Identification of an increased lifetime risk of major adverse cardiovascular events in UK Biobank participants with scoliosis

Abstract

Background: Structural changes caused by spinal curvature may impact the organs within the thoracic cage, including the heart. Cardiac abnormalities in idiopathic scoliosis patients are often studied post-corrective surgery or secondary to diseases. To investigate cardiac structure, function, and outcomes in participants with scoliosis, phenotype and imaging data of the UK Biobank (UKB) adult population cohort was analysed.
Methods: Hospital episode statistics of 502,324 adults were analysed to identify participants with scoliosis. Summary 2D cardiac phenotypes from 39,559 cardiac magnetic resonance imaging (CMR) scans were analysed alongside a 3D surface-to-surface (S2S) analysis.
Results: A total of 4,095 (0.8%, 1 in 120) UKB participants were identified to have all-cause scoliosis. These participants had increased lifetime risk of major adverse cardiac events (MACE) (HR=1.45, P<0.001), driven by heart failure (HR=1.58, P<0.001) and atrial fibrillation (HR=1.54, P<0.001). Increased radial and decreased longitudinal peak diastolic strain rates were identified in participants with scoliosis (+0.29, Padj<0.05; -0.25, Padj<0.05; respectively). Cardiac compression of the top and bottom of the heart and decompression of the sides was observed through S2S analysis. Additionally, associations between scoliosis and older age, female sex, heart failure, valve disease, hypercholesterolemia, hypertension, and decreased enrolment for CMR, were identified.
Conclusion: The spinal curvature observed in participants with scoliosis alters the movement of the heart. The association with increased MACE may have clinical implications for whether to undertake surgical correction. This work identifies, in an adult population, evidence for altered cardiac function and increased lifetime risk of MACE in participants with scoliosis. Future genetic analyses would aid to assess causality.

Full text

1
Identification of an increased lifetime risk of major adverse cardiovascular 1
events in UK Biobank participants with scoliosis 2
Valentina Q. Santofimio1, Adam Clement2, Declan P. O’Regan2, 3
James S. Ware1,2,3, Kathryn A. McGurk1,2,* 4
1National Heart and Lung Institute, Imperial College London, UK 5
2MRC London Institute of Medical Sciences, Imperial College London, Hammersmith 6
Hospital Campus, London, UK 7
3Royal Brompton & Harefield Hospitals, Guy’s and St. Thomas’ NHS Foundation 8
Trust, London, UK. 9
*Corresponding author: Dr. Kathryn A. McGurk; k.mcgurk@imperial.ac.uk 10
Word count: 2,669 11 12
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

2
Abstract 13
Background: Structural changes caused by spinal curvature may impact the organs 14
within the thoracic cage, including the heart. Cardiac abnormalities in idiopathic 15
scoliosis patients are often studied post-corrective surgery or secondary to diseases. 16
To investigate cardiac structure, function, and outcomes in participants with 17
scoliosis, phenotype and imaging data of the UK Biobank (UKB) adult population 18
cohort was analysed. 19
Methods: Hospital episode statistics of 502,324 adults were analysed to identify 20
participants with scoliosis. Summary 2D cardiac phenotypes from 39,559 cardiac 21
magnetic resonance imaging (CMR) scans were analysed alongside a 3D surface-22
to-surface (S2S) analysis. 23
Results: A total of 4,095 (0.8%, 1 in 120) UKB participants were identified to have all-24
cause scoliosis. These participants had increased lifetime risk of major adverse 25
cardiac events (MACE) (HR=1.45, P<0.001), driven by heart failure (HR=1.58, 26
P<0.001) and atrial fibrillation (HR=1.54, P<0.001). Increased radial and decreased 27
longitudinal peak diastolic strain rates were identified in participants with scoliosis 28
(+0.29, Padj<0.05; -0.25, Padj<0.05; respectively). Cardiac compression of the top and 29
bottom of the heart and decompression of the sides was observed through S2S 30
analysis. Additionally, associations between scoliosis and older age, female sex, 31
heart failure, valve disease, hypercholesterolemia, hypertension, and decreased 32
enrolment for CMR, were identified. 33
Conclusion: The spinal curvature observed in participants with scoliosis alters the 34
movement of the heart. The association with increased MACE may have clinical 35
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

3
implications for whether to undertake surgical correction. This work identifies, in an 36
adult population, evidence for altered cardiac function and increased lifetime risk of 37
MACE in participants with scoliosis. Future genetic analyses would aid to assess 38
causality. 39
Keywords: Scoliosis, heart, MACE, UK Biobank 40
41
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

