Eur Spine J
DOI 10.1007/s00586-014-3333-8
ORIGINAL ARTICLE
Vertebral endplate change as a feature of intervertebral disc
degeneration: a heritability study
Juhani H. Määttä • Minna Kraatari • Lisa Wolber •
Jaakko Niinimäki • Sam Wadge • Jaro Karppinen •
Frances M. K. Williams
Received: 25 October 2013 / Revised: 23 April 2014 / Accepted: 24 April 2014
Ó Springer-Verlag Berlin Heidelberg 2014
Abstract
Purpose To study the prevalence of vertebral endplate or
Modic change (MC), the progression of MC over a 10-year
follow-up and the heritability of MC prevalence in a
classical twin study.
Methods The study population was recruited from
TwinsUK register between 1996 and 2000. MC was eval-
uated from T2-weighted lumbar magnetic resonance
imaging (MRI) at baseline and follow-up. Heritability was
estimated using variance components analysis. Baseline
MRI with appropriate data was available for 831 twins and
J. Määttä, M. Kraatari contributed equally to this work.
J. H. Määttä (&)  J. Karppinen
Department of Physical and Rehabilitation Medicine, Institute of
Clinical Medicine, University of Oulu, and Medical Research
Center Oulu, PL 5000, 90114 Oulu, Finland
e-mail: juhani.maatta@fimnet.fi
M. Kraatari
Faculty of Medicine, Institute of Health Sciences,
University of Oulu, Oulu, Finland
M. Kraatari
Biocenter and Faculty of Biochemistry and Molecular Medicine,
Oulu Center for Cell–Matrix Research, University of Oulu,
Oulu, Finland
L. Wolber  S. Wadge  F. M. K. Williams
Department of Twin Research and Genetic Epidemiology,
King’s College London, London, UK
J. Niinimäki
Department of Diagnostic Radiology, Institute of Diagnostics,
Oulu University Hospital, Oulu, Finland
J. Karppinen
Health and Work Ability, and Disability Prevention Center,
Finnish Institute of Occupational Health, Oulu, Finland
follow-up for 436 twins. In total, both baseline and follow-
up imaging were available for 347 twins.
Results Mean age of the study population was 54.1 years
(range 45.7–62.5) and females comprised 96 %. The
prevalence of MC at baseline was 32.1 % and at follow-up
48.4 %. The incidence of MC during the 10-year follow-up
was 21.6 % and was highest at L4–5 and L5–S1. MC
regressed totally in 3.5 % of twins. Twins with prevalent
MC at baseline demonstrated a higher incidence of MC at
upper lumbar levels during follow-up compared to twins
without baseline MC (p = 0.009). Probandwise concor-
dance rates were higher in monozygotic (0.56) than dizy-
gotic twin pairs (0.39) suggestive of familial influence.
Heritability of MC prevalence was estimated at 30
(16–43) %.
Conclusions The results suggest that MC is generally
progressive in middle age and furthermore is heritable.
Since MC is associated with disc degeneration, which is
also heritable, further work on potential shared mecha-
nisms is needed.
Keywords Disc degeneration  Heritability  Magnetic
resonance imaging  Modic change  Vertebral endplate
Introduction
Low back pain (LBP) is a common disorder in the western
world and accounts for disability and considerable work
absenteeism [1]. Lumbar disc degeneration (DD) is mod-
estly associated with LBP although greater in advanced DD
with disc height narrowing [2, 3]. It has been suggested that
different DD phenotypes should be ‘‘split’’ rather than
‘‘lumped’’ into a single phenotype (DD) so as to allow a
more meaningful interrogation of the pathogenic processes
123

