenesis Imperfecta: Intrinsic Lung Disease May Contribute More Than Scoliosis

Abstract

Background

Osteogenesis imperfecta (OI) is a heterogeneous group of collagen-related disorders characterized by osteopenia, bone fractures, spine deformities, and nonskeletal complications. Cardiopulmonary complications are the major cause of morbidity and mortality in adults with OI. The cause of such problems was often attributed solely to the presence of large scoliosis curves affecting pulmonary function and, indirectly, cardiovascular health. However, recent studies suggest this may not be the case. Therefore, determining the relationships and causative agents of cardiopulmonary problems in patients with OI, specifically pulmonary impairment, is important to improving the overall wellbeing, quality of life, and survival of these patients.

Questions/purposes

(1) Is cardiopulmonary fitness in OI solely related to the presence of scoliosis? (2) What is the prevalence of heart and lung complications in this adult population? (3) Does the presence of pulmonary impairment impact quality of life in adults with OI?

Methods

This is a prospective observational cross-sectional study. Within 1 year, each participant (n = 30) completed pulmonary function testing, echocardiogram, ECG, chest CT, AP spine radiography, and quality-of-life assessments (SF-36, St. George’s Respiratory Questionnaire, Functional Outcomes of Sleep Questionnaire, and Pittsburgh Sleep Quality Index). In terms of pulmonary function, we differentiated restrictive and obstructive physiology using the ratio of forced expiratory volume over one second to forced vital capacity (FEV1/FVC), with restrictive lung physiology defined as FEV1/FVC > 0.8 and obstructive lung physiology as FEV1/FVC < 0.7. Spine radiographs were evaluated for scoliosis. Chest CT images were reviewed to qualitatively assess the lungs. The statistical analysis involved a Kruskall-Wallis test with Bonferroni’s correction and a bivariate correlation analysis using Spearman’s rho correlation coefficient (p < 0.05).

Results

Sixteen of 23 participants with restrictive lung physiology had scoliosis; their ages ranged from 19 years to 67 years. There was no correlation between the magnitude of the scoliosis curve and deficient pulmonary function (R = 0.08; p = 0.68). Seven participants had normal pulmonary function. The average scoliosis curve was 44 ± 29°. Thirteen participants had abnormal ECG findings while 10 had abnormal echocardiogram results. All but two individuals with abnormal chest CT results were found to have bronchial wall thickening. There were no differences in pulmonary or cardiac findings between OI types, except for FVC and total lung capacity, which were lower in individuals with Type III OI than in those with other types of OI. FEV1/FVC correlated with St. George’s Respiratory Questionnaire (R = 0.429; p = 0.02) but not with Functional Outcomes of Sleep Questionnaire (R = -0.26; p = 0.19) or SF-36 scores (physical component summary: R = -0.037, p = 0.85; mental component summary: R = -0.204, p = 0.29).

Conclusions

The lack of a relationship between decreased pulmonary function and the severity of scoliosis suggests that restrictive lung physiology in this population is likely because of factors intrinsic to OI and not entirely because of thoracic cage deformities. The fact that pulmonary impairment influences self-perceived quality of life exemplifies how detrimental such complications may be to everyday functioning. This also reinforces the importance of determining the underlying cause of cardiopulmonary impairment in this population to set clear clinical guidelines of care.

Level of Evidence

Level II, prognostic study.

Introduction

Osteogenesis imperfecta (OI) is a group of heritable connective tissue disorders with varied phenotypic and genotypic presentations [20, 35]. Its prevalence is approximately 1 in 15,000 births [35]. Although a number of genes directly or indirectly affect type I collagen metabolism resulting in OI, autosomal dominant mutations in type I procollagen genes, Col1α1 and Col1α2, are detected in approximately 90% of affected individuals [13, 20, 22]. The Sillence classification scheme divides OI into four types based on phenotypic severity: mild (Type I), moderate (Type IV), severe (Type III), and perinatally lethal (Type II) [32, 37]. The hallmark of OI is osteopenia and the subsequent susceptibility to fracture. Nonskeletal complications may include generalized deficits in dental, hearing, and, perhaps most importantly, cardiopulmonary health [10, 36].

