Musculoskeletal Imaging Original Research

Joe et al.
MRI of Lumbar Endplate Herniation

Musculoskeletal Imaging
Original Research
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved

Herniation of Cartilaginous
Endplates in the Lumbar Spine:
MRI Findings
Eugene Joe1,2 OBJECTIVE. The purpose of this study was to determine reliable MRI findings suggest-
Joon Woo Lee1 ing disk herniation with cartilage endplate herniation in the lumbar spine.
Kun Woo Park 3 MATERIALS AND METHODS. The records of 73 patients who underwent lumbar spi-
Jin Sup Yeom 3 nal MRI and lumbar microdiskectomy between March 2005 and January 2009 were searched
Eugene Lee1 to find those with the diagnosis of disk herniation with cartilage endplate herniation con-
firmed during surgery. The following morphologic features were assessed: posterior corners
Guen Young Lee1
(posterior marginal nodes, dorsal corner defects, Modic changes, and posterior osteophytes),
Heung Sik Kang1 mid endplates (endplate irregularities, Modic changes), and heterogeneous low signal inten-
Joe E, Lee JW, Park KW, et al. sity of extruded material. The chi-square test and multiple logistic regression analysis with
age, body mass index, and sex as covariates were used for the analysis. The ROC curve was
obtained with scores of the statistically significant MRI findings.
RESULTS. Posterior marginal nodes, posterior osteophytes, Modic changes in posterior
corners, mid endplate irregularities, Modic changes in mid endplates, and heterogeneous low
signal intensity of extruded material were significantly more frequent in patients with disk
herniation with cartilage endplate herniation (0.000 < p < 0.009). The AUC for diagnosing
disk herniation with cartilage endplate herniation with our scoring system of the six MRI
findings (06) was 0.888.
CONCLUSION. The presence of disk herniation with cartilage endplate herniation
could be ascertained with the following MRI findings: posterior marginal nodes, posterior
osteophytes, mid endplate irregularities, heterogeneous low signal intensity of extruded ma-
terial, and Modic changes in posterior corners and mid endplates.

Keywords: cartilage endplate, disk herniation, lumbar
MRI, lumbar spine, MRI
DOI:10.2214/AJR.14.13319
Received June 15, 2014; accepted after revision
September 22, 2014.
1
Department of Radiology, Seoul National University
Bundang Hospital, 300 Gumidong, Bundag-Gu,
Seongnam, Gyeongi-do 463-707, Republic of Korea.
Address correspondence to J. W. Lee
(joonwoo2@gmail.com).
2
Department of Radiology, Gachon University Gil Medical
Center, Incheon, Republic of Korea.
3
Department of Orthopedics, Seoul National University
Bundang Hospital, Seongnam, Republic of Korea.
AJR 2015; 204:10751081
0361803X/15/20451075
American Roentgen Ray Society
L
umbar disk herniation is a condi-
tion in which a tear of the annu-
lus fibrosus allows the nucleus
pulposus to bulge and exert me-
chanical compression on the nerve root. It is
considered one of the key causes of back
pain. An intervertebral disk has three major
structural components: nucleus pulposus,
annulus fibrosus, and cartilaginous endplate
(CEP). Many reports have appeared on the
anatomy, mechanics, and chemistry of the
nucleus and annulus, but little is known
about the MRI characteristics of CEPs and
CEP herniation because their average thick-
ness of the CEP is only 0.6 mm, making it
difficult to recognize at routine MRI exami-
nations [13] (Fig. 1).
A CEP consists of a thin layer of hyaline
cartilage between a disk and a vertebral body.
Functionally, a CEP is considered a gateway
for nutrient transport from blood vessels into
its central disk [4]. It is the most important
factor in controlling diffusion of the disk
proper (nucleus pulposus and annulus fibro-
sus), the only source of nutrition to the nucle-
us pulposus [5]. Endplate damage, which can
start as site-specific focal breaks, can result
in a cascade of events resulting in degenera-
tion. In such a case, a CEP becomes sclerotic
and loses contact with blood vessels, provid-
ing less nutrition to the disk and the CEP it-
self. As a result, the proteoglycan content de-
creases within the disk; the result is loss of
water and osmotic pressure in the disk [6, 7].
The association between the severity of end-
plate damage and disk degeneration has been
documented, and CEP loss is also associat-
ed with disk calcification and rupture (e.g.,
Schmorl nodes, a protrusion of a disk into the
adjacent vertebral body) [5, 8, 9].
Willburger et al. [10] found that a great-
er proportion of the CEP in a herniated disk
correlated with increased pain intensity val-
ue. Brock et al. [11] reported that multiple

AJR:204, May 2015 1075

 Joe et al.

jection period in days were also gathered by retro-

spective chart review.