underlying the constituent traits [4]. One trait of special
interest is vertebral endplate or Modic change (MC), which
comprises vertebral endplate and bone marrow lesions seen
on magnetic resonance imaging (MRI) but not evident on
plain radiographs [5, 6]. This trait has also been docu-
mented to be associated with LBP in some studies [7, 8],
but not all [9]. MC is classified as type I lesions (low T1-
and high T2-signals in MRI), which are associated with
vascular granulation tissue within the bone marrow indi-
cating an ongoing degeneration; type II lesions (high T1-
and T2-signals) representing fatty replacement of the bone
marrow; and type III lesions (low T1- and T2-signals) with
endplate sclerosis [5]. Interestingly, it has been proposed
that MC may represent a specific degenerative phenotype
of LBP [7].
Heritability is defined as the proportion of phenotypic
variation in a population due to underlying genetic fac-
tors. Demonstrating significant heritability is an essential
first step before attempting to identify the genetic vari-
ants underlying any given trait. In classical twin studies,
lumbar DD as determined by MRI changes in the disc
and anterior osteophyte formation has been shown to
have a quite marked genetic influence: depending on trait
studied and sample characteristics, heritability estimates
range between 0 and 74 % [3, 4, 10, 11]. Longitudinal
change in DD traits has also been shown to be heritable
for all traits except disc height, but there is a significant
influence of age, which varies across traits. In keeping
with other traits, genetic influence in lumbar DD is at its
greatest among individuals younger than 50 years [4].
Whether MC is entirely environmental in origin, or
whether it is influenced by underlying genetic factors,
has not been established. We examined the cross-sec-
tional and longitudinal lumbar MRI available in Twin-
sUK, to address several questions: first, the prevalence of
MC in a relatively healthy population of volunteer twins,
the majority of whom are female; second, the change
over time in a 10-year follow-up; and third, heritability
of MC in a classical twin study.
Materials and methods
Study subjects
Female twin pairs were recruited from TwinsUK register
(www.twinsuk.ac.uk). This sample represented part of an
adult twin cohort recruited since 1992 from the general UK
population [12]. Between 1996 and 2000 unselected twins
were invited to attend for lumbar MRI, a nurse-led inter-
view and collection of demographic data. The twins were
neither selected nor excluded if there was a history of LBP,
neck pain, or disc disease. Similarly, MRI was performed
123
Eur Spine J
on a subsample approximately a decade later. Appropriate
ethics permission had been obtained and twins gave
informed written consent.
Magnetic resonance imaging
Lumbar baseline and follow-up MRI was performed using a
1.0-T equipment (Siemens, Munich, Germany) as previ-
ously reported [11]. Sagittal T2-weighted images of the
lumbar spine were used, using a fast spin-echo sequence of
time to recovery (TR) 5000–4500 ms and time to echo (TE)
112 ms, with a slice thickness of 4 mm. In total, 1,005 twins
attended at baseline and 445 at follow-up for MRI. Baseline
MRI with appropriate data was available for 831 twins and
follow-up for 436 twins. In total, both baseline and follow-
up imaging were available for 347 twins.
A single observer (J.M.) coded all available scans with
no prior knowledge of the clinical status and re-coded a
subset of scans. A second reader (S.W.) also coded a subset
of scans (n = 50) chosen at random for calculation of
inter-rater reliability. MC of both rostral and caudal end-
plates of each lumbar segment was coded using the same
method at baseline and follow-up (Fig. 1a, b). Rostral and
caudal endplates of each of the five lumbar segments were
summed to get the prevalence of MC by segment. The
maximum depth of each MC was graded from all sagittal