Cardiopulmonary impairment is the major cause of morbidity and mortality in individuals with OI [9, 21]. Pulmonary complications range from pulmonary hypoplasia causing neonatal death to restrictive lung disease to pulmonary hypertension. Pulmonary impairment, commonly caused by restrictive lung physiology in patients with OI, may cause shortness of breath, fatigue, increased susceptibility to lower-respiratory infections, and sleep apnea [30, 36]. Over time, chronic pulmonary insufficiency may progress to cardiac issues such as cor pulmonale [1, 24, 25]. Mild tricuspid regurgitation, aortic root dilation, and increased ventricular mass are also commonly reported complications in adults with OI [24, 25, 29]. Unfortunately, the etiology of the cardiac and pulmonary manifestations of OI remain unclear, and guidelines for clinical management are vague [17, 23, 27]. To provide care, it is essential that the underlying causes of cardiopulmonary issues in OI are understood.

Pulmonary deficit may often be the result of chest wall abnormalities [15, 16, 39]. In the general population, pulmonary compromise is strongly correlated with thoracic scoliosis of > 60°, while individuals with smaller curves generally do not present with compromised lung function [16, 26, 39]. Pectus excavatum can affect the function of the heart and lungs and pectus carinatum may affect the function of respiratory muscles and overall ventilation [15, 18]. In OI, hyperkyphosis, scoliosis, rib fractures, and pectus carinatum affect the structure of the chest wall, and movement of the diaphragm might be restricted by upward compression of abdominal contents because of short stature [18, 30]. These physical differences may result in poor airway clearance that can often cause an increased infection risk [15, 30]. Even so, the presence of pulmonary issues may not depend on the presence of such physical features. Because scoliosis is a common OI complication, it was initially suggested that pulmonary compromise observed in the OI population was likely because of its presence [8, 14, 39, 42]. However, recent studies proposed that restrictive lung disease is not exclusive to individuals with severe spinal deformities [18, 42, 43]. Some individuals with OI who have mild or no scoliosis have similar pulmonary outcomes as those with more severe scoliosis [4, 43]. The presence of restrictive lung disease in individuals with mild OI suggests that pulmonary impairment in those with OI may not be secondary to scoliosis but due to the underlying abnormal collagen composition in lung parenchyma and airways. Although there has been a greater focus on extraosseous OI complications recently, there has been a relative dearth of published studies regarding cardiopulmonary abnormalities in this population. We have previously shown that the presence of scoliosis may not solely determine respiratory impairment in individuals with OI [4].

Building on this research, in the current study, we asked: (1) Is cardiopulmonary fitness in OI solely related to the presence of scoliosis? (2) What is the prevalence of heart and lung complications in this adult population? (3) Does the presence of pulmonary impairment impact quality of life in adults with OI?

Patients and Methods

Study Design

Adults 18 years and older with a clinical and/or genetic diagnosis of OI were enrolled in this prospective cross-sectional, observational study between August 2017 and May 2019 during their standard-of-care visit at the Kathryn O. and Alan C. Greenberg Center for Skeletal Dysplasias at Hospital for Special Surgery, New York, NY, USA. To avoid confounding results and selection bias, individuals were excluded from the study if they had a respiratory illness within 6 weeks of enrollment or were undergoing diagnostic studies for an active respiratory illness. Within 1 year, participants completed pulmonary function tests, spine radiography (AP and lateral), chest CT, ECG, echocardiogram, and ultrasound (abdominal and carotid).

Pulmonary function test measures included lung volumes assessed by body plethysmography, spirometry, and diffusing capacity of the lung for carbon monoxide. The primary values evaluated included the ratio of forced expiratory volume in 1 second to forced vital capacity (FEV1/FVC), diffusing capacity of the lung for carbon monoxide corrected for altitude, hemoglobin, and alveolar volume, and total lung capacity, a measure of the maximum air capacity of the lungs. Predicted values for participants with a short stature was calculated using arm-span measurement instead of height.

Scoliosis (≥ 10°) was measured by a single reviewer (CLR) using the Cobb method [7]. If more than one curve was present, the largest curve was used for analysis. Chest CT images were qualitatively evaluated by two reviewers (RAS, DD). Normal bronchial wall thickness-to-airway diameter ratio on CT is 0.2 [3]. This criterion was used as a reference point to identify bronchial wall thickening in our cohort.

Participants completed a variety of quality-of-life assessments including the SF-36, St. George’s Respiratory Questionnaire, Pittsburgh Sleep Quality Index, and Functional Outcomes of Sleep Questionnaire [5, 11, 40, 41]. Basic demographic and health information was obtained, including OI type, age, sex, ambulatory status, physical activity level, assistive device use, genotype, pulmonary comorbidities, cardiac comorbidities, and bisphosphonate and other bone-related treatment history.