4
Introduction 42
Scoliosis is the lateral curvature of the spine with a Cobb angle >10°, primarily 43
diagnosed in adolescents[1]. There are multiple aetiologies of scoliosis including 44
neuromuscular, congenital, syndromic, or secondary to other diseases such as 45
muscular dystrophy or Friedreich's ataxia[2]. However, the most common type is 46
idiopathic scoliosis, with a prevalence of 8% in adults aged over 40 years old[3–5]. In 47
later life, scoliosis can result from skeletal muscle diseases such as sarcopenia. 48
Sarcopenia is the loss of muscle mass associated with age that may cause 49
imbalance and alteration on the supportive muscles of the spine, contributing to the 50
progression of degenerative scoliosis in elderly patients[6,7]. Degenerative scoliosis 51
is observed in 68% of adults aged over 60 years old, as the joints and disks of the 52
spine begin to deteriorate[7]. Osteopenia, loss of bone density, is more frequently 53
observed in females and contributes to the severity of the curvature of the spine[8]. 54
Furthermore, the coexistence of congenital heart disease (CHD) and spinal 55
curvature is found in up to 12% of infant and juvenile scoliosis patients presenting 56
with CHD[9,10]. This is likely due to shared developmental aetiology, whereas 57
studies are lacking on the impact of age-related scoliosis on the functioning of a 58
developmentally normal heart. 59
60
In addition to impacting physical day-to-day activities, structural disruption of the 61
thoracic cage (pectus deformity) can influence the organs within, such as the 62
heart[11]. Pectus deformities lead to the displacement of the heart towards the left 63
side of the body which can result in right-sided spinal curvatures as the beating heart 64
pushes the thoracic vertebrae to the right[7,11]. Furthermore, the degree of 65
curvature of the spine can greatly increase the risk of restrictive lung disease, which 66
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

5
in conjunction with intrathoracic organ displacement, increases the risk of 67
comorbidities with high mortality rates, such as right heart failure[3,12]. 68
69
The impact of scoliosis on adult cardiac function has not been extensively studied 70
and the relationship between scoliosis and non-congenital cardiac manifestations is 71
not well characterised. In the UK Biobank adult population cohort, we explored 72
whether all-cause scoliosis has an impact on or relationship with cardiac phenotypes 73
of the matured adult heart. We identify altered radial and longitudinal peak diastolic 74
strain rates and an increased lifetime risk of major adverse cardiovascular effects 75
(MACE) in participants with scoliosis in the UK Biobank. In addition, we observe 76
associations between scoliosis and older age, female sex, heart failure, valve 77
disease, hypercholesterolemia, diagnosis of hypertension, and decreased enrolment 78
for cardiac magnetic resonance imaging. 79
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

6
Methods 80
The UK Biobank population cohort 81
The UK Biobank (UKB) recruited over 500,000 participants aged 40-69 years across 82
the UK between 2006-2010 (National Research Ethics Service, 11/NW/0382, 83
21/NW/0157)[13]. This project was conducted under the UK Biobank applications 84
47602 and 40616. All participants provided written informed consent[14]. 85
86
UK Biobank codes for identification of scoliosis 87
A diagnosis of scoliosis was identified for UKB participants through the first 88
occurrence of scoliosis trait (M41*), a composite trait of the first reported date 89
derived by the UKB that incorporates data from primary care, hospital inpatient 90
admissions, death records, and self-reported medical conditions (Table 2). 91
92
Cardiac magnetic resonance imaging data 93
Amongst all UKB participants, 39,559 participants had cardiac MRI (CMR) data 94
available[15]. Imaging was performed using a 1.5 Tesla machines (MAGNETOM 95
Aera, Siemens Healthcare)[15], and 2D summary CMR traits were analysed for an 96
association with scoliosis. This includes left ventricular ejection fraction (LVEF), left 97
ventricular end systolic volume (LVESV), left ventricular end diastolic volume 98
(LVEDV), and measures of cardiac strain: Eulerian radial strain (Err), Eulerian 99
longitudinal strain (EII), Eulerian circumferential strain (Ecc), radial peak diastolic 100
strain rate (PDSRrr), and longitudinal peak diastolic strain rate (PDRSll). All CMR 101
traits were adjusted for age at the time of imaging, sex, White British ancestry, 102
systolic blood pressure (SBP), and body surface area (BSA). 103
104
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

