cervical spinal stenosis: determination with vertebral ... · the cervical spine and last afew...

5
Helene Pavlov, MD #{149} Joseph S. Torg, MD #{149} Bruce Robie, MS #{149} Caren Jahre, MD Cervical Spinal Stenosis: Determination with Vertebral Body Ratio Method’ 771 Transient bilateral sensory and mo- tor symptoms after trauma, includ- ing complete paralysis, have been identified in patients with cervical spinal stenosis. Radiographs of 23 patient athletes with cervical spinal neurapraxia were used for measure- ment of the cervical spinal canal. Two methods of measurement were used. In the conventional method, sagittal diameter is measured from the posterior surface of the verte- bra! body to the nearest point of the corresponding laminar line. In the ratio method, the sagittal diameter of the spinal canal is divided by the sagittal diameter of the correspond- ing vertebra! body. Results indicate the ratio method is reliable for de- termining cervical spinal stenosis and is independent of technical fac- tor variables. Index terms: Spinal canal, stenosis, 31.77 #{149} Spine, abnormalities #{149} Spine. injuries, 31.4 S Spine, radiography, 31.11 Radiology 1987; 164:771-775 1 From the Departments of Radiology (H.P., C.J.) and Biomechanics (BR.), Hospital for Spe- cial Surgery. 535 E. 70th Street, New York, NY iOO2l, affiliated with New York Hospital-Cor- nell University Medical College. New York; and the Department of Orthopaedic Surgery, University of Pennsylvania. Philadelphia (J.S.T.). Received September 1 1, 1986; revision requested October 22; revision received April 6, 1987; accepted April 8. Address reprint re- quests to H.P. e RSNA, 1987 W E have previously reported a clinical and radiographic me- view of the data on 31 patients with transient cervical spinal neurapraxia (1). These patients are young athletes who experienced sudden sensory changes that may be associated with motor symptoms in both arms, both legs, on all four extremities. Sensory changes include numbness, burning, tingling, and paresthesia. Motor symptoms include weakness and complete paralysis. These symptoms are precipitated by sudden hypenex- tension, hyperflexion, or axial load to the cervical spine and last a few sec- onds to 36 hours, with complete me- covemy. Cervical spine radiogmaphs in these patients demonstrate congeni- tal anomalies, disk disease, instabil- ity, or “normal” findings. Although cervical spinal stenosis was suspected clinically, standard radiographic measurement of the sagittal canal di- ameten was not always confirmatory. The purposes of this paper are (a) to introduce the ratio method of deter- mining cervical spinal stenosis to the radiographic literature, (b) to deter- mine the ratio values in male and fe- male control groups, and (c) to ex- plain the radiographic rationale for the use of this method. REVIEW OF LITERATURE Multiple investigators have report- ed the measurement of the sagittal diameter of the cervical spinal cord as a means of diagnosing spinal ste- nosis. These reports have resulted in inconsistencies in “normal” and “ab- normal” values for the sagittal diam- eten of the cervical spine. Of the van- ous techniques and measurements reported, the most commonly em- ployed method for determining the sagittal diameter of the spinal canal makes use of a lateral view of the cer- vical spine and measures the distance from the cephalocaudal midpoint (middle of the posterior surface) of the vertebral body to the nearest point of the corresponding spinal laminar line, which Wilkinson et al. (2) called the preexisting sagittal di- ameter (2-8). Using this technique, Boijsen (3) reported the average non- mal sagittal diameter for C-4 to C-6 to be 18.5 mm (range, 14.2-23 mm), and Hinck et al. (4) reported the average measurement for C-3 to C-5 to be 17.0 mm (range, 13.9-20.3 mm); in both studies, the investigators used a tan- get distance of 5 feet. Moiel et al. (5), Payne and Spillane (6), and Wolf et al. (8) determined the average nor- mal sagittal diameter of the cervical spine to be 17.0 mm ± 5; these inves- tigatons used a target distance of 6 feet. Using this method, Epstein et al. (9, 10) and Countee and Vijayanathan (ii) reported severe spinal stenosis at less than 13 mm and less than 14 mm, respectively. Wilkinson et al. (2) reported mea- sunements from four clinical sub- groups: one group of patients with neck and arm pain but without neu- nologic symptoms and three groups of patients with neunologic symp- toms. The cervical sagittal diameter was largest in those patients without radiculan pain; the average measure- ment for C-3 to C-5 was 16.6 mm. A target distance of 6 feet was used. Chnispin and Lees (12) found that on the lateral radiograph, the project- ed canal area of myelopathic patients was smaller than the projected verte- bral body area, but that in nonmyelo- pathic subjects, the projected canal and vertebral body areas were ap- proximately equal. They determined the actual weights of these areas by using tracings and cutouts of the ma- diognaphs. They concluded that in patients with cervical spondylolysis and myelopathy, the ratio of the area of the spinal canal to the area of the vertebral body was less than 85%. Ehni (13) simplified their observa- tion by assuming that since the verti- cal heights of the spinal canal and