MR Image Acquisition
At our institution, MRI was performed with ei-
ther a 1.5-T (Gyroscan, Philips Healthcare) or a 3-T
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved

(Intera Achieva, Philips Healthcare) system with

a Sense spine coil for the 3-T system or a Syner-
gy spine coil (both Philips Healthcare) for the 1.5-
T system. The patients were placed in the supine
position with a cushion under both knees. Sagittal
and axial T1-weighted spin-echo images (TR/TE,
500/15) and sagittal and axial T2-weighted fast
spin-echo images (TR/TE, 3600/120) were ob-
tained. The parameters were as follows: slice thick-
ness, 4 mm; slice gap, 0.4 mm; FOV, 32 cm for sag-
Fig. 1Diagram shows disk herniation without (left) and with (right) with cartilage endplate herniation. ittal images and 16 cm for axial images; matrix,
512 512; flip angle, 90; and number of signals
and recurrent sequestered fragments almost in the admission records of the patients electronic acquired, 3. The imported MR images obtained at
always consist of endplate material. Carreon medical charts: age, sex, body mass index (BMI), various hospitals before the referral to our hospital
et al. [12] observed differences in the absorp- symptom duration in days, sensory changes scored were obtained with various MRI systems and vari-
tion of herniated CEPs and annulus fibro- 0100, motor grade scored 05, straight leg raising ous protocols. All images were deidentified before
sus after cervical spinal trauma and report- (SLR) test result, visual analog scale (VAS) score analysis and were randomly numbered.
ed that CEPs exhibited less absorption and for leg pain and back pain, and the Oswestry Dis-
seemed to be the same size. These findings ability Index (ODI) score. Data were gathered by Analysis of MR Images
may reflect metabolic alterations that occur a radiologist after retrospective chart review. The The MR images were assessed by a radiologist
in the course of disk degeneration and ex- number of previous epidural injections and the in- with 10 years of experience as a staff spinal radi-
plain the importance of having a method of
assessing CEPs and CEP herniation. TABLE 1: Demographic Data on 73 Patients With Disk Herniation With or
Some MRI studies have been conduct- Without Cartilage Endplate (CEP) Herniation
ed on CEPs and the relation between CEP
Disk Herniation Without Disk Herniation With
herniation and recurrent disk herniation [5, Characteristic CEP Herniation CEP Herniation pa
1315]. Schmid et al. [15] reported that CEP
herniations are common and that changes in Age (y) 45.58 (10.93) 50.82 (15.28) 0.180
marrow signal intensity of the vertebral end- 60 33 (55) 27 (45) 0.037b
plate on MR images are indicative of carti- > 60 3 (23) 10 (77)
laginous material in the extruded disk herni-
Male-to-female ratio 10:26 18:19 0.067b
ation material. To our knowledge, however,
a reliable method of identifying CEP hernia- Body mass index 24.07 (3.01) 24.27 (3.84) 0.658
tion has not been described. Symptom duration (d) 63.67 (52.44) 74.27 (81.99) 0.675
The main aim of our study was to find a Sensory score (0100) 80.56 (20.56) 72.70 (28.05) 0.312
method of diagnosing disk herniation with
Motor score (05) 4.60 (0.59) 4.63 (0.45) 0.941
CEP herniation and differentiating it from
disk herniation without CEP herniation. Straight leg raising test 48.19 (19.90) 55.68 (22.30) 0.113
result ()