7
Surface to surface analysis 105
A mass univariate regression was utilised to explore associations between the three-106
dimensional (3D) mesh-derived phenotype and scoliosis[16–19]. The underlying 107
principle of this approach is the implementation of a linear regression at each vertex 108
of the 3D atlas to derive a regression coefficient associated with the variable of 109
interest, which results in a map of beta-coefficients showing the strength and 110
direction of these associations. The analysis was adjusted for age at the time of 111
imaging, sex, White British ancestry, BSA, diastolic blood pressure (DBP), and SBP. 112
113
Lifetime risk and survival analyses 114
Lifetime risk of disease, from birth month to January 2021, was assessed using the 115
first occurrence of scoliosis trait (M41*). The first occurrence of health outcomes 116
summary data in the UKB is reported using ICD10 codes[13]. The fields analysed for 117
the major adverse cardiovascular events (MACE) composite trait were as follows: 118
cardiac arrest, I46*; atrial fibrillation and flutter/arrhythmia: I48* and I49*; heart 119
failure, I50*; and stroke, I64*, as previously published[20]. The survival analysis was 120
conducted using MACE and death as the primary outcome. Survival and survminer 121
R packages were used to estimate hazard ratios (HR). 122
123
Statistical analysis 124
R programming language (version 3.6.0) and RStudio software (version 1.3.1073) 125
were used for analyses. Categorical variables were assessed using Chi-Squared 126
Test or Fisher's Exact Test and expressed as percentages. Continuous variables 127
were assessed using Student’s t-test and expressed as mean ± standard deviation 128
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

8
(SD). All P-values were adjusted using Bonferroni correction for multiple 129
comparisons where Padj<0.05 was deemed significant. 130
In the association of disease analysis, all variables were adjusted for age at 131
recruitment (UKB ID: 21022-0.0) and genetic data-derived sex (UKB ID:22001-0.0) 132
using a multiple linear regression model. A separate multiple linear regression model 133
for CMR traits was used to adjust for covariates at the time of imaging: age, sex, 134
White British ancestry, SBP, and BSA. 135
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

9
Results 136
Prevalence of scoliosis in the UK Biobank 137
The prevalence of all-cause scoliosis in 502,324 participants of the UKB was 1 in 138
120 (n=4,095; 0.8%). Of the 4,095 participants with scoliosis, 1,489 (37%) were 139
diagnosed with scoliosis prior recruitment. From the total participants with scoliosis, 140
10 (0.2%) reported congenital scoliosis, 23 (0.6%) reported childhood scoliosis 141
(infantile and juvenile) and the rest reported scoliosis due to other causes later in life. 142
The most common scoliosis subtype was unspecified scoliosis reported by 2,685 143
(65.6%) participants (Table 2). 144
145
Participants with scoliosis were significantly older than the rest of the population 146
(59.4 years old ± 7.58; mean age 56.5 years old ± 8.10; Padj<0.01) and were more 147
female (69% female in scoliosis cohort; 55% female in the rest of the UKB 148
population; Padj<0.01). Lifetime risk analysis showed that participants with scoliosis 149
had significantly longer lives when compared to the rest of the population, regardless 150
of sex (Figure 1). 151
152
We observed a significantly increased burden of heart failure (+4%, Padj<0.001), 153
valve disease (+1%, Padj<0.001), hypercholesterolemia (+7%, Padj<0.001) and 154
diagnosis of hypertension (+13%, Padj<0.001) in scoliosis participants, adjusted for 155
age and sex (Table 2). Although diagnosis of hypertension was significantly 156
increased, no significant differences were found with SBP or DBP, which may be due 157
to correction of diagnosed hypertension through medication. Additionally, the 158
proportion of scoliosis participants with CMR available was significantly decreased (-159
3%, Padj<0.001) compared to the rest of the population. 160
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

10
161
CMR analysis of participants with scoliosis identifies altered cardiac diastolic strain 162
2D summary CMR traits were available on 39,559 participants[15]. 224 participants 163
with scoliosis had CMR available and had significantly increased radial PDSR 164
(Padj<0.05) and decreased longitudinal PDSR (Padj<0.05) compared to participants 165
without a diagnosis of scoliosis (Table 3). Adequate diastolic function is essential 166
during ventricular filling and maintenance of optimum stroke volume. PDSR is a 167
diastolic function trait, that has been previously associated with major adverse 168
cardiovascular events, increased mortality, increased blood pressure and altered left 169
atrial function [21]. For example, decreased PDRSrr corresponds to stiffer ventricle, 170
impairing its relaxation and increasing the risk of heart failure[21]. No additional 171
significant associations were found with the other CMR measures available for 172
analysis and no association was observed between scoliosis and diabetes (Table 1). 173
174
Surface to surface analysis of participants with scoliosis shows increased cardiac 175
compression 176
A 3D surface-to-surface analysis was performed on 21,088 participants of the cohort. 177
The three-dimensional cardiac modelling of scoliosis patients showed increased 178
strain at the top and bottom of the heart (Figure 2). Radial cardiac decompression 179
(sides of the heart) was also observed in participants with scoliosis. However, these 180
were not significant when adjusting for the number of comparisons and covariates. 181
182
Lifetime risk of MACE is increased in participants with scoliosis 183
A significantly increased lifetime risk of MACE was observed for UKB participants 184
with scoliosis (Figure 3; HR=1.45, P<0.001), mainly driven by heart failure 185
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