Upload: others

Post on 23-Oct-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

  • Helene Pavlov, MD #{149}Joseph S. Torg, MD #{149}Bruce Robie, MS #{149}Caren Jahre, MD

    Cervical Spinal Stenosis: Determinationwith Vertebral Body Ratio Method’

    771

    Transient bilateral sensory and mo-tor symptoms after trauma, includ-ing complete paralysis, have beenidentified in patients with cervicalspinal stenosis. Radiographs of 23patient athletes with cervical spinalneurapraxia were used for measure-ment of the cervical spinal canal.Two methods of measurement wereused. In the conventional method,sagittal diameter is measured fromthe posterior surface of the verte-bra! body to the nearest point of thecorresponding laminar line. In theratio method, the sagittal diameterof the spinal canal is divided by thesagittal diameter of the correspond-ing vertebra! body. Results indicatethe ratio method is reliable for de-termining cervical spinal stenosisand is independent of technical fac-tor variables.

    Index terms: Spinal canal, stenosis, 31.77

    #{149}Spine, abnormalities #{149}Spine. injuries, 31.4

    S Spine, radiography, 31.11

    Radiology 1987; 164:771-775

    1 From the Departments of Radiology (H.P.,C.J.) and Biomechanics (BR.), Hospital for Spe-cial Surgery. 535 E. 70th Street, New York, NYiOO2l, affiliated with New York Hospital-Cor-nell University Medical College. New York;and the Department of Orthopaedic Surgery,University of Pennsylvania. Philadelphia(J.S.T.). Received September 1 1, 1986; revision

    requested October 22; revision received April

    6, 1987; accepted April 8. Address reprint re-quests to H.P.

    e RSNA, 1987

    W E have previously reported aclinical and radiographic me-

    view of the data on 31 patients with

    transient cervical spinal neurapraxia

    (1). These patients are young athletes

    who experienced sudden sensory

    changes that may be associated with

    motor symptoms in both arms, both

    legs, on all four extremities. Sensory

    changes include numbness, burning,

    tingling, and paresthesia. Motor

    symptoms include weakness and

    complete paralysis. These symptoms

    are precipitated by sudden hypenex-

    tension, hyperflexion, or axial load to

    the cervical spine and last a few sec-

    onds to 36 hours, with complete me-

    covemy. Cervical spine radiogmaphs in

    these patients demonstrate congeni-

    tal anomalies, disk disease, instabil-

    ity, or “normal” findings. Although

    cervical spinal stenosis was suspected

    clinically, standard radiographic

    measurement of the sagittal canal di-

    ameten was not always confirmatory.

    The purposes of this paper are (a) to

    introduce the ratio method of deter-

    mining cervical spinal stenosis to the

    radiographic literature, (b) to deter-

    mine the ratio values in male and fe-

    male control groups, and (c) to ex-

    plain the radiographic rationale for

    the use of this method.