slices compared to height of the vertebral body as (0) no
MC, (1) MC in the endplate, (2) MC less than 25 %, (3)
MC from 25 to 50 % and (4) MC more than 50 % of the
height of the vertebral body. The width of MC was eval-
uated dividing the vertebra axially in four peripheral
quadrants and one central circular area. Thereby, width
ranged from 0 to 5. The size index of MC was calculated
multiplying the depth and width of each MC. A segmental
sum score was calculated by adding size of MC of rostral
and caudal endplates together. The subtype of MC was not
recorded due to lack of T1-weighted images.
Reliability of MC coding
For the presence of MC, both the intraobserver reliability
of the first reader (J.M.) and the interobserver reliability
were very good (kappa-values 0.92 and 0.92, respectively).
The intraobserver reliability for the depth and width of MC
ranged from 0.61 to 1.00, and the interobserver reliability
from 0.69 to 1.00, depending on vertebral endplate.
Statistical methods
Probandwise concordance rates were estimated separately
for MZ and DZ twin pairs and for all twin pairs. The
probandwise concordance rate is the number of affected
individuals having an affected co-twin divided by the
Eur Spine J
Fig. 1 a Modic change in both rostral and caudal endplates at L5–S1
level, twin A. b Modic change in both rostral and caudal endplates at
L5–S1 level, twin B
number of affected participants. Heritability for MC was
estimated through standard variance components analysis.
The approach considers MC reported in the twins to have a
potential contribution from additive (A) genetic factors,
environmental factors shared by twin pairs (C) and unique
environmental factors (E). According to the classical twin
model, MZ twin pairs share 100 % of their additive genetic
factors (A), while DZ twin pairs share on average half. MZ
and DZ twin pairs are exposed to the same common
environment (C). The covariance within MZ and DZ twin
pairs can, therefore, be described as follows
CovMZ ¼ A þ C
CovDZ ¼ 0:5A þ C
Accordingly, the influence of additive genetic factors on
a trait (A) can be estimated as twice the difference between
MZ and DZ twin pair covariance. Estimates for C and
E can be achieved as
c2 ¼ C ¼ ICCMZ  H 2 and e2 ¼ E ¼ 1  H 2 þ c2 . The
described formulae have been applied in the structural
equation modelling software Mx [13], where estimates for A,
C and E are determined based on maximum likelihood
estimation [14]. The most suitable model was selected by
comparing models having different combinations of variance
components A, C, and E with the final choice representing
optimal balance between parsimony and goodness of fit.
Results
Data were available for 831 same-sex twin pairs of con-
firmed zygosity, comprising 244 pairs of MZ twins, 446
pairs of DZ twins and 141 unpaired twins (Table 1).
Thirty-five subjects (4.2 %) were male. Mean age at
baseline was 54.1 years (range 45.7–62.5), with mean age
in MZ twins 57.1 years, DZ twins 53.0 years and unpaired
twins 52.2 years.
Modic change
The prevalence of MC (individuals manifesting at least a
single endplate with MC) at baseline was 32.1 %
(Table 2). The incidence of MC during the 10-year follow-
up was 21.6 %, with total prevalence of MC at follow-up
being 48.4 %. At most lumbar levels, the prevalence of MC
was slightly higher in the rostral than in the caudal endplate
(difference not significant). The volume of the MC lesions
increased during follow-up: the median of the MC size
index summary score was 10.0 at baseline and 13.5 at
follow-up (p = 0.012, Table 2).
The incidence of MC was highest at L4–5 and L5–S1 at
follow-up (Table 2). In 12 twins (3.5 %), MC regressed
totally during follow-up (Table 2). Half of these had only
minor MC (MC depth and width at most 2). Regression of
MC was greatest at L5–S1 (Table 2). Comparing MC
incidence among twins with and without MC at baseline,
the total incidence did not differ (31.4 vs. 31.0 %,
123
 Eur Spine J