11
(HR=1.58, P<0.001) and atrial fibrillation (HR=1.54, P<0.001). The probability of 186
MACE doubled in males into older age (from 60 years of age). This may be caused 187
through the altered cardiac diastolic strain rates observed in participants with 188
scoliosis. However, we emphasise caution regarding the causality of this association 189
of scoliosis and MACE at this stage. 190
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

12
Discussion 191
Participants with scoliosis have increased lifetime risk of MACE 192
To the best of our knowledge, this is the first study to identify an increased lifetime 193
risk of MACE, through atrial fibrillation and heart failure, in participants with scoliosis. 194
This may be through the identified, significantly increased PDSRrr and decreased 195
PDSRll. The abnormal curvature of the spine can increase mechanical constraint on 196
the heart which may result in diastolic dysfunction[22] and the severity of the spinal 197
deformity has been shown to aggravate ventricular and right atrial pressure[3].
198
199
It is unlikely that the observed heart failure is due to blood pressure abnormalities as 200
no significant differences were found in measured SBP and DBP between 201
participants with scoliosis and the rest of the population. However, it is possible that 202
pulmonary dysfunction may be eliciting cardiac dysfunction alongside atrial 203
fibrillation, in patients with advanced scoliosis[3,11,23]. These findings suggest that 204
early medical intervention in patients with scoliosis through surgery, may decrease 205
the risk of a future MACE, however future validation of this finding would need to be 206
confirmed in a scoliosis case cohort. It may be possible that scoliosis develops 207
secondary to other diseases that could also increase the risk of MACE. Future 208
genetic analyses would be beneficial to assess causality through Mendelian 209
randomisation techniques. Likewise, studies are required to determine whether the 210
cardiovascular changes observed are reversible with scoliosis treatment surgery. 211
212
A previous study assessed 201 scoliosis patients for cardiopulmonary changes 213
following corrective scoliosis surgeries and suggested that untreated scoliosis can 214
result in pulmonary dysfunction and subsequently lead to right heart failure, 215
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

13
increasing mortality[3]. Postoperative normalisation of cardiac measures was 216
observed in this study, highlighting the potential benefits of surgical correction of 217
scoliosis on cardiac function[3]. Additionally, a long-term follow-up study showed 218
increased risk of pulmonary limitations such as shortness of breath in patients with 219
untreated scoliosis, aggravated by spinal curvature[24]. 220
221
It is challenging to accurately assess the cardiac function of scoliosis patients, due to 222
the heart being displaced[22] and participants with scoliosis in the UKB had 223
significantly less CMR scans available, possibly due to lack of comfort for prolonged 224
periods of time in MRI machines. Adaptations may be required to allow patients with 225
scoliosis to undergo MRI scans more comfortably, to further assess any alteration in 226
cardiac function in scoliosis patients. 227
228
Participants with scoliosis have altered cardiac diastolic strain rates 229
In the UKB, participants with scoliosis showed increased pressure at the top and 230
bottom of the heart, as well as elongation of the sides of the heart. These results 231
concur with the 2D-derived CMR findings of significantly increased PDSRrr and 232
decreased PDSRll. A reduced PDSRll has been previously associated with reduced 233
left atrial function[21]. No significant associations were found with the other studied 234
CMR traits, suggesting that this compression does not alter blood flow through the 235
heart. This altered cardiac strain may be explained by the deformity of the thoracic 236
cage in scoliosis limiting cardiac diastolic movement. The mechanical abnormalities 237
of the thoracic spine as well as the impact on the pulmonary system, may be the 238
primary cause of the heart involvement[25,26]. Secondary involvement via altered 239
pulmonary haemodynamics may be possible, where spinal curvature impacts 240
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