Due to the small study population, missing data outside of pulmonary function tests and scoliosis radiographs were ignored. Nonparametric statistical testing included a Kruskal-Wallis test with Bonferroni’s correction, Pearson’s chi-square test, and a bivariate correlation analysis using Spearman’s rho coefficient. Results were considered significant at p < 0.05. All statistical analyses were conducted using SPSS version 22 for Windows (IBM Corp, Armonk, NY, USA).

Study Population

Thirty-four individuals enrolled in the study, but four were excluded (three were lost to follow-up and did not complete minimum study requirements, one underwent genetic testing and was found to not have a diagnosis of OI). Thirty individuals (average age 39 years, range 19 to 67) were included because they, at minimum, completed a pulmonary function test and had a scoliosis radiograph. Nine of 30 patients had Type I OI, eight had Type III OI, 12 had type IV OI, and one had Bruck syndrome-2. Sixteen of 30 patients had genetic confirmation (Table 1). Most participants were female (24 of 30). Nineteen of 30 individuals were short-statured (< 147 cm; the average height was 135 ± 27 cm). Four of 30 participants reported having asthma, and three had sleep apnea. Four of 30 participants had diagnosed cardiac comorbidities, including coronary artery disease and/or high blood pressure. Two of 30 participants were smokers. Sixteen of 30 individuals used an assistive device (10 used power wheelchairs, two used manual wheelchairs, three used canes, and one used underarm crutches).

Table 1.

Study participant diagnoses

Study ID	OI Type	Gene
1	III	No testing
2	I	COL1A1
3	IV	No testing
5	III	COL1A2
6	I	COL1A1
7	III	COL1A1
8	BRUCK	PLOD2
9	IV	COL1A2
10	I	No testing
11	IV	COL1A1
12	III	No testing
13	III	No testing
14	IV	COL1A1
15	I	No testing
16	IV	COL1A2
17	IV	COL1A2
18	IV	No testing
19	I	COL1A1
20	I	No testing
21	IV	No testing
22	I	COL1A1
23	IV	COL1A1
24	IV	COL1A1
25	IV	No testing
27	I	COL1A1
28	III	No testing
29	IV	No testing
30	I	No testing
32	III	COL1A2
34	III	No testing

Results

Seventy percent (21 of 30) of the population had scoliosis, with the average largest curve measuring 31 ± 31°). The largest measurable curve was in either the thoracic (55%) or thoracolumbar regions (45%). Average curve measurements and pulmonary function test values for the groups varied (Table 2). Individuals with Type III OI had larger curves than those with Type I OI, but not larger than those with Type IV (Table 3). Seventy-seven percent of participants (23 of 30) had restrictive lung physiology (FEV1/FVC ≥ 80%). The average forced vital capacity was 2 ± 1 L, which was 83 ± 29% of the predicted values. The average total lung capacity was 4 ± 2 L. Forty-three percent (13 of 30) of the study participants had lower total lung capacity than the reported reference value, which was based on the participant’s age, height, sex, and race. Forced vital capacity and total lung capacity were lower in individuals with Type III OI than in those with Types I and IV (Table 3). There was no significant difference in FEV1/FVC or percent reference diffusing capacity of the lung for carbon monoxide corrected for altitude, hemoglobin, and alveolar volume between OI Types (Table 3). Of those with scoliosis, 70% (16 of 23) had restrictive lung physiology (FEV1/FVC ≥ 80%). However, FEV1/FVC did not correlate with scoliosis curve magnitude (R = 0.08; p = 0.68). All six individuals with scoliosis curves > 60° had restrictive lung physiology. The magnitude of the scoliosis curve correlated with forced vital capacity (R = -0.753; p < 0.001) and total lung capacity (R = -0.763; p < 0.001).

Table 2.