Table 1 Characteristics of the twins

Zygosity N Age (mean ± SD) Age range Gender (F/M) Baseline MC (controls/cases) Probandwise concordance rate

MZ 244 57.11 ± 8.12 33–73 240/4 162/82 0.56

DZ 446 53.00 ± 8.06 21–70 418/28 303/143 0.39
Singleton 141 52.20 ± 8.49 37–72 138/3 99/42 –
Total 831 54.08 ± 8.38 21–73 796/35 564/267 0.45
The number, mean age, gender and probandwise concordance rates of the twins
MZ monozygotic, DZ dizygotic, SD standard deviation, F female, M male, MC Modic change

Table 2 The prevalence, incidence and regression of MC at lumbar 35

endplates, including rostral and caudal part of every disc, and MC size 31.4 31,0
index sum score of endplates
30
BL MC
MC, MC, New MC Regressed No BL MC
baseline follow-up at follow- MC at 25
(n = 831) (n = 347) up (%) follow-up

Percentage
(%) (%) (%) 20

Total 267 (32.1) 168 (48.4) 75 (21.6) 12 (3.5) 15 14.3 14,0

13,3 13,2
12.4
L1–2 16 (1.9) 21 (6.1) 18 (5.2) –
Rostral 14 (1.7) 16 (4.6) 13 (3.7) – 10 9,1
7.6
6.7
Caudal 11 (1.3) 20 (5.8) 18 (5.2) – 5,4
4,5
5
L2–3 32 (3.9) 43 (12.4) 36 (10.4) 3 (0.9)
Rostral 31 (3.7) 38 (11.0) 31 (8.9) 3 (0.9) 0
Caudal 26 (3.1) 33 (9.5) 27 (7.8) 1 (0.3) L1-L2 L2-L3 L3-L4 L4-L5 L5-S1 Total
Lumbar Level
L3–4 40 (4.8) 36 (10.4) 28 (8.1) 5 (1.4)
Rostral 37 (4.5) 32 (9.2) 24 (6.9) 4 (1.2) Fig. 2 The incidence rates of MC by endplate at follow-up between
Caudal 32 (3.9) 31 (8.9) 25 (7.2) 4 (1.2) twins with MC and without MC at baseline. Twins with prevalent MC
L4–5 114 (13.7) 80 (23.1) 47 (13.5) 4 (1.2) at baseline demonstrated higher incidence at the three highest lumbar
levels during follow-up compared to twins without prevalent MC at
Rostral 110 (13.2) 76 (21.9) 43 (12.4) 4 (1.2) baseline (L1–L4 incidence 25.7 vs. 14.0 %, respectively, p = 0.009)
Caudal 101 (12.2) 73 (21.0) 43 (12.4) 3 (0.9) while at the two lowest lumbar levels no significant difference was
L5–S1 151 (18.2) 92 (26.5) 40 (11.5) 18 (5.2) observed (L4–S1 incidence 25.7 vs. 22.7 %, respectively, p = 0.55).
The total incidence rate did not differ between the groups. BL
Rostral 149 (17.9) 89 (25.6) 39 (11.2) 19 (5.5)
baseline, MC Modic change
Caudal 135 (16.2) 81 (23.3) 35 (10.1) 18 (5.2)
MC size index 10.0 (5.0, 13.5 (8.0,
sum score, 18.0) 21.75) levels, no significant difference was observed (L4–S1
median
(IQR) incidence 25.7 vs. 22.7 %, respectively, p = 0.55, Fig. 2).

The prevalence of MC at baseline and at follow-up is given for each
endplate of the lumbar spine. The third column reports new MC
developing during follow-up (incidence). The fourth column reports
number (%) of regressed MC during follow-up. The rostral and caudal
endplates have been reported separately; however, MC was present at
a given disc level if rostral or caudal or both endplates were affected.
The size index sum score of MC is calculated by multiplying the
depth and width of each MC and summing MC across the lumbar
spine
MC Modic change, SD standard deviation, IQR interquartile range
respectively). Twins with prevalent MC at baseline dem-
onstrated higher incidence at the three highest lumbar
levels during follow-up compared to twins without pre-
valent MC at baseline (L1–L4 incidence 25.7 vs. 14.0 %,
respectively, p = 0.009) while at the two lowest lumbar
Heritability of Modic changes
The prevalence of MC at baseline was 33.6 % in MZ,
32.1 % in DZ and 29.8 % in unpaired twins. Probandwise
concordance rates were higher in case of MZ (0.56) than in
DZ (0.39) suggestive of familial influence. In general, the
risk of MC in a co-twin of an affected individual was 0.45
(Table 1). Heritability estimates of MC were determined
using variance components analysis implemented in Mx
[13] and unpaired twins were included in the modelling to
increase the variance in the MC phenotype. Model fit was
determined for the full ACE model using a Chi squared
statistic and compared to that of reduced models (CE, AE,
E) under the principle of parsimony (Table 3). In the full
model, additive genetic effects [A (95 % CI) = 30

123
Eur Spine J

Table 3 Results of the structural equation modelling

Trait Model fit Model comparison Univariate estimates % (95 % CI)
2 2
Model v df Dv Ddf p value AIC A C E

bl_mc (binary) ACE 1.113 3 – – – – 30 (1–43) 0 (0–20) 70 (57–84)

AE – – 0.000 1 1.000 22.00 30 (16–43) – 70 (57–84)
CE – – 4.107 1 0.043 2.107 – 19 (9–29) 81 (71–91)
E – – 16.890 2 0.000 12.890 – – 100 (100–100)
In the structural equation modelling, an ACE model was fitted to the observed phenotypic variance in Modic change based on a maximum
likelihood estimation. For each trait, three nested models were compared to the full ACE model, taking into account different causal factors: AE
additive genetics and unshared environmental factors, CE shared and unshared environmental factors and E unshared environmental factors.
Model fit for the full and respective reduced models was compared in a likelihood ratio test (Model comparison). Significance of the likelihood
ratio test (p value) is based on the difference in Chi-square statistic (Dv2) and difference in degrees of freedom (Ddf) between the compared
models. In addition, the Akaike’s information criterion (AIC) gives a measure of parsimony of a reduced model in comparison to the full model.
Estimated variances explained by the specific causal factors are given with 95 % confidence intervals for each model. The AE model was found,
by AIC, to provide the best balance between model fit and parsimony (bold)
A additive genetics, C shared environment, E unshared environment