14
pulmonary pressures leading to pulmonary hypertension. Likewise, direct 241
compression of the myocardium could occur in conjunction to pulmonary 242
involvement. These events can contribute to the development of cardiac 243
consequences in scoliosis patients[27]. It would be beneficial to include cardiac 244
follow-ups in patients with scoliosis to observe any cardiac alterations as scoliosis 245
progresses, ensuring early intervention. 246
247
Prevalence of scoliosis in the UK Biobank 248
We report a prevalence of scoliosis of 0.8% in the UKB population cohort. Previous 249
literature has reported a scoliosis prevalence of 8% in adult volunteers aged over 40 250
years old[5]. Patients included in the previous clinical study were recruited on 251
evaluation of bone mineral density and thus were likely at increased risk of scoliosis. 252
Additionally, this discrepancy may be due to limitations of the UKB cohort (see 253
Limitations). 254
255
Participants with scoliosis in the UKB are more elderly compared to the rest of the 256
population, regardless of sex. Scoliosis occurs alongside other diseases such as 257
osteoporosis or degenerative spine disorders with increasing age[5]. 258
259
In the UKB cohort, scoliosis is more commonly found in females than males (2.5:1), 260
which agrees with a previous scoliosis case cohort study of adolescents and adults 261
that reported a ratio of scoliosis in females to males as 2:1[28]. Although not fully 262
understood, there are different theories linking the higher prevalence of scoliosis in 263
females, including implication of the autonomic nervous system (ANS) in skeletal 264
growth and/or puberty through leptin hormone, which has been found decreased in 265
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

15
female patients with idiopathic scoliosis[1,29,30]. The role of leptin in scoliosis is not 266
fully understood, and analyses to determine causality would aid our current 267
understanding of the mechanisms behind scoliosis. Furthermore, it has been 268
reported that approximately 30% of females (n=324) suffer from systemic osteopenia 269
which strongly contributes to the progression of spinal curvature in adolescent 270
females following skeletal maturity[8]. 271
272
Limitations 273
The participants in the UKB cohort were recruited at 40-69 years of age and most 274
participants are of European ancestry. Selection bias for inclusion in CMR sub cohort 275
may have excluded more severe scoliosis participants from study. In addition, 276
discrimination of scoliosis aetiology was mostly unspecified. 277
278
Conclusions 279
This work describes for the first time in an adult population, evidence for altered 280
cardiac function in adult participants with scoliosis. We identified altered diastolic 281
strain, increased lifetime risk of MACE driven by HF and atrial fibrillation and 282
observed cardiac compression in the UK Biobank participants with scoliosis. Further 283
research is required to follow up the role of scoliosis in cardiac manifestations in a 284
clinical setting. 285
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

16
Data and code availability 286
UK Biobank (https://www.ukbiobank.ac.uk/) population reference datasets are 287
publicly available. All analysis code will be available on GitHub 288
(https://github.com/ImperialCardioGenetics/Scoliosis) 289
Competing interests 290
J.S.W. has consulted for MyoKardia, Inc., Foresite Labs, and Pfizer. D.P.O. has 291
consulted for Bayer. 292
Acknowledgements and funding sources 293
This work was supported by the Wellcome Trust [107469/Z/15/Z; 200990/A/16/Z], 294
Medical Research Council (UK), British Heart Foundation [RG/19/6/34387, 295
RE/18/4/34215], and the NIHR Imperial College Biomedical Research Centre. The 296
views expressed in this work are those of the authors and not necessarily those of 297
the funders. For open access, the authors have applied a CC BY public copyright 298
license to any Author Accepted Manuscript version arising from this submission. 299
CRediT statement 300
Conceptualization: K.A.M.; Methodology: K.A.M, V.Q.S.; Formal Analysis: V.Q.S., 301
A.C.; Resources: J.S.W., D.O’R.; Data curation: V.Q.S., K.A.M.; Writing – original 302
draft: V.Q.S.; Writing – review & editing: A.C., D.O’R., J.S.W., K.A.M.; Visualization: 303
V.Q.S.; Supervision: K.A.M., J.S.W. 304
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