Scoliosis and pulmonary function test measurements

Measure	Type I	Type III	Type IV	Pulmonary comorbidities	No pulmonary comorbidities	Pulmonary scarring	No scarring	Bronchial wall thickening	No bronchial wall thickening	Scoliosis	No scoliosis	Total
Largest curve measurement (°)	15 ± 14	60 ± 23	28 ± 34	39 ± 33	30 ± 31	56 ± 36	18 ± 18	33 ± 32	25 ± 32	44 ± 29	4 ± 4	31 ± 31
Forced expiratory volume (FEV1) (L)	2.7 ± 0.8	0.8 ± 0.4	2.4 ± 1.0	1.5 ± 1.2	2.2 ± 1.1	1.3 ± 1.0	2.5 ± 1.0	1.9 ± 1.1	2.6 ± 1.1	1.5 ± 0.9	3.3 ± 0.6	2.1 ± 1.1
Total lung capacity (L)	4.5 ± 1.3	1.7 ± 0.2	2.4 ± 1.0	2.9 ± 1.8	3.8 ± 1.6	2.5 ± 1.5	4.2 ± 1.5	3.6 ± 1.7	4.1 ± 1.8	2.9 ± 1.3	5.3 ± 1.2	3.7 ± 1.6
Forced vital capacity (FVC) (L)	3.2 ± 1.0	0.9 ± 0.4	2.9 ± 1.3	1.7 ± 1.4	2.7 ± 1.3	1.6 ± 1.2	2.9 ± 1.3	2.3 ± 1.4	3.0 ± 1.4	1.9 ± 1.1	3.8 ± 1.0	2.5 ±1.4
FEV1/FVC (%)	85 ± 7	85 ± 5	83 ± 5	84 ± 6	84 ± 6	85 ± 6	84 ± 5	84 ± 6	87 ± 4	84 ± 5	86 ± 6	84 ± 6
Diffusing capacity (mL/mmHg/min/L)	6.3 ± 3.2	6.5 ± 0.9	5.3 ± 0.7	6.3 ± 0.9	5.8 ± 2.2	6.1 ± 1.0	5.8 ± 2.5	6.1 ± 2.2	5.1 ± 0.6	5.7 ± 0.9	6.4 ± 3.4	5.9 ± 2.0
Percent reference diffusing capacity (%)	106 ± 20	144 ± 82	97 ± 34	153 ± 76	100 ± 27	125 ± 58	98 ± 29	108 ± 43	92 ± 12	107 ± 46	106 ± 25	107 ± 39

Table 3.

Between groups comparisons scoliosis and pulmonary function test measurements

Comparison	Type I vs III	Type I vs IV	Type III vs IV	Pulmonary Comorbidities vs no pulmonary comorbidities	Scarring vs no scarring (CT Scan)	Bronchial wall thickening vs no bronchial wall thickening	Scoliosis vs no scoliosis
	Mean difference (95% confidence interval); p value
Largest curve measurement (°)	-45 (-64 to -26); p = 0.002	-13 (-38 to 11); p = 0.26	32 (1 to 63); p = 0.03	9 (-20 to 39); p = 0.36	37 (16 to 58); p = 0.005	8 (-27 to 43); p = 0.41	41 (20 to 61); p < 0.001
Forced expiratory volume (FEV1) (L)	1.9 (1.3 to 2.5); p = 0.001	0.3 (-05 to 1.1); p = 0.39	-1.6 (-2.4 to -0.8); p = 0.001	-0.8 (-1.7 to 0.2); p = 0.09	-1.1 (-1.9 to -0.3); p = 0.007	-0.7 (-1.9 to 0.6); p = 0.23	-1.7 (-2.4 to -1.1); p < 0.001
Total lung capacity (L)	2.8 (1.8 to 3.9); p = 0.001	0.4 (-0.8 to 1.7); p = 0.36	-2.4 (-3.6 to -1.3); p = 0.001	-0.9 (-2.4 to 0.6); p = 0.22	-1.7 (-2.9 to -0.5); p = 0.01	-0.5 (-2.4 to 1.4); p = 0.51	-2.3 (-3.3 to -1.3); p = 0.001
Forced vital capacity (FVC) (L)	2.3 (1.5 to 3.1); p < 0.001	0.3 (-0.8 to 1.3); p = 0.53	-2.0 (-3.0 to -1.0); p = 0.001	-0.9 (-2.1 to 0.2); p = 0.11	-1.4 (-2.3 to -0.4); p = 0.009	-0.7 (-2.3 to 0.8); p = 0.23	-2.0 (-2.8 to -1.1); p < 0.001
FEV1/FVC (%)	1 (-6 to 7); p = 0.89	2 (-3 to 7); p = 0.67	1 (-3 to 6); p = 0.59	0 (-5 to 5); p = 0.98	1 (-4 to 5); p = 0.86	-3 (-8 to 3); p = 0.25	-2 (-7 to 2); p = 0.24
Diffusing capacity (mL/mmHg/min/L)	-0.2 (-3.5 to 3.1); p = 0.07	1.0 (-1.1 to 3.1); p = 0.79	1.2 (0.3 to 2.1); p = 0.03	0.5 (-1.6 to 2.7); p = 0.04	0.3 (-1.5 to 2.2); p = 0.08	0.9 (-1.4 to 3.3); p = 0.16	-0.8 (-2.6 to 1.1); p = 0.60
Percent reference diffusing capacity (%)	-38 (-99 to 22); p = 0.78	10 (-17 to 36); p = 0.57	48 (-17 to 112); p = 0.59	53 (8 to 99); p = 0.14	28 (-11 to 66); p = 0.46	16 (-30 to 63); p = 0.20	1 (-36 to 37); p = 0.85