    REVIEW OF LITERATURE

    Multiple investigators have report-

    ed the measurement of the sagittal

    diameter of the cervical spinal cord

    as a means of diagnosing spinal ste-

    nosis. These reports have resulted in

    inconsistencies in “normal” and “ab-

    normal” values for the sagittal diam-

    eten of the cervical spine. Of the van-

    ous techniques and measurements

    reported, the most commonly em-

    ployed method for determining the

    sagittal diameter of the spinal canalmakes use of a lateral view of the cer-

    vical spine and measures the distance

    from the cephalocaudal midpoint

    (middle of the posterior surface) of

    the vertebral body to the nearest

    point of the corresponding spinal

    laminar line, which Wilkinson et al.

    (2) called the preexisting sagittal di-

    ameter (2-8). Using this technique,

    Boijsen (3) reported the average non-

    mal sagittal diameter for C-4 to C-6 to

    be 18.5 mm (range, 14.2-23 mm), andHinck et al. (4) reported the average

    measurement for C-3 to C-5 to be 17.0

    mm (range, 13.9-20.3 mm); in both

    studies, the investigators used a tan-

    get distance of 5 feet. Moiel et al. (5),

    Payne and Spillane (6), and Wolf et

    al. (8) determined the average nor-

    mal sagittal diameter of the cervical

    spine to be 17.0 mm ± 5; these inves-

    tigatons used a target distance of 6

    feet. Using this method, Epstein et al.

    (9, 10) and Countee and Vijayanathan

    (ii) reported severe spinal stenosis at

    less than 13 mm and less than 14 mm,

    respectively.

    Wilkinson et al. (2) reported mea-

    sunements from four clinical sub-

    groups: one group of patients with

    neck and arm pain but without neu-

    nologic symptoms and three groups

    of patients with neunologic symp-

    toms. The cervical sagittal diameter

    was largest in those patients without

    radiculan pain; the average measure-

    ment for C-3 to C-5 was 16.6 mm. A

    target distance of 6 feet was used.

    Chnispin and Lees (12) found that

    on the lateral radiograph, the project-

    ed canal area of myelopathic patients

    was smaller than the projected verte-

    bral body area, but that in nonmyelo-

    pathic subjects, the projected canal

    and vertebral body areas were ap-

    proximately equal. They determined

    the actual weights of these areas by

    using tracings and cutouts of the ma-

    diognaphs. They concluded that in

    patients with cervical spondylolysis

    and myelopathy, the ratio of the area

    of the spinal canal to the area of the

    vertebral body was less than 85%.

    Ehni (13) simplified their observa-

    tion by assuming that since the verti-

    cal heights of the spinal canal and

  • 2.0

    .�u

    0

    0

    .:

    11

    q

    �. .�‘ .-.0b. b. b. -

    I I I

    0’ 0’. CO ‘-4

    - - �

    0000

    -H-H-H-H� b. 0C’) - Lfl -

    �, � c4� �.

    * 4- 4� ‘.0�

    0 0 0’. 0(‘4 (‘.4 - (‘4

    I I I I,,� b. �m c’� � U)

    - ,�. �U) c�) c#{149}�

    -H-H-H-Ha’. c’.4 �- 0’. 0 U)b. ‘0 b.

    - - b. LI)

    C’) (‘1 -m m0 0 0

    0 0 0 0

    -H41-H41(4 �U) �L‘0 ‘.0 ‘0

    0 0 0 ci

    If) � U)’0

    I I I I‘-4 U) 0’ b.00 � 00 �0 0 0 ci‘.0 ‘-4 b. 000 b. U) b.- 0� 0 00 0 0 0

    -H-H-H-H00 ‘-4 ‘0’0

    04u

    .0�u

    U)� 4..

    kUa)a

    U U

    r

    a)

    LI) U)C’) C’)

    c�;Ie;I

    C’) �

    00 U)

    00 0’.