Materials and Methods Visual analog scale

score for pain
Patients
This study was approved by our institutional re- Leg 7.62 (1.63) 8.08 (1.16) 0.221
view board; informed consent was not required be- Back 4.29 (3.28) 4.86 (3.31) 0.477
cause of the retrospective nature of the study. Sev- Oswestry Disability 61.63 (21.83) 62.85 (22.22) 0.825
enty-three patients (45 women, 28 men; mean age, Index score
48 years; range, 2782 years) who underwent mi-
No. of previous 0.97 (1.42) 1.27 (1.56) 0.485
crodiskectomy for lumbar disk herniation between injections
March 2005 and January 2009 and had information
Injection period (d) 58.63 (76.78) 36.24 (42.38) 0.614
in the surgical record about whether CEP herniation
was present were included. All patients underwent NoteExcept for age group and male-to-female ratio, which are numbers of patients with percentages in
parenthesis, data are mean with SD in parentheses.
preoperative lumbar spinal MRI. Before the op- aMann-Whitney test for equality of means unless otherwise indicated.

erations, the following clinical data were assessed bChi-square test with Bonferroni correction.

1076 AJR:204, May 2015


ologist and by an orthopedist, and the two achieved
consensus. They focused on the spinal level that
was operated on after MRI was performed. Both
observers were blinded to the surgical results, radi-
ologic reports, and clinical information except for
operation level. The examinations were reviewed
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved
at random to avoid bias. After the first session of
MRI review, the two readers determined which
items would be assessed on spinal MR images.
These items included lumbar posterior marginal
nodes, dorsal vertebral corner defects, the presence
of Modic changes in the posterior corners, poste-
rior osteophytes, irregularities in the midportion of
an endplate, the presence of Modic changes in the
midportion of an endplate, and heterogeneous low
signal intensity of extruded material.
A lumbar posterior marginal node has been de-
fined as a posterior premarginal intraspongeous her-
niated disk displacing the marginal border posteri-
A B
Fig. 267-year-old woman with disk herniation with cartilage endplate herniation.
A, Sagittal T2-weighted MR image (TR/TE, 3057/100) shows changes in posterior corner: lumbar posterior
marginal node (white arrow), posterior osteophyte (black arrow), and Modic changes (changes in subchondral
bone marrow signal intensity in endplate) (arrowhead).
B, Sagittal T2-weighted MR image (TR/TE, 3057/100) shows mid endplate Modic changes (white arrows) and
endplate irregularity (black arrow) of vertebral body.
Fig. 340-year-old man with disk herniation with
cartilage endplate herniation. Sagittal T2-weighted
MR image (TR/TE, 3300/115) shows changes of
extruded material: heterogeneous low signal
intensity of extruded disk (arrows).
AJR:204, May 2015 1077
MRI of Lumbar Endplate Herniation
orly as an intraosseous hernia with focal avulsion
[1620]. The readers classified an intraosseous her-
niated disk at the posterior corner of a vertebral end-
plate with a bony avulsion, particularly on a sagit-
tal MR image, as a lumbar posterior marginal node.
A dorsal vertebral corner defect was defined, as de-
scribed by previous researchers [15, 21], as a loss
and rounding off of the posterior corner of a verte-
bra on a sagittal MR image. The presence of a Mod-
ic change in the posterior corner was defined as a
change in subchondral bone marrow signal intensity
in the vertebral endplate that included the posterior
corner of a vertebral body [22, 23]. A posterior os-
teophyte was considered present when a bony pro-
jection was seen at the posterior corner of a verte-
bral body [24, 25]. A mid endplate irregularity was
considered present when the linearity and integrity
of the middle portion of an endplate was lost on a
sagittal MR image. The presence of a mid endplate
Fig. 4 45-year-old man with disk herniation with
cartilage endplate herniation. Sagittal T2-weighted
MR image (TR/TE, 3708/120) shows dorsal vertebral
corner defect (arrow).
Modic change was also defined as a change in sub-
chondral bone marrow signal intensity in a vertebral
endplate that included the midportion of the end-
plate [22, 23]. Heterogeneous low signal intensity
of extruded material was recorded when the signal
intensity of a herniated disk was heterogeneous and
had an area of low signal intensity compared with
the signal intensity of the mother disk.
The assessed MRI findings were further classi-
fied into three categories: morphologic features of
the posterior corners (lumbar posterior marginal
nodes, dorsal corner defects, Modic changes, and
posterior osteophytes), mid endplates (endplate ir-
regularities, Modic changes), and extruded material
(Figs. 24).
Data and Statistical Analyses
Patients were assigned to one of two groups:
disk herniation with CEP herniation or disk her-
niation without CEP herniation. To determine
the differences between the two groups, a chi-
square test was used to compare categoric vari-
ables (sex, grouped clinical data, MRI findings).
Differences between the quantitative variables
(age, height, weight, BMI, symptom duration,
motor power, sensory change, angle at SLR test,
VAS score, ODI score) of the two groups were
analyzed with the Mann-Whitney test. Statisti-
cally significant variables were also analyzed by
multiple logistic regression to check for the in-
fluence of age, BMI, and sex as covariates. Lin-
ear by linear association was used for the trend
assessment of the MRI scoring system to deter-
mine the presence of disk herniation with CEP
herniation. ROC analysis was also performed to
differentiate disk herniation with and disk herni-
ation without CEP herniation. The optimal cutoff
point was defined as the value at which the sum
of the sensitivity and specificity was maximized.
SPSS software (version 17, IBM SPSS) was used
for all calculations. The level of statistical signif-
icance was set at 0.05.
Results
Demographic Data
The demographic data are summarized
in Table 1. Thirty-seven patients (18 men,
19 women; mean age, 51 years; range, 27
82 years) were found at surgery to have disk
herniation with CEP herniation. Disk hernia-
tion with CEP herniation was more common
in patients older than 60 years (10/13, 77%)
than in those younger than 60 years (27/60,
45%) (p = 0.037). There was no significant
difference in terms of mean symptom dura-
tion, sensory change, motor weakness, SLR
test result, VAS score, ODI score, or number
of previous epidural injections.
 Joe et al.