Thirteen of 24 participants with an ECG had abnormal findings (tachycardia, abnormal ventricular relaxation, and T-wave abnormalities), while 10 of 23 with an echocardiogram had abnormal findings (aortic root dilation, atria dilation, moderate tricuspid insufficiency, and atrial septal aneurysm). Eighteen individuals had some level of mitral, tricuspid, and/or pulmonic regurgitation. There were no abnormalities found on abdominal and carotid ultrasound images. Of the individuals with a completed chest CT, 27 of 29 had abnormal findings in addition to skeletal abnormalities. Twenty-five of 29 individuals had bronchial wall thickening, 16 had ground glass opacities, 12 had atelectasis, 11 had scarring, and five had mosaic air trapping (Table 4). Bronchial wall thickening affected study participants of all ages ranging from 19 years (youngest person in the study) to 67 years (oldest person in the study). Of the 25 individuals with bronchial wall thickening, 19 had restrictive physiology. The four individuals who had a chest CT without bronchial wall thickening still had restrictive physiology. Bronchial wall thickening did not result in differences between FEV1/FVC or the scoliosis curve (Table 3). Participants with scarring also had lower forced vital capacity and total lung capacity (Table 3).

Table 4.

Chest CT abnormalities by osteogenesis imperfecta type

Osteogenesis imperfecta type	I (n = 9)	III (n = 8)	IV (n = 9)	All types (n = 29)
Bronchial wall thickening	7	7	11	25
Ground glass opacities	4	6	6	16
Atelectasis	4	4	4	12
Scarring	3	6	2	11
Mosaic air trapping	0	3	2	5
Bronchiectasis	1	3	0	4
Microfibrosis	1	1	0	2
Focal centrilobular clusters of nodules	0	1	0	1
Reticulation in right peripheral lobe	0	1	0	1
Paraseptal emphysema	0	1	0	1
Infection/aspiration	0	0	1	1
Small airways disease with centrilobular nodularity	1	0	0	1
Total	21	33	26	80

For this study, we collected the average scores from quality of life measures from various groups (Table 5). In addition, we obtained between-group comparisons (Table 6). The mental component summary (49 ± 9) and the physical component summary (45 ± 9) scores for the SF-36 were within 1 SD of those of the general population (Table 5). The average St. George’s Respiratory Questionnaire total scores were 16 ± 15, more than double that of the general population. Higher St. George’s Respiratory Questionnaire scores indicated worse quality of life. St. George’s Respiratory Questionnaire Impact subscores were higher for individuals with Type III OI than for those with Type I OI (Table 6), indicating that individuals with Type III OI are more negatively impacted by their respiratory health status than those with Type 1 OI. FEV1/FVC correlated with St. George’s Respiratory Questionnaire scores (R = 0.429; p = 0.02) but did not correlate with Functional Outcomes of Sleep Questionnaire (R = -0.26; p = 0.19) or SF-36 scores (physical component summary: R = -0.037, p = 0.85; mental component summary: R = -0.204, p = 0.29). Participants with the abnormal chest CT finding of scarring scored lower on the SF-36 physical component summary score than did those without this complication (Table 6). St. George’s Respiratory Questionnaire scores did not correlate with any of the abnormalities found on chest CT, suggesting that the abnormal findings of chest CT do not directly contribute to respiratory-related quality of life. People with pulmonary comorbidities scored higher on the St. George’s Respiratory Questionnaire symptom subscore and impact subscores and total St. George’s Respiratory Questionnaire score than did those without these complications, as expected. These individuals also had worse physical component summary and mental component summary scores, suggesting that pulmonary impairment negatively impacts overall physical and mental functioning in OI.

Table 5.