    -H-H�r_4 U)a’ 00

    U) U)

    C’) t�)c’.4r;It.&0� LI)

    C’) C’)

    00 0’

    00

    -H-H- C’)

    00 0’

    Figure 1. The sagittal diameter of the spi-nal canal (a) is measured from the posteriorsurface of the vertebral body to the nearestpoint of the corresponding spinal laminarline. The sagittal diameter of the vertebral

    body (b) is measured at the midpoint, fromthe anterior surface to the posterior surface.The spinal canal/vertebral body ratio is de-termined with the formula a/b. The normalratio is approximately 1.00.

    the vertebral body are the same, theheight can be eliminated in the for-

    mula and, instead of area, the anteno-

    posterior diameter of the spinal canal

    can be compared directly with that of

    the vertebral body. He reported that

    in the absence of spondylolytic my-

    elopathy, the canal depth is equal to

    on greaten than the vertebral body di-

    mension, whereas a canal depth of

    80%-85% that of the vertebral body

    dimension increases the probability

    that spondylolytic myelopathy is

    present. He also reported that if the

    canal depth is only 80% that of the

    vertebral body, spondylolytic my-

    elopathy is probable, and that if the

    canal depth is 50%-75% that of the

    body, spondylolytic myelopathy is a

    near certainty.

    Multiple authors have reported the

    association of neunologic symptoms

    with spinal stenosis (2, 5-30). Kessler

    (14) reported that a diameter of the

    cervical spinal canal less than i4 mm

    is significant and described two types

    of associated symptoms: (a) the sud-

    den onset of myelopathy without me-

    gression and (b) a brief transient epi-

    sode lasting minutes or hours. It was

    0 0 0

    00 0 ‘-4

    0 0 0

    -H-H-H00 C’) U)0 b.000’- 0 0

    .*

    ci

    -H00

    0’0

    (a0

    � �?‘�‘.‘?‘�0

    � uuuu:aC,)

    772 #{149}Radiology September 1987

    �Lfi�� -�a�.oq�I $ tO OO�

    �ooo c�)e4-Lrio.Iri� c�)cicfiO%O�� OO�O

    r��-t�- �O%�e1 SO--� �0000 OO�O

    �H-H�H41 -H�1-H-H�-t’�t�,�e’1 tnmc’�c�4o’�oo C40ttiiri’� t-..e-� .- ,- - ,- 0 0 � 0

    InLflOLfl �ootnc�4 c’ia�a� �0� C�’� ‘-I � � � 0 0u � � � I I I I

    LnLnoo o�t�2�.. rfic’it’iN �,�o- - - � .- 0 0 0 0uN inmt-��. ‘.0-b.

    u� � � � - In ‘.0 �4.#� ei�-c�i -005-.-, �

    .� -H�H41-H -H-H-H44(5 G’ (‘I ‘C LI) in c’ � C’.

    a, � � m ‘.o 0 ‘.0 c’�b:’.o’.oIf� �---- 0000

  • MATERIALS AND METHODS

    Volume 164 Number 3 Radiology #{149}773

    Figure 2. Lateral view of the cervical spine

    of a patient with an episode of transient cer-

    vical spinal neurapraxia. In this patient withcervical spinal stenosis, the spinal canal/

    vertebral body ratio was approximately 0.50.

    his observation that either type of

    myelopathy could occur with in-

    creased physical activity without

    trauma or hyperextension of the

    neck; he attributed intermittent

    symptoms to cord claudication. In

    older people, neunologic symptoms

    have been attributed to thickening of

    the ligamentum flavum on hypentro-

    phic proliferative spurs; in younger

    people, to developmental stenosis.

    Countee and Vijayanathan (11) have

    reported the cases of young patients

    with quadniplegia after trauma on

    falls who had cervical spinal canal di-

    ametens of 14 mm or less. Moiel et al.

    (5) reported the case of a young pa-

    tient with congenital narrowing of

    13 mm at C-4 to C-6. Specific case me-

    ports of quadnipanesis in football

    players have also been reported:

    Grant and Puffer (24) reported the

    case of an 18-year-old man with de-

    velopmental stenosis; Stratford (25)

    reported that of a 23-year-old man

    with a sagittal diameter of 12 mm;

    Funk and Wells (26) reported the

    cases of two players with quadnipane-

    sis, one that occurred after hyperfiex-

    ion and one after hypenextension;

    and Maroon (27) reported the symp-

    toms in a patient with “normal” ma-

    diognaphs.