TABLE 2: Comparison of MRI Results Among 73 Patients With Disk MRI Findings
Herniation With or Without Cartilage Endplate (CEP) Herniation The MRI findings are summarized in Ta-
ble 2. Lumbar posterior marginal nodes were
Disk Herniation Disk Herniation
Without CEP With CEP significantly more frequent in patients with
Characteristic Herniation Herniation pa pb (23/37, 62%) than in those without (6/36,
17%) CEP herniation (p < 0.001). Posteri-
Lumbar posterior marginal nodes < 0.001 < 0.001
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved

or osteophytes were also significantly more

Absent 30 14 frequent in patients with disk herniation with
Present 6 23 (16/37, 43%) than in those without (3/36, 8%)
Posterior osteophytes 0.001 0.001 CEP herniation (p = 0.001). Modic changes
in the posterior corners were more common
Absent 33 21
in patients with (26/37, 70%) than in those
Present 3 16 without (14/36, 39%) CEP herniation (p =
Modic changes < 0.001 0.001 0.009). Mid endplate Modic changes (21/37,
Absent 20 6 57%) and mid endplate irregularities (24/37,
65%) were significantly more frequent in pa-
Present 16 31
tients with CEP herniation than in those with-
Modic changes in posterior corners 0.007 0.009 out CEP herniation (6/36, 17%; 8/36, 22%)
Absent 22 11 (p = 0.001; p = 0.001). Heterogeneous low
Present 14 26 signal intensity of herniated disks was sig-
nificantly more frequent in patients with disk
Modic changes in mid endplate < 0.001 0.003
herniation with CEP herniation (19/37, 51%)
Absent 30 16 than in those without CEP herniation (5/36,
Present 6 21 14%) (p = 0.001). The dorsal corner defects,
Mid endplate irregularities < 0.001 0.001 disk levels, volume of disk herniation, pres-
ence of disk migration, and disk degenera-
Absent 28 13
tion in patients with disk herniation with and
Present 8 24 those without CEP herniation were not sig-
Heterogeneous signal intensity of 0.001 0.001 nificantly different (Figs. 5 and 6).
extruded disks MRI scoring was conducted to diagnose
Absent 31 18 disk herniation with CEP herniation accord-
Present 5 19 ing to the six statistically significant MRI
findings (lumbar posterior marginal nodes,
Dorsal vertebral corner defects 0.172 0.263
presence of Modic changes in posterior cor-
Absent 13 8 ners, posterior osteophytes, mid endplate ir-
Present 23 29
TABLE 3: Scores of Six Statistically
Disk level 0.121 0.761 Significant MRI Findings
L2-3 1 0 in Patients With Disk
L3-4 3 1
Herniation With or
Without Cartilage Endplate
L4-5 14 24 (CEP) Herniation
L5-S1 18 12
Disk
Volume of disk herniation 0.436 0.443 Herniation Disk
Without Herniation
Mild 15 11
CEP With CEP
Moderate 18 24 Score Herniation Herniation Total
Severe 3 2 0 12 1 13
Disk migration 0.935 0.476 1 12 1 13
Absent 12 12 2 8 9 17
Present 24 25 3 2 6 8
Disk degeneration (Pfirrmann type) 0.208 0.988 4 2 11 13
C 24 19 5 0 6 6
D 12 6 6 0 3 3
E 0 2 Total 36 37 73
NoteData are numbers of patients. NoteSignificant positive trend was found for MRI
aChi-square test.
score and CEP herniation (p < 0.001) (linear by linear
bMultiple logistic regression analysis with age, body mass index, and sex as covariates.
association).