Quality of life measures (mean ± SD)

Measure	OI type	I	III	IV	Pulmonary comorbidities	No pulmonary comorbidities	Pulmonary scarring (CT scan)	No scarring (CT scan)	Bronchial wall thickening	No bronchial wall thickening	Scoliosis	No scoliosis	Total	General population [5, 11, 40, 41]
SF-36	Mental component summary score	48 ± 9	51 ± 7	48 ± 11	42 ± 7	51 ± 9	46 ± 6	50 ± 11	48 ± 9	49 ± 10	50 ± 9	45 ± 9	49 ± 9	50 ± 10
	Physical component summary score	47 ± 8	41 ± 10	45 ± 9	36 ± 6	48 ± 8	39 ± 8	47 ± 7	43 ± 9	49 ± 5	44 ± 9	46 ± 9	45 ± 9	50 ± 10
St. George’s Respiratory Questionnaire	Symptom	19 ± 18	27 ± 18	27 ± 16	35 ± 16	21 ± 16	29 ± 17	24 ± 16	24 ± 16	32 ± 15	21 ± 15	33 ± 20	25 ± 17	12 (9-15)
	Activity	13 ± 17	32 ± 26	22 ± 23	35 ± 26	18 ± 20	28 ± 26	20 ± 22	22 ± 23	27 ± 25	21 ± 20	25 ± 30	22 ± 23	9 (7-12)
	Impact	7 ± 13	18 ± 15	5 ± 6	23 ± 16	5 ± 7	16 ± 16	5 ± 7	10 ± 13	8 ± 11	9 ± 12	10 ± 14	9 ± 12	2 (1-3)
	Total	10 ± 15	24 ± 17	15 ± 11	29 ± 17	12 ± 11	23 ± 18	13 ± 12	16 ± 15	18 ± 15	15 ± 14	18 ± 18	16 ± 15	6 (5-7)
Sleep questionnaires	Functional outcomes of sleep questionnaire	18 ± 2	17 ± 2	18 ± 2	17 ± 2	18 ± 2	17 ± 2	18 ± 2	18 ± 2	18 ± 2	18 ± 2	18 ± 2	18 ± 2	≥ 17 = normal
	Pittsburgh Sleep Quality index	6 ± 3	5 ± 3	9 ± 4	9 ± 6	6 ± 3	7 ± 3	8 ± 5	7 ± 4	7 ± 1	7 ± 4	7 ± 2	7 ± 4	2.67 ± 1.70; ≤ 5 = normal

Table 6.

Between-group comparisons quality of life measures^a

Measure	Comparison	Type I vs III	Type I vs IV	Type III vs IV	Pulmonary comorbidities vs no pulmonary comorbidities	Scarring vs no scarring (CT Scan)	Bronchial wall thickening vs no bronchial wall thickening	Scoliosis vs no scoliosis
SF-36	Mental component summary score	-3 (-11 to 5); p = 0.63	0 (-9 to 9); p = 0.83	3 (-6 to 12); p = 0.59	-9 (-17 to -2); p = 0.02	-3 (-10 to 4); p = 0.23	0 (-11 to 10); p = 0.84	4 (-3 to 12); p = 0.26
	Physical component summary score	6 (-3 to 16); p = 0.18	2 (-6 to 10); p = 0.57	-4 (-13 to 5); p = 0.40	-12 (-18 to -5); p = 0.003	-8 (-14 to -2); p = 0.01	-6 (-15 to 4); p = 0.17	-2 (-10 to 6); p = 0.53
St. George’s Respiratory Questionnaire	Symptom	-7 (-26 to 11); p = 0.36	-8 (-23 to 7); p = 0.20	-1 (-16 to 15); p = 0.97	14 (0 to 28); p = 0.03	5 (-8 to 18); p = 0.48	-8 (-26 to 10); p = 0.33	-12 (-26 to 2); p = 0.12
	Activity	-20 (-42 to 3); p = 0.04	-9 (-29 to 11); p = 0.28	11 (-13 to 34); p = 0.36	17 (-2 to 17); p = 0.06	8 (-11 to 27); p = 0.39	-5 (-31 to 21); p = 0.68	-4 (-24 to 16); p = 0.90
	Impact	-12 (-26 to 3); p = 0.01	1 (-8 to 10); p = 0.22	13 (3 to 23); p = 0.03	18 (10 to 27); p = 0.003	11 (2 to 20); p = 0.08	2 (-12 to 16); p = 0.97	-1 (-12 to 10); p = 0.96
	Total	-14 (-30 to 3); p = 0.02	-4 (-16 to 8); p = 0.10	10 (-4 to 23); p = 0.16	17 (6 to 29); p = 0.009	10 (-2 to 21); p = 0.12	-2 (-18 to 15); p = 0.73	-3 (-16 to 9); p = 1.00
Sleep questionnaires	Functional outcomes of sleep questionnaire	1 (-1 to 3); p = 0.37	1 (-1 to 2); p = 0.43	0 (-2 to 2); p = 0.82	-1 (-3 to 1); p = 0.15	-1 (-2 to 0); p = 0.10	-1 (-3 to 1); p = 0.43	-1 (-2 to 1); p = 0.74
	Pittsburgh Sleep Quality Index	1 (-2 to 5); p = 0.36	-3 (-7 to 1); p = 0.19	-4 (-8 to -1); p = 0.03	3 (0 to 7); p = 0.12	2 (-4 to 2); p = 0.94	0 (-4 to 5); p = 0.81	0 (-4 to 4); p = 0.77