    To obtain a control population, radio-graphs of the lateral cervical spine ob-tainted in the emergency room were col-lected consecutively over several months.Patients with cervical spinal abnormali-ties, fractures, disclocations, or a historyof previous or current neurologic symp-toms related to the cervical spine wereeliminated. Only radiographs of patientsaged 15-38 years were included. The re-sultant control group consisted of 49 maleand 25 female patients.

    For comparison, nadiographs of the lat-enal cervical spine of 23 of 31 patientswith documented transient cervical spi-nal neurapraxia occurring in athletes

    were examined. These patients are thoseof one of the authors (J.S.T.) or of the Na-tional Football Head and Neck InjuryRegistry and are the same populationused in an extensive clinical and radio-graphic review (1). The patients were allmale and ranged in age from 15 to 32years (mean, 20.2 years). The primary ath-letic involvement was football; 29 pa-tients were involved in football (ninehigh school, 16 university, and four pro-fessional), one was a professional boxer,and one was a university hockey player.

    The radiographic measurements wereperformed at C-3 through C-6 on the rou-tine lateral view of the cervical spine.Cervical spinal measurements were deter-mined with two different methods. Thefirst method, designated for this report asthe “conventional method,” is the onemost often referred to in the literature.With this method, the sagittal spinal Ca-nal diameter is measured from the middleof the posterior surface of the vertebralbody to the nearest point of the come-sponding spinal laminar line (Fig. 1) (2-8). The second method, designated as the

    “ratio method,” compares the sagittal spi-nal canal diameter, measured as just de-scnibed, with the sagittal diameter of thecorresponding vertebral body, measured

    at its midpoint (Fig. 1).Statistical analysis was performed on

    the conventional and ratio values withuse of the Mann-Whitney U test and rela-tive operating characteristic (ROC)curves. The Mann-Whitney nonparamet-nc test does not assume normal distnibu-tion of the data. The conventional and ma-tio measurements of the symptomaticpatients were compared with those of themale control group. Measurements of themale control group were also comparedwith those of the female control group.ROC curves (31) were used to analyzeboth the conventional and the ratio meth-

    ods of evaluating cervical spinal stenosis.An ROC curve is a graphic method of ex-pressing the effectiveness of a decisionfor different values of the decision van-able. The decision variables in this in-stance are the measurement of sagittal ca-nal diameter (conventional methodvalue) and the sagittal spinal canal-verte-bral body ratio (ratio method value). TheROC curves were determined with use ofall the measurements made on the control

    male subjects (195 vertebrae) and thesymptomatic patients (91 vertebrae).

    RESULTS

    Measurements of the sagittal diam-

    eter of the cervical spinal canal of C-3

    through C-6 in the control groups are

    detailed in Table 1 . For the male con-

    trol group, the mean sagittal diame-

    ten was 18.9 mm (13.7-23.5 mm), and

    the mean sagittal ratio was 0.98 (0.69-

    1 .27). For the female control group,

    the mean sagittal diameter was 17.2

    mm (13.3-20.4 mm), and the mean

    sagittal ratio was 1.02 (0.81-1.26). The

    overall mean sagittal diameter for the

    entire control group was 18.2 mm,

    and the mean sagittal ratio was 0.99

    (Fig. 1).

    Review of the radiognaphs avail-able for the 23 symptomatic patients

    demonstrated no obvious radio-

    graphic anomalies in 12 patients

    (subgroup 1); of the 11 patients in

    subgroup 2, four had congenital

    anomalies, two had instability, and

    five had intervertebral disk disease.

    Of the four patients with congenitalanomalies, three patients had failure

    of segmentation at C2-3, C3-4, and

    C2-3 and C3-4, respectively, and onepatient had extensive bony prolifera-

    tion at the anterior aspect of the yen-

    tebral bodies. Measurements of the

    sagittal diameter of the vertebral

    bodies in the latter patient did not

    include the proliferative bony mass.