1078 AJR:204, May 2015

 MRI of Lumbar Endplate Herniation

regularities, presence of Modic changes in

mid endplates, and heterogeneous low signal
intensity of extruded material) (Fig. 7). The
presence of each finding was given 1 point,
so scores ranged from 0 to 6. A statistically
significant positive trend was found between
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved

the MRI score and disk herniation with CEP

herniation (p < 0.001). No patient with disk
herniation without CEP herniation had a
score of 5 or 6 (Table 3 and Fig. 8). The ROC
curve of this scoring system was also gener-
ated. The AUC for diagnosing disk hernia-
tion with CEP herniation using the scoring
system was 0.888. When the MRI scoring
Fig. 537-year-old man with disk herniation without Fig. 667-year-old woman with disk herniation with
criterion for disk herniation with CEP herni- cartilage endplate herniation. Sagittal T2-weighted MR cartilage endplate herniation. Sagittal T2-weighted
ation was set at 2.5, the sensitivity and speci- image (TR/TE, 3100/100) shows only dorsal vertebral MR image (TR/TE, 3057/100) shows mid endplate
ficity were 70% and 90% (Fig. 9). corner defect at superior corner of S1 endplate. irregularity, Modic changes (change in subchondral
bone marrow signal intensity in endplate) in mid
endplate and posterior corner, posterior osteophyte,
Discussion be considered if patients with radiculopathy small lumbar posterior marginal node, and
The results of this study suggest that the have MRI findings suggestive of disk hernia- heterogeneous signal intensity of extruded material.
following six MRI findings are useful for the tion with CEP herniation.
diagnosis of disk herniation with CEP herni- There have been several reports on the tis- microsurgical removal of herniated material,
ation and for differentiating it from disk her- sue composition of herniated material [11, 27 25 patients had hyaline cartilaginous materi-
niation without CEP herniation: lumbar pos- 29]. Schmid et al. [15] compared preoperative al in their extrusions. Similarly, in our study,
terior marginal nodes, Modic changes in the MRI findings and the tissue composition of among 73 patients who had herniated mate-
posterior corners, posterior osteophytes, mid herniated disks. They reported that among 51 rial microsurgically removed, 37 patients had
endplate irregularities, Modic changes in the patients with herniated disks who underwent cartilaginous material in their extruded disks.
mid endplates, and heterogeneous low signal
intensity of extruded material. According to
results of previous studies [12, 26], herniated
disks with CEP herniation are less absorp- 12
tive than herniated disks without CEP herni-
ation. Consequently, patients with disk her-
niation with CEP herniation are more likely
10
to have recurrent episodes of radiculopa-
thy and to need surgical treatment [11, 26].
Therefore, we suggest that early operation
8
No. of Patients

Fig. 7Diagram shows six MRI findings suggesting 0

disk herniation with cartilage endplate herniation: 0 1 2 3 4 5 6
posterior marginal nodes, posterior osteophytes, mid
endplate irregularities, heterogeneous low signal MRI Score
intensity of extruded material, and Modic changes
(changes in subchondral bone marrow signal Fig. 8Histogram shows number of patients who had disk herniation with (gray) and without (black) cartilage
intensity in endplate) in posterior corners and mid endplate (CEP) herniation according to scoring system with six statistically significant MRI findings (06). No
endplates. patient with score of 5 or 6 had CEP herniation.