aThe values are given as mean difference (95% CI); p value.

Discussion

Although cardiopulmonary issues in patients with OI are the major cause of morbidity and mortality, they are less frequently recognized and studied than other complications [9, 12, 21]. This study was designed to further elucidate the relationship among scoliosis, cardiopulmonary health, and quality of life in individuals with OI. This study lends support to the view that OI collagenopathies have effects beyond the bone [10, 22]. The myocardium is approximately 75% Type I collagen. Type I collagen also plays an integral role in the structure of blood vessels. In the lungs, Type I collagen provides the structural framework for the alveoli and bronchial walls and plays an important role in lung elasticity [32]. Thus, inherent differences in the composition of lung and heart collagen may disturb the biomechanics of breathing and circulation [1]. Our results suggest that (1) cardiopulmonary impairment goes beyond abnormal chest wall architecture and scoliosis in OI, (2) there is a high prevalence of cardiopulmonary complications in all OI Types, and (3) pulmonary impairment negatively impacts quality of life in adults with OI.

Osteogenesis imperfecta is a rare genetic disorder, and the study was conducted at a single site. The study is limited by a small study population, which resulted in an underpowered analysis. Although there is a small sample, there is a fairly even distribution across OI types and age groups. Large cohorts of patients with rare diseases are often not available for study. This is especially true for adults, who may be lost to coordinated follow-up care once they age out of pediatric hospitals. Additionally, data collection performed at a single site offers the advantage of uniformity compared with a multicenter approach. One of the greatest limitations of this study is its cross-sectional design. Longitudinal data may be the best method to accurately assess how cardiopulmonary issues affect the health of individuals with OI. In conditions with clinical variability and where data normalization is limited, such as OI, changes over time can provide the most relevant data. The percent predicted values of pulmonary function test results (raw values normalized for comparison) must be approached with healthy skepticism in this population. Raw pulmonary function test values are normalized to controls that match in age, sex, race, and height [34]. Although it is appropriate to use percent predicted data in individuals with Type I OI, who are usually of average stature, there are no appropriate controls for individuals of short stature, whose percent predicted values may underestimate pulmonary dysfunction [34]. Even studies that suggest substituting arm span for height in shorter individuals fail to acknowledge that there is no standard method of measuring arm span in an individual with long-bone deformities [42]. In addition to arm span, trunk height was collected for a number of study participants. However, this measurement was not reliable due to variable anatomical deformities such as a pectus carinatum, which falsely increased the measurement. Trunk height measurements on radiograph were also unreliable due to variable distortion of the pelvis. This study addresses this limitation by using the FEV1/FVC ratio as the primary assessment of restrictive physiology, as it does not require normalization. Future studies to identify other, more consistent and reliable surrogate measurements that relate to diminished pulmonary function would be useful. Although the determination of bronchial wall thickening on chest CT scans is subjective, the definition of normal bronchial wall thickness to airway diameter on CT is used as a reference. Two independent reviewers (RAS, DD) with clinical expertise in related disciplines (pulmonology and thoracic radiology) reviewed the chest CT scans in this study. However, future studies may require review by a wider panel of experts to confirm CT findings. Despite its limitations, this is the first study we know of that prospectively evaluates cardiopulmonary status and its relationship with scoliosis in adults with OI.