(1–43) %] and unique environmental effects [E (95 %
CI) = 70 (57–84) %] explained most of the variance in
MC, whereas shared environment [C (95 % CI) = 0
(0–20) %] had only minor effect. Comparing the full model
(ACE) to reduced models (CE, AE, E) showed that the AE
model provided equivalent model fit but required fewer
latent parameters (Ddf = 1, AIC = -2.00). So, the nested
AE model was selected as providing the most parsimonious
description of the data with changes in MC variance best
described by additive genetic [A (95 % CI) = 30
(16–43) %] and unique environmental effects [E (95 %
CI) = 70 (57–84) %].
Discussion
It is widely believed that MC contributes to LBP. Associ-
ation with LBP has been found to be strong in many studies
from different populations, particularly for MC type I [15–
17]. LBP is a huge problem socially and is estimated to
cost the UK economy alone over £12 billion per annum.
Lumbar DD is one of the major risk factors of LBP [2, 3],
but few of the genetic variants underlying lumbar DD have
been reliably described [18] and little is known of the
biological pathways leading to DD [19]. While a number of
features of lumbar DD have been shown to be heritable [4],
no study has yet evaluated the heritability of MC.
In this study, probandwise concordance rates were higher
in monozygotic (0.56) than dizygotic twin pairs (0.39), in
keeping with greater sharing of predisposition among MZ
than DZ twins. In general, the risk of MC in a co-twin of an
affected individual was 0.45 (Table 1). These results suggest
that MC is not entirely environmental but does have a small
heritable component. This is an important finding as previ-
ously MC was considered occupationally linked [20]. The
findings indicate that it is reasonable to search for genetic
variants. This would shed light on the biology of endplate
changes, which may actually underpin the degenerative
process in the disc [16].
The prevalence of MC in this sample was 32.1 % at
baseline and 48.4 % at follow-up, at a mean baseline age of
54 years. This is similar to prevalence found in Danish
population: 39 % at age 40 [21]. In another study of gen-
eral population aged 40, the prevalence of MC was 22 %
[17] and in a study of 300 pairs of male twins, the preva-
lence was 55.6 % [22]. During the 10-year follow-up, the
prevalence of MC increased in our study, with incidence of
21.6 %. Kuisma et al. [23] found that 6 % of study subjects
with no MC at baseline had a new incident MC at the
3-year follow-up. The prevalence of MC varies between
studies depending on the consistence of study sample:
subjects’ age, nationality and clinical picture. In addition,
the definition of MC probably varies between the individ-
ual studies. In this study, we excluded minor MC, such as
MC only in one sagittal slice. To the authors’ knowledge,
the current study is the largest follow-up study of MC and
the follow-up period was long enough to enable the natural
course of MC over time.
MC can convert from one type to another. Modic has
demonstrated conversion of type I to type II and type II
changes were found to be stable [6]. We have shown
regression of MC, which is of particular interest because
few studies have reported their natural history. Regression
of MC has been previously described in Danish population;
MC regressed in 23 % of 40-year-old Danes with MC after
4-year follow-up and the majority of those were of type I or
located in the endplate only [21]. In our study, MC was
seen regressed totally in 12 twins (3.5 %), although there
was more regression in individual endplates. In half of
these twins with total MC regression, MC was only minor