AJR:204, May 2015 1079


1.0
0.8
0.6
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved
Sensitivity
0.4
0.2
0.0
0.0 0.2 0.4 0.6 0.8
1 Specificity
Fig. 9Graph shows ROC curve determined with
scoring system using six statistically significant MRI
findings (06) for diagnosis of disk herniation with
cartilage endplate. Area under ROC curve is 0.888.
In both studies, approximately 50% of the
patients with disk herniation had CEP her-
niation. Both studies also showed a positive
correlation between marrow signal intensity
changes along CEPs and the presence of carti-
laginous material in herniated disks, and both
showed that when the changes extended to
33% of the vertebral endplates, CEP material
was present in the extruded disks. Modic et al.
[22] previously found that changes in marrow
signal intensity in vertebral endplates are as-
sociated with vertebral endplate fissures. Be-
cause an endplate avulsion injury must have
preceded the process of disk herniation, if hy-
aline cartilage is found in the herniated ma-
terial, signs of endplate degeneration, such as
Modic changes, can be regarded as signs of
CEP herniation. Thus, even though we cannot
directly see the CEPs themselves on the MR
images, because they are extremely thin, we
can diagnose CEP herniation using the MRI
findings that indicate endplate degeneration.
In addition to Modic changes, lumbar poste-
rior marginal nodes, endplate irregularities,
dorsal vertebral corner defects, and posteri-
or osteophytes were thought to be reasonable
signs of endplate degeneration.
According to our results, heterogeneous
disk signal intensity was significantly re-
lated to CEP herniation. Either the CEPs
themselves affect the heterogeneous signal
intensity of extruded material, or disk degen-
eration affects the heterogeneous signal in-
tensity of herniated disks.
Schmid et al. [15] reported that vertebral
corner defects were more common in CEP
herniation. We also evaluated dorsal verte-
1080 AJR:204, May 2015
Joe et al.
bral corner defects, but contrary to our ex-
pectation, they were not significantly corre-
lated with CEP herniations. Dorsal vertebral
corner defects were easily seen in the corner
of disk herniations simply as herniated disk
material covering the posterior corner of the
vertebral body on MR images, so they may
not be related to endplate degeneration.
In our study, disk herniation with CEP
herniation was more commonly found in
patients older than 60 years. This finding is
concordant with those in previous histopath-
ologic studies showing that disk herniation
with CEP herniation seems to be the pre-
1.0
dominant type of disk herniation in elderly
persons [15, 27, 30, 31].
Willburger et al. [10] reported that a high
percentage of cartilage in herniated disks cor-
related with increased pain intensity values, but
in our study there was no significant correla-
tion between CEP herniation and VAS or ODI
scores. Kokkonen et al. [2] reported no corre-
lation between endplate degeneration and pain
provocation at diskography and a positive cor-
relation between disk rupture and pain provoca-
tion. It seems that endplate degeneration or the
presence of CEP in extruded disks is not direct-
ly correlated with pain intensity. Our study did
not show any positive correlation between CEP
herniation and clinical symptoms, such as pain,
sensory changes, motor changes, or the results
of an SLR test. These results could be explained
by the inflammatory effect of the nucleus pulp-
osus as a confounder [3237]. The nucleus