FEV1/FVC was not correlated with the magnitude of scoliosis curves. This important result suggests that the severity of scoliosis may not solely determine the severity of pulmonary dysfunction in patients with OI. Restrictive lung function was common across the entire cohort (23 of 30 patients), 43% of whom had either mild (< 25°) curves (2 of 21) or no scoliosis at all (7 of 21). The minority (6 of 30) of individuals had curves > 60°. Although individuals with Type III OI had the largest curves, their pulmonary impairment was not worse than those with other Types (p = 0.85). In addition, no association was found between the presence of scoliosis and the presence of restrictive physiology. This is not to say that chest wall deformities play no role in respiratory impairment. Forced vital capacity and total lung capacity correlated with curve magnitude, which may indicate the impact of chest wall architecture on lung physiology. Hyperkyphosis (> 50°), larger waist circumference, age, and level of physical activity might also influence pulmonary function [6, 19, 28]. Still, our data show no relationship between pulmonary function and the extent of kyphosis, age, weight, and activity level. Therefore, we suggest that connective tissue pathology effects cardiopulmonary impairment in patients with OI.

Most participants (18 of 30) had some level of mild mitral, tricuspid, and/or pulmonic valve regurgitation, which is consistent with previous reports [22, 24, 25]. The aortic root diameter and ejection fraction were within normal limits for all evaluated participants. The modest severity of cardiovascular problems among participants suggests that pulmonary complications may play a larger role in morbidity. More than two thirds of study participants (23 of 30) across all OI types demonstrated restrictive physiology, which results in decreased pulmonary ventilation and reserve. This is consistent with predictions of increased lung stiffness in OI because of abnormal collagen production in the lungs [30, 33]. Most CT images showed evidence of atelectasis (41%) and bronchial wall thickening (86%). Atelectasis is often seen in individuals with altered chest architecture without OI, but could also be related to the underlying Type I collagen disorder in OI and contribute to a predisposition to pulmonary compromise in adults with OI. If severe enough, atelectasis may lead to arterial shunting of unoxygenated blood through the lung and pneumonia because of impaired clearance [9, 38, 44]. Bronchial wall thickening, which is a new finding, was seen in individuals with both normal and restrictive physiology. The significance of this finding should be further explored. We speculate that bronchial wall thickening is an inherent pulmonary phenotype in OI, affecting individuals of all types and possibly contributing to pulmonary dysfunction. Bronchial wall thickening, commonly found in individuals with airway inflammation, chronic bronchitis or chronic obstructive pulmonary disease, and asthma, leads to a more fixed and narrower airway that does not respond to changes in respiratory demand, increasing the load on the airway smooth muscles and causing poor airway clearance [2, 31]. Poor ventilation combined with poor airway clearance might lead to an increased risk of infection [30].

St. George’s Respiratory Questionnaire total scores correlated with FEV1/FVC and were worse than those of the general population. In contrast, the largest scoliosis curve did not correlate with St. George’s Respiratory Questionnaire scores, suggesting that quality of life was influenced by pulmonary impairment. St. George’s Respiratory Questionnaire activity (22 ± 23) and impact scores (9 ± 12), which measure the effects of pulmonary disease on daily activities and psycho-social function, respectively, were worse than those of the average population. FEV1/FVC values correlated with St. George’s Respiratory Questionnaire activity and impact scores. Study participants with pulmonary comorbidities not only scored worse on the St. George’s Respiratory Questionnaire but also had worse physical component summary and mental component summary scores, suggesting that their pulmonary impairment negatively impacts overall physical and mental functioning in addition to their respiratory-related quality of life.

Cardiopulmonary diseases are the leading cause of death in people with OI. However, most participants (21 of 30) did not indicate that they were concerned about their cardiac and pulmonary health. Although a recent publication provided evidence of a shift toward a systematic, multivariable, and longitudinal evaluation of the comorbidities associated with OI [34], there is no current consensus on how to approach monitoring the extraskeletal complications in adults. We suggest that adults with OI should be followed regularly because the OI diagnosis has implications beyond fracture. Primary care physicians can refer these patients for a baseline cardiopulmonary work-up. Orthopaedic surgeons should understand that all patients with OI are at risk for underlying pulmonary and possibly cardiac abnormalities, and should refer these patients for total care by a multidisciplinary group. Future studies must focus on elucidating the mechanisms by which extraosseous complications manifest in people with OI, with a focus on collagen abnormalities in heart and lung tissue on a histologic and molecular level. Studies such as this will serve as the basis for the development of appropriate, evidence-based clinical guidelines to improve overall health, quality of life, and survival.