    The cervical sagittal canal diameter

    measurements of C-3 through C-6 in

    the symptomatic patients are also de-

    tailed in Table 1. In subgroup 1, the

    mean sagittal diameter was 14.0 mm

    (8.5-17.0 mm), and the mean sagittal

    ratio was 0.646 (0.31-0.81). For sub-

    group 2, the mean sagittal diameter

    was 16.6 mm (12.0-23.5 mm), and the

    mean sagittal ratio was 0.745 (0.55-

    1.18). For the overall symptomatic

    patient population, the mean sagittal

    diameter was 15.2 mm, and the mean

    sagittal ratio was 0.69 (Fig. 2).

    With both the conventional and

    the ratio methods, there was statisti-

    cally significant (P < .000i) spinalstenosis at each level from C-3

    through C-6 in each subgroup and in

    the entire symptomatic patient group

    compared with the male control

    group.

    Statistical analysis performed to

    compare the spinal canal sagittal di-

    ameters of male and female control

    subjects as measured with the con-

    ventional method indicated a statisti-

    cal difference (Table 1). It was alsodetermined that the sagittal diameter

    of the canal was proportional to the

  • 1 .0-

    .9 -

    .7 -

    .5 -

    .3 -

    +r�’;.���4”-

    ,‘

    HIT

    RATE

    I �

    .5 .7 .9

    774 #{149}Radiology September 1987

    sagittal diameter of the vertebral

    body and that the vertebral bodies

    enlarge proportionally with the ca-

    nals. No statistical difference was

    identified in the spinal canal/verte-

    bra! body ratios between the male

    and female control groups.

    ROC curves were determined forboth the conventional and the ratio

    values (Fig. 3). The conventional val-

    ue and the ratio value were the deci-

    sion variables. For a given cutoff val-

    ue of a decision variable, there are a

    certain number of correct choices

    whereby a stenotic spine is diag-

    nosed as a stenotic spine; these are

    called hits. Additionally, theme are a

    number of incorrect choices whereby

    a normal spine is diagnosed as a ste-

    notic spine; these are called false

    alarms. The ratio of hits to the num-

    ben of stenotic vertebrae is called the

    hit rate. The ratio of false alarms to

    the number of normal vertebrae is

    called the false-alarm rate. Each

    change in the cutoff value produces a

    new hit rate and an associated false-

    alarm rate. Selection of a series of

    cutoff values yields a set of hit rates

    and false-alarm rates. An ROC curve

    is a plot of these sets of pairs with the

    hit mate plotted on the ordinate axis

    and the false-alarm rate plotted on

    the abscissa. The points on the curves

    were generated by varying the cutoff

    value for the decision variables such

    that each change in the cutoff value

    increased the hit rate, false-alarm

    rate, or both, by the smallest amount

    possible; this produced the maximal

    number of points to plot the curves.

    A perfect decision would have a hitrate of i and a false-alarm rate of 0.

    Therefore, as the curve shifts toward

    the upper left corner of the graph,

    the more sensitive and specific the

    test (32).

    With use of an operating point of a

    sagittal canal diameter of less than 14

    mm to indicate stenosis, the hit rate

    was 0.35 and the false-alarm rate was

    0.01. Hence, 65% of all stenotic canals

    would have been misinterpreted as

    normal, and i% of the normal canals

    would have been interpreted as ste-

    notic. With use as an operating point

    a spinal canal/vertebral body ratio of

    less than 0.82 to indicate stenosis, the

    hit rate was 0.92 and the false-alarm

    rate was 0.06. Hence, 8% of all stenot-

    ic canals would have been misintem-

    preted as normal and 6% of all nor-

    ma! canals would have been inter-

    preted as stenotic. Based on the cu-

    mulative data (195 normal vertebrae,

    91 abnormal vertebrae), a ratio value

    of less than 0.76 represented a stenot-

    ic canal 98.5% of the time, and a ratio

    value of less than 0.80 represented a

    stenotic canal 96.3% of the time.