17
References 305
1 Cheng JC, Castelein RM, Chu Winnie, et al. Adolescent idiopathic scoliosis. 306 Nat Rev Dis Primers 2015;1:15030. doi:10.1038/nrdp.2015.30 307 2 Rummey C, Flynn JM, Corben LA, et al. Scoliosis in Friedreich’s ataxia: 308 longitudinal characterization in a large heterogeneous cohort. Ann Clin Transl 309 Neurol 2021;8:1239–50. doi:10.1002/acn3.51352 310 3 Sarwahi V, Galina J, Atlas A, et al. Scoliosis Surgery Normalizes Cardiac 311 Function in Adolescent Idiopathic Scoliosis Patients. Spine (Phila Pa 1976) 312 2021;46:E1161–7. doi:10.1097/BRS.0000000000004060 313 4 Trobisch P, Suess O, Schwab F. Idiopathic Scoliosis. Dtsch Arztebl 314 2010;107:875–84. doi:10.3238/arztebl.2010.0875 315 5 Kebaish KM, Neubauer PR, Voros GD, et al. Scoliosis in adults aged forty 316 years and older: Prevalence and relationship to age, race, and gender. Spine 317 (Phila Pa 1976) 2011;36:731–6. doi:10.1097/BRS.0b013e3181e9f120 318 6 Aftzoglou P. Sarcopenia and falls in patients with adult scoliosis. J Frailty 319 Sarcopenia Falls 2017;02:83–7. doi:10.22540/jfsf-02-083 320 7 Konieczny MR, Senyurt H, Krauspe R. Epidemiology of adolescent idiopathic 321 scoliosis. J Child Orthop 2013;7:3–9. doi:10.1007/s11832-012-0457-4 322 8 Hung VWY, Qin L, Cheung CSK, et al. Osteopenia: A new prognostic factor of 323 curve progression in adolescent idiopathic scoliosis. Journal of Bone and Joint 324 Surgery 2005;87:2709–16. doi:10.2106/JBJS.D.02782 325 9 Bülbül S, Benl
İ
İ
T, Güler M, et al. Scoliosis prevalence in congenital heart 326 disease patients. Turkish Spine 2013;24:3–12. 327 10 Kaito T, Shimada M, Ichikawa H, et al. Prevalence of and Predictive Factors 328 for Scoliosis After Surgery for Congenital Heart Disease in the First Year of 329 Life. Journal of Bone and Joint Surgery 2018;3:e0045. 330 doi:10.2106/JBJS.OA.17.00045 331 11 Wang Y, Chen G, Xie L, et al. Mechanical factors play an important role in 332 pectus excavatum with thoracic scoliosis. J Cardiothorac Surg 2012;7:2–4. 333 doi:10.1186/1749-8090-7-118 334 12 Swank SM, Winter RB, Moe JH. Scoliosis and Cor Pulmonale. Spine (Phila Pa 335 1976) 1982;7:343–54. doi:0.1097/00007632-198207000-00004 336 13 Sudlow C, Gallacher J, Allen N, et al. UK Biobank: An Open Access Resource 337 for Identifying the Causes of a Wide Range of Complex Diseases of Middle 338 and Old Age. PLoS Med 2015;12:1–10. doi:10.1371/journal.pmed.1001779 339 14 Szustakowski JD, Balasubramanian S, Sasson A, et al. Advancing Human 340 Genetics Research and Drug Discovery through Exome Sequencing of the UK 341 Biobank. medRxiv Published Online First: 2020. 342 doi:10.1101/2020.11.02.20222232 343 15 Bai W, Suzuki H, Huang J, et al. A population-based phenome-wide 344 association study of cardiac and aortic structure and function. Nat Med 345 2020;26:1654–62. doi:10.1038/s41591-020-1009-y 346
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

18
16 Schafer S, de Marvao A, Adami E, et al. Titin-truncating variants affect heart 347 function in disease cohorts and the general population. Nat Genet 2017;49:46–348 53. doi:10.1038/ng.3719 349 17 Biffi C, de Marvao A, Attard MI, et al. Three-dimensional cardiovascular 350 imaging-genetics: A mass univariate framework. Bioinformatics 2018;34:97–351 103. doi:10.1093/bioinformatics/btx552 352 18 Duan J, Ghalib B, Schlemper J, et al. Europe PMC Funders Group Automatic 353 3D bi-ventricular segmentation of cardiac images by a shape-refined multi-task 354 deep learning approach. 2019;38:2151–64. 355 doi:10.1109/TMI.2019.2894322.Automatic 356 19 Meyer H v, Dawes TJW, Serrani M, et al. Europe PMC Funders Group Genetic 357 and functional insights into the fractal structure of the heart. Nature 358 2021;584:589–94. doi:10.1038/s41586-020-2635-8.Genetic 359 20 McGurk KA, Zheng SL, Henry A, et al. Correspondence on “ACMG SF v3.0 list 360 for reporting of secondary findings in clinical exome and genome sequencing: 361 a policy statement of the American College of Medical Genetics and Genomics 362 (ACMG)” by Miller et al. Genetics in Medicine 2022;24:744–6. 363 doi:10.1016/j.gim.2021.10.020 364 21 Thanaj M, Mielke J, McGurk KA, et al. Genetic and environmental 365 determinants of diastolic heart function. Nature Cardiovascular Research 366 2022;1:361–71. doi:10.1038/s44161-022-00048-2 367 22 Li S, Yang J, Zhu L, et al. Ventricular and atrial mechanics and their interaction 368 in patients with congenital scoliosis without clinical heart failure. Cardiol Young 369 2014;760:976–83. doi:10.1017/S1047951114001504 370 23 Mauro A lo, Aliverti A. Physiology of respiratory disturbances in muscular 371 dystrophies. Breathe 2016;12:318–27. doi:10.1183/20734735.012716 372 24 Hawes MC, Weinstein SL. Health and Function of Patients with Untreated 373 Idiopathic Scoliosis. J Am Med Assoc 2003;289:2644–5. 374 doi:10.1001/jama.289.20.2644-a 375 25 Janicki JA, Frcsc BA. Scoliosis: Review of diagnosis and treatment. Paedriatric 376 Child Health 2007;12:771–6. 377 26 Prof A. Pulmonary Functions in Patients With Idopathic Scoliosis. 378 2017;90:123–8. 379 27 Li Q, Zeng F, Chen T, et al. Management of Severe Scoliosis with Pulmonary 380 Arterial Hypertension: A Single-Center Retrospective Case Series Study. 381 Geriatr Orthop Surg Rehabil 2022;13:1–11. doi:10.1177/21514593221080279 382 28 Lang C, Wang R, Chen Z, et al. Incidence and Risk Factors of Cardiac 383 Abnormalities in Patients with Idiopathic Scoliosis. World Neurosurg 384 2019;125:e824–8. doi:10.1016/j.wneu.2019.01.177 385 29 Man GCW, Tam EMS, Wong YS, et al. Abnormal Osteoblastic Response to 386 Leptin in Patients with Adolescent Idiopathic Scoliosis. Sci Rep 2019;9:1–7. 387 doi:10.1038/s41598-019-53757-3 388
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