pulposus has been reported to provoke physi-
ologic changes in the nerve root regardless of
the mechanical compression effect [35]. On the
other hand, a correlation has been reported be-
tween compression pressure and the duration
of impaired nerve root conduction [3840].
Therefore, it may be a premature determination
that there was no correlation between the com-
position of the extruded disk material and clini-
cal symptoms.
Our study had several limitations. First, the
study population was relatively small. Fur-
ther prospective studies with a larger number
of patients may confirm our results. Second,
MRI findings were assessed in a consensus-
based manner. We primarily focused on ex-
amining meaningful MRI findings to detect
disk herniation with CEP herniation rather
than assessing intraobserver or interobserv-
er reliability. Third, the presence of CEP was
determined in the operative field on the ba-
sis of gross morphologic features because ex-
perienced surgeons can discern the CEP at
gross morphologic inspection.
Conclusion
The presence of CEP herniation can be as-
certained with the following MRI findings:
lumbar posterior marginal nodes, posterior
osteophytes, mid endplate irregularities, het-
erogeneous low signal intensity of extruded
material, and Modic changes in the posterior
corners and midportion of the endplates.
References
1. Roberts S, Menage J, Urban JP. Biochemical and
structural properties of the cartilage end-plate and
its relation to the intervertebral disc. Spine 1989;
14:166174
2. Kokkonen SM, Kurunlahti M, Tervonen O, Ilkko
E, Vanharanta H. Endplate degeneration observed
on magnetic resonance imaging of the lumbar
spine: correlation with pain provocation and disc
changes observed on computed tomography disk-
ography. Spine 2002; 27:22742278
3. Grant JP, Oxland TR, Dvorak MF. Mapping the
structural properties of the lumbosacral vertebral
endplates. Spine 2001; 26:889896
4. Roberts S, Menage J, Eisenstein SM. The carti-
lage end-plate and intervertebral disc in scoliosis:
calcification and other sequelae. J Orthop Res
1993; 11:747757
5. Rajasekaran S, Babu JN, Arun R, Armstrong BR,
Shetty AP, Murugan S. ISSLS prize winner: a
study of diffusion in human lumbar discsa serial
magnetic resonance imaging study documenting
the influence of the endplate on diffusion in normal
and degenerate discs. Spine 2004; 29:26542667
6. Lyons G, Eisenstein SM, Sweet MB. Biochemical
changes in intervertebral disc degeneration. Bio-
chim Biophys Acta 1981; 673:443453
7. Raj PP. Intervertebral disc: anatomy-physiology-
pathophysiology-treatment. Pain Pract 2008;
8:1844
8. Boos N, Weissbach S, Rohrbach H, Weiler C, Spratt
KF, Nerlich AG. Classification of age-related
changes in lumbar intervertebral discs: 2002 Volvo
Award in basic science. Spine 2002; 27:26312644
9. Roberts S, Urban JP, Evans H, Eisenstein SM.
Transport properties of the human cartilage end-
plate in relation to its composition and calcifica-
tion. Spine 1996; 21:415420
10. Willburger RE, Ehiosun UK, Kuhnen C, Kramer
J, Schmid G. Clinical symptoms in lumbar disc
herniations and their correlation to the histologi-
cal composition of the extruded disc material.
Spine 2004; 29:16551661
11. Brock M, Patt S, Mayer HM. The form and structure
of the extruded disc. Spine 1992; 17:14571461
12. Carreon LY, Ito T, Yamada M, Uchiyama S, Taka-
hashi H, Ikuta F. Histologic changes in the disc
after cervical spine trauma: evidence of disc ab-
sorption. J Spinal Disord 1996; 9:313316