123

in size, with height and width at most 2. Unfortunately, we
could not assess accurately the type of MC because for
funding reasons we had only T2-weighted scans available
and thus could not evaluate the proportion of type I of
regressed MC.
It has been suggested that DD comprises two phenotypes:
endplate driven and annulus driven [24]. Endplate-driven
DD usually occurs in upper lumbar level and thoracic spine
and tends to occur at a younger age as structural defects in the
endplate can be congenital [24]. Annulus-driven DD is more
common in the lower lumbar spine and tends to increase with
age as it depends more on accumulating fatigue damage and
age-related biomechanical changes in the annulus [24].
Battié et al. [25] found also the difference of genetic and
environmental influences on DD between upper and lower
lumbar levels. We found the incidence higher at three highest
lumbar levels among twins with baseline MC while at two
lowest lumbar levels no significant difference was observed.
This is in accord with endplate-driven disc degeneration
characteristic of the upper lumbar spine.
We acknowledge a number of limitations to this
study. The sample was predominantly female so it does
not contain subjects reporting heavy manual occupa-
tions. There was no significant difference of MC prev-
alence or age between men and women thus it was
reasonable to include both genders in the sample. These
twins have been shown to be representative of the gen-
eral UK population [26]. The prevalence of MC was not
so high despite reasonable sample size of MZ twins, but
nevertheless is in keeping with other samples of similar
study population age profile. Also, the heritability esti-
mation of the sample was not large and confidence
intervals were wide. It would have been of interest to
examine heritability of MC according to the extent of
lesions and possible differences of heritability between
higher and lower lumbar endplates but we found the
sample size and MC prevalence too small. We were
unable to assess precisely the MC type owing to lack of
T1-weighted images (for funding reasons) so the overall
prevalence and incidence rates represent the sum total of
type I and type II combined. However, we believe that
having only T2-weighted images does not affect the total
MC prevalence assessment and for estimating heritabil-
ity the lack of knowledge on Modic type is not a major
concern particularly as there is no reason to believe
differential expression of MC between monozygotic
(MZ) and dizygotic (DZ) twins. Of course, some of the
MC may have been small type I changes with perhaps a
more benign biological background. However, we have
evaluated some subjects 10 years later and the propor-
tion of MC regression was minimal, suggesting that the
changes seen in most subjects were longstanding.
123
Eur Spine J
Conclusions
In conclusion, this study of a sample of female twins
suggests that MC is heritable. Since MC is associated with
lumbar DD, further work is needed to define the relation-
ship of lumbar DD and LBP. That MC shows a degree of
heritability suggests that it is reasonable to search for
genetic variants. This would provide further explanation of
the biology of endplate change and shed further light on the
biological pathways influencing the degenerative progress
in the disc.
Acknowledgments The authors would like to thank the twins who
volunteered to take part in the study. The study was funded by the
Wellcome Trust; European Community’s Seventh Framework Pro-
gramme (FP7/2007-2013). The study also received support from the
National Institute for Health Research (NIHR) Clinical Research
Facility and Biomedical Research Centre based at Guy’s and St
Thomas’ NHS Foundation Trust and King’s College London, and the
Finnish Academy (Jaro Karppinen, project number 121620). The
authors would like to thank Dr Michael T. Modic, MD for his
invaluable comments on the interpretation of the MR scans.
Conflict of interest None.
Reference
1. Murray CJ, Vos T, Lozano R et al (2013) Disability-adjusted life
years (DALYs) for 291 diseases and injuries in 21 regions,
1990–2010: a systematic analysis for the Global Burden of Dis-
ease Study 2010. Lancet 380:2197–2223
2. Livshits G, Popham M, Malkin I, Sambrook PN, Macgregor AJ,
Spector T, Williams FM (2011) Lumbar disc degeneration and