19
30 Wang Q, Wang C, Hu W, et al. Disordered leptin and ghrelin bioactivity in 389 adolescent idiopathic scoliosis (AIS): a systematic review and meta-analysis. J 390 Orthop Surg Res 2020;15:1–9. doi:10.1186/s13018-020-01988-w 391 392
393
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

20
Tables 394
Table 1 Summary statistics of participants with scoliosis compared to the rest 395 of the UK Biobank population cohort. 396
Data is presented as mean ± SD or count (%); all the variables are adjusted for age 397 and sex; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood 398 pressure; CMR, cardiac magnetic resonance imaging; CM, cardiomyopathy. White 399 British ancestry was genetically inferred. Padj, P-value adjusted for multiple 400 comparisons via Bonferroni correction; Significant, Padj<0.05; NS, not significant, 401 Padj>0.05. 402
403
Characteristic General
population
(n=502,324)
Scoliosis
carriers
(n=4,095)
Padj
Age at recruitment 56.50 ± 8.10 59.40 ± 7.58 <0.001
White British 406,131 (81%) 3344 (82%) NS
Sex, female 261,862 (55%) 2,845 (69%) <0.001
BMI 27.40 ± 4.80 27.00 ± 5.16 NS
Auto SBP 140.00 ± 19.60 141.00 ± 20.00 NS
Auto DBP 82.20 ± 10.70 81.50 ± 10.80 NS
Diabetes 27,771 (5%) 212 (5%) NS
Hypercholesterolaemia 92,824 (18%) 1,022 (25%) <0.001
Hypertension 167,817 (33%) 1,902 (46%) <0.001
Included in UKBB CMR
cohort 39,319 (8%) 224 (5%) <0.001
Heart failure 10,504 (2%) 237 (6%) <0.001
Any CM 3,602 (0.7%) 49 (2%) NS
Valve disease 5,145 (1%) 96 (2%) <0.001
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

21
Table 2: Codes used for identification of all-cause scoliosis. 404 The HES data is coded using the International statistical Classification of Disease 405 (ICD) codes, versions 9 and 10. The self-reported data (SR) has a UKB-specific 406 coding system. SR, non-cancer illness code, self-reported data; ICD9, summary 407 diagnoses (main and secondary); ICD10, underlying primary cause of death, 408 contributory cause of death, and external cause of death, summary diagnoses (main 409 and secondary); GP, general practitioner. 410
411
Disease/phenotype Participants
(n=4,095) Coding system
Self-reported scoliosis 468 (11.4%) SR: 1535
Kyphoscoliosis and scoliosis
Congenital postural scoliosis 5 (0.1%)
0 (%) ICD9: 7373, 75420
Congenital scoliosis
Post-radiation scoliosis
Childhood scoliosis
(infantile and juvenile)
Other idiopathic scoliosis
Thoracogenic scoliosis
Neuromuscular scoliosis
Secondary scoliosis
Other forms of scoliosis
Unspecified scoliosis
10 (0.2%)
10 (0.2%)
23 (0.6%)
56 (1.4%)
24 (0.6%)
22 (0.6%)
79 (2%)
226 (5.5%)
2685 (65.5%)
ICD10: Q763, M965, M41,
M410, M4100-109, M411,
M4110-119, M412, M4120-
129, M413, M410-139, M414,
M1410-149, M415, M1450-
159, M418, M4180-189,
M419, M4190-199
Unspecified scoliosis 487 (11.9%) GP-derived
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