13. Becker GT, Willburger RE, Liphofer J, Koester O,
Schmid G. Distribution of MRI signal alterations
of the cartilage endplate in pre-operated patients
with special focus on recurrent lumbar disc her-
niation [in German]. Rofo 2006; 178:4654
14. Bae WC, Statum S, Zhang Z, et al. Morphology of
Downloaded from www.ajronline.org by 36.82.102.157 on 03/04/17 from IP address 36.82.102.157. Copyright ARRS. For personal use only; all rights reserved
the cartilaginous endplates in human interverte-
bral disks with ultrashort echo time MR imaging.
Radiology 2013; 266:564574
15. Schmid G, Witteler A, Willburger R, Kuhnen C,
Jergas M, Koester O. Lumbar disk herniation:
correlation of histologic findings with marrow
signal intensity changes in vertebral endplates at
MR imaging. Radiology 2004; 231:352358
16. Laredo JD, Bard M, Chretien J, Kahn MF. Lumbar
posterior marginal intra-osseous cartilaginous
node. Skeletal Radiol 1986; 15:201208
17. Dietemann JL, Runge M, Badoz A, et al. Radiology
of posterior lumbar apophyseal ring fractures: re-
port of 13 cases. Neuroradiology 1988; 30:337344
18. Ikata T, Morita T, Katoh S, Tachibana K, Maoka
H. Lesions of the lumbar posterior end plate in
children and adolescents: an MRI study. J Bone
Joint Surg Br 1995; 77:951955
19. Leroux JL, Fuentes JM, Baixas P, Benezech J,
Chertok P, Blotman F. Lumbar posterior marginal
node (LPMN) in adults: report of fifteen cases.
Spine 1992; 17:15051508
20. Cao LB, Zhan AL, Cao QX. Imaging study of
lumbar posterior marginal intraosseous node: an
analysis of 36 cases. Chin Med J (Engl) 1992;
105:866869
21. Vincent JM, Baldwin JE, Sims C, Dixon AK. Ver-
tebral corner defect associated with lumbar disk
herniation shown by magnetic resonance imag-
ing. Spine 1993; 18:109113
22. Modic MT, Steinberg PM, Ross JS, Masaryk TJ,
AJR:204, May 2015 1081
MRI of Lumbar Endplate Herniation
Carter JR. Degenerative disk disease: assessment
of changes in vertebral body marrow with MR
imaging. Radiology 1988; 166:193199
23. Jones A, Clarke A, Freeman BJ, Lam KS, Grevitt
MP. The Modic classification: inter- and intraob-
server error in clinical practice. Spine 2005;
30:18671869
24. Vernon-Roberts B, Pirie CJ. Degenerative changes
in the intervertebral discs of the lumbar spine and
their sequelae. Rheumatol Rehabil 1977; 16:1321
25. Farfan HF. The pathological anatomy of degener-
ative spondylolisthesis: a cadaver study. Spine
1980; 5:412418
26. Ross P, Jelsma F. Postoperative analysis of 366
consecutive cases of herniated lumbar discs. Am J
Surg 1952; 84:657662
27. Kokubun S, Sakurai M, Tanaka Y. Cartilaginous
endplate in cervical disc herniation. Spine 1996;
21:190195
28. Eckert C, Decker A. Pathological studies of interver-
tebral discs. J Bone Joint Surg Am 1947; 29:447454
29. Moore RJ, Vernon-Roberts B, Fraser RD, Osti
OL, Schembri M. The origin and fate of herniated
lumbar intervertebral disc tissue. Spine 1996;
21:21492155
30. Tanaka M, Nakahara S, Inoue H. A pathologic
study of discs in the elderly: separation between
the cartilaginous endplate and the vertebral body.
Spine 1993; 18:14561462
31. Harada Y, Nakahara S. A pathologic study of lum-
bar disc herniation in the elderly. Spine 1989;
14:10201024
32. Kawakami M, Weinstein JN, Spratt KF, et al. Ex-
perimental lumbar radiculopathy: immunohisto-
chemical and quantitative demonstrations of pain
induced by lumbar nerve root irritation of the rat.
Spine 1994; 19:17801794
33. Kawakami M, Tamaki T, Matsumoto T, Kuriba
yashi K, Takenaka T, Shinozaki M. Role of leuko-
cytes in radicular pain secondary to herniated
nucleus pulposus. Clin Orthop Relat Res 2000;
376:268277
34. McCarron RF, Wimpee MW, Hudkins PG, Laros
GS. The inflammatory effect of nucleus pulposus:
a possible element in the pathogenesis of low-back
pain. Spine 1987; 12:760764
35. Olmarker K, Rydevik B, Nordborg C. Autologous
nucleus pulposus induces neurophysiologic and
histologic changes in porcine cauda equina nerve
roots. Spine 1993; 18:14251432
36. Olmarker K, Rydevik B. Selective inhibition of tu-
mor necrosis factor-alpha prevents nucleus pulpo-
sus-induced thrombus formation, intraneural ede-
ma, and reduction of nerve conduction velocity:
possible implications for future pharmacologic
treatment strategies of sciatica. Spine 2001;
26:863869
37. Takebayashi T, Cavanaugh JM, Cuneyt Ozaktay
A, Kallakuri S, Chen C. Effect of nucleus pulpo-
sus on the neural activity of dorsal root ganglion.
Spine 2001; 26:940945
38. Lind B, Massie JB, Lincoln T, Myers RR, Swen-
son MR, Garfin SR. The effects of induced hyper-
tension and acute graded compression on impulse
propagation in the spinal nerve roots of the pig.
Spine 1993; 18:15501555
39. Olmarker K, Holm S, Rydevik B. Importance of
compression onset rate for the degree of impair-
ment of impulse propagation in experimental
compression injury of the porcine cauda equina.
Spine 1990; 15:416419
40. Pedowitz RA, Garfin SR, Massie JB, et al. Effects
of magnitude and duration of compression on spi-
nal nerve root conduction. Spine 1992; 17:194199