genetic factors are the main risk factors for low back pain in
women: the UK Twin Spine Study. Ann Rheum Dis
70:1740–1745
3. MacGregor AJ, Andrew T, Sambrook PN, Spector TD (2004)
Structural, psychological, and genetic influences on low back and
neck pain: a study of adult female twins. Arthritis Rheum
51:160–167
4. Williams FM, Popham M, Sambrook PN, Jones AF, Spector TD,
MacGregor AJ (2011) Progression of lumbar disc degeneration
over a decade: a heritability study. Ann Rheum Dis
70:1203–1207
5. Modic MT, Masaryk TJ, Ross JS, Carter JR (1988) Imaging of
degenerative disk disease. Radiology 168(1):177–186
6. Modic MT, Steinberg PM, Ross JS, Masaryk TJ, Carter JR (1988)
Degenerative disk disease: assessment of changes in vertebral
body marrow with MR imaging. Radiology 166:193–199
7. Jensen TS, Karppinen J, Sorensen JS, Niinimaki J, Leboeuf-Yde
C (2008) Vertebral endplate signal changes (Modic change): a
systematic literature review of prevalence and association with
non-specific low back pain. Eur Spine J 17:1407–1422
8. Kuisma M, Karppinen J, Niinimaki J et al (2007) Modic changes
in endplates of lumbar vertebral bodies: prevalence and associ-
ation with low back and sciatic pain among middle-aged male
workers. Spine (Phila Pa 1976) 32:1116–1122
9. Lim CH, Jee WH, Son BC, Kim DH, Ha KY, Park CK (2005)
Discogenic lumbar pain: association with MR imaging and CT
discography. Eur J Radiol 54:431–437
Eur Spine J
10. Battie MC, Videman T, LE Gibbons, Fisher LD, Manninen H,
Gill K (1995) 1995 Volvo Award in clinical sciences. Determi-
nants of lumbar disc degeneration. A study relating lifetime
exposures and magnetic resonance imaging findings in identical
twins. Spine (Phila Pa 1976) 20:2601–2612
11. Sambrook PN, MacGregor AJ, Spector TD (1999) Genetic
influences on cervical and lumbar disc degeneration: a magnetic
resonance imaging study in twins. Arthritis Rheum 42:366–372
12. Moayyeri A, Hammond CJ, Hart DJ, Spector TD (2013) The UK
adult twin registry (twins UK resource). Twin Res Hum Genet
16:144–149
13. Neale MC, Boker SM, Xie G, Maes HH (2006) Mx: statistical
modeling, 7th edn. VCU, Department of Psychiatry, Richmond
14. Rijsdijk FV, Sham PC (2002) Analytic approaches to twin data
using structural equation models. Brief Bioinform 3:119–133
15. Albert HB, Manniche C (2007) Modic changes following lumbar
disc herniation. Eur Spine J 16:977–982
16. Kerttula L, Luoma K, Vehmas T, Gronblad M, Kaapa E (2012)
Modic type I change may predict rapid progressive, deforming
disc degeneration: a prospective 1-year follow-up study. Eur
Spine J 21:1135–1142
17. Kjaer P, Leboeuf-Yde C, Korsholm L, Sorensen JS, Bendix T
(2005) Magnetic resonance imaging and low back pain in adults:
a diagnostic imaging study of 40-year-old men and women. Spine
(Phila Pa 1976) 30:1173–1180
18. Ryder JJ, Garrison K, Song F et al (2008) Genetic associations in
peripheral joint osteoarthritis and spinal degenerative disease: a
systematic review. Ann Rheum Dis 67:584–591
19. Urban JP, Roberts S (2003) Degeneration of the intervertebral
disc. Arthritis Res Ther 5:120–130
20. Mefford J, Sairyo K, Sakai T et al (2011) Modic type I changes of
the lumbar spine in golfers. Skeletal Radiol 40:467–473
21. Jensen TS, Bendix T, Sorensen JS, Manniche C, Korsholm L,
Kjaer P (2009) Characteristics and natural course of vertebral
endplate signal (Modic) changes in the Danish general popula-
tion. BMC Musculoskelet Disord 10:81
22. Wang Y, Videman T, Battié MC (2012) Modic changes: preva-
lence, distribution patterns, and association with age in white
men. Spine J 12:411–416
23. Kuisma M, Karppinen J, Niinimaki J, Kurunlahti M, Haapea M,
Vanharanta H, Tervonen O (2006) A three-year follow-up of
lumbar spine endplate (Modic) changes. Spine (Phila Pa 1976)
31:1714–1718
24. Adams MA, Dolan P (2012) Intervertebral disc degeneration:
evidence for two distinct phenotypes. J Anat 221:497–506
25. Battie MC, Videman T, Levalahti E, Gill K, Kaprio J (2008)
Genetic and environmental effects on disc degeneration by phe-
notype and spinal level: a multivariate twin study. Spine (Phila Pa
1976) 33:2801–2808
26. Andrew T, Hart DJ, Snieder H, de Lange M, Spector TD, Mac-
Gregor AJ (2001) Are twins and singletons comparable? A study
of disease-related and lifestyle characteristics in adult women.
Twin Res 4:464–477
123