22
Table 3: Summary of the altered cardiac PDSR in participants of the UKB with 412 scoliosis. 413 Mean and standard deviation for adjusted PDSRs are described for the general UKB 414 population and participants with all-cause scoliosis. The p-value was adjusted for 415 multiple comparisons of CMR traits. WT, wall thickness; LVEDV, left ventricular end-416 diastolic volume; LVEF, left ventricular ejection fraction; LVM, left ventricular mass; 417 Ecc, circumferential; Err, radial; Ell, longitudinal; PDSRll, longitudinal peak diastolic 418 strain rate; PDSRrr, radial peak diastolic strain rate. 419 420
CMR trait General population
(n=38,319) Scoliosis
(n=224) Padj
Maximum WT
9.40 ± 1.62 9.36 ± 1.70 NS
LVEDV 148.00 ± 33.90 136.00 ± 32.80 NS
LVEF 59.50 ± 6.16 59.80 ± 6.03 NS
LVM 86.0 ± 22.30 78.00 ± 20.70 NS
Ecc Global (-22.33) ± 3.35 (-22.10) ± 3.77 NS
Err Global 45.00 ± 8.31 45.20 ± 8.73 NS
Ell Global (-18.50) ± 2.76 (-17.90 ) ± 2.44 NS
PDSRll 1.66 ± 0.61 1.45 ± 0.58 <0.05
PDSRrr -5.64 ± 2.09 -5.08 ± 2.13 <0.05
421
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

23
Figure legends 422
Figure 1: Scoliosis is associated with longer life compared to the rest of the 423 UKB cohort, adjusted for sex. 424
Years of life is shown on the x-axis from date of birth. Counts of UKB participants are 425 shown in the box depicting number at risk. The median survival time (a survival 426 probability of 0.5) was significantly increased in participants with scoliosis, regardless 427 of sex. Median survival time for females with scoliosis vs females without scoliosis (2 428 years difference); median survival time for males with scoliosis vs males without 429 scoliosis (1.5 years difference). 430
Figure 2 Surface to surface analysis suggests a compressed heart in 431 participants with scoliosis. 432
3D models of left ventricle geometry with standardized beta-coefficients which show 433 the association between scoliosis and regional surface-to-surface distance. Blue, 434 increased inward pressure; red, increased outwards pressure. 435 436 Figure 3: Increased lifetime risk of MACE in UKB participants with scoliosis. 437
A, cumulative incidence curve depicts an increase in lifetime risk of MACE in 438 participants with scoliosis over time (HR=1.45, P<0.001). B, incidence curve 439 stratified by sex (HR=1.63, P<0.001). 440
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

p < 0.0001
0.00
0.25
0.50
0.75
1.00
40 60 80
Time (Years)
Survival probability
Status Females without scoliosis Males without scoliosis Females with scoliosis Males with scoliosis
12723 12502 11116 6827 446
18978 18720 16925 10540 686
276 274 255 174 22
160 159 152 105 10
Males with scoliosis
Females with scoliosis
Males without scoliosis
Females without scoliosis
40 50 60 70 80
Time (Years)
Status
Number at risk
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

A B
C D
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

+++++++++++++++++++++++++++++++++++ + +
+++++++++++++++++++++++++++++
+++ + +
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
40 50 60 70 80
Years
Probability freedom from events
498224 498143 498006 497635 496408 491495 391400 223111 24212
4094 4090 4088 4077 4060 4010 3465 2317 333
Participants with scoliosis
Participants without scoliosis
40 50 60 70 80
Years
Status
Number at risk
Status ++
Participants without scoliosis Participants with scoliosis
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint

++++++++++++++++++++++++++++++++++ +
+++++++++++++++++++++++++++++++++++ + +
+++++++++++++++++++++++++++++
+++ +
+++++++++++++++++++++++
++++
++
+
+
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
40 50 60 70 80
Years
Probability freedom from events
261858 261820 261746 261545 261001 259023 208016 119022 12880
222265 222225 222168 222004 221368 218578 172305 97942 10537
2845 2842 2840 2834 2823 2797 2422 1644 253
1113 1112 1112 1107 1102 1080 933 603 66
Males with scoliosis
Females with scoliosis
Males without scoliosis
Females without scoliosis
40 50 60 70 80
Years
Status
Number at risk
Status ++++
Females without scoliosis Males without scoliosis Females with scoliosis Males with scoliosis
. CC-BY 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted November 22, 2022. ; https://doi.org/10.1101/2022.11.21.22282578doi: medRxiv preprint