    DISCUSSION

    Spinal stenosis of the lumbar spine

    is best determined with computed to-

    mogmaphy; in the cervical spine,

    however, lordosis may create a false

    appearance of spinal stenosis, and

    measurement of the sagittal dimen-

    sion of the cervical spine on a con-

    ventional radiograph is more reliable

    (33). Use of the actual measure-

    ment in millimeters on the lateral

    view of the cervical spine to docu-

    ment spinal stenosis, however, is

    misleading, and the dimensions con-

    sidened to be significant as reported

    in the literature vary. Boijsen (3) in-

    vestigated the cause of variation in

    the reported radiographic measure-

    ments of the sagittal diameter of the

    cervical spinal canal and evaluated

    the effects of the following two fac-

    tons: (a) the focus-to-film (i.e., the tan-

    get) distance and (b) the object-to-

    film (i.e., the cervical spine to the

    cassette) distance. He reported that

    the effect of a difference between a 1-

    and 1 .5-m focus-to-film distance on

    the resultant sagittal canal measure-

    ment is 0.5 mm. To evaluate the ef-

    fect of the object-to-film distance, he

    calculated that the average difference

    in shoulder breadth between men

    and women is 10 cm (approximately

    a 5-cm difference in the object-to-

    film distance). The effect of a 5-cmdifference in the object-to-film dis-

    tance on the resultant sagittal canal

    measurement is 1.2 mm at a focus-to-

    film distance of 1 m, and 0.7 mm at a

    focus-to-film distance of 1.5 m.

    In our series, the average sagittal

    dimension of the vertebral body in

    the symptomatic patient group was

    22 mm, compared with 19.2 mm in

    the control group. This increased di-

    mension of the vertebral body may

    be actual, projected, on a combina-

    tion: actual because the symptomatic

    subjects were athletically active,

    well-developed football players who

    may have had increased skeletal size;

    projected, because the patients in the

    symptomatic group may have had an

    increased shoulder girth compared

    with that of the average population,

    which increases the object-to-film

    distance and thereby the magnifica-

    tion. Despite the effect of magnifica-

    tion on the sagittal dimension of the

    vertebral body, however, the average

    sagittal spinal canal dimension was

    14 mm in the symptomatic patients,

    compared with 18.9 mm in the con-

    trol group. Therefore, even with the

    0.0-0.0

    .1 .3

    FALSE ALARM RATE

    Figure 3. ROC curve for all data. Solid lineand + indicate value obtained with ratiomethod. Broken line and 0 indicate valueobtained with conventional method.

    contributing effects of magnification,

    the canal measurement was signifi-

    cantly smaller in the symptomatic pa-

    tient group compared with the con-

    trol group. It would be interesting to

    obtain lateral radiographs of the cer-

    vical spine of asymptomatic high

    school, university, and professional

    football players of the same size and

    weight as our symptomatic patient

    group to determine if the sagittal di-

    mensions of the vertebra! bodies and

    the spinal canal are proportinally en-

    larged compared with those of our

    control group; however, in this liti-

    gious age, radiation exposure of

    young, healthy, asymptomatic adults

    is difficult to justify. The size of the

    spinal cord was not evaluated as pant

    of this study, and it is unknown if

    the spinal cord size increases propon-

    tionally with that of the vertebral

    bodies, although there have been re-

    ports indicating that the sagittal di-

    mension of the spinal cord is rela-

    tively constant from the cervical

    level to the midthonacic level (34-38).

    In our series, statistically signifi-

    cant cervical spinal stenosis was de-

    tected with both the conventional

    and the ratio methods in the symp-

    tomatic patients compared with the

    male control group. However, the ac-

    tual measurement of the sagittal di-

    ameten of the cervical spinal canal

    was within normal limits, that is, 14

    mm or more at two on more levels, in

    2i of the 23 symptomatic patients. By

    comparison, the ratio value was with-

    in normal limits (i.e., greater than

    0.82 at two on more levels) in only

    two of the 23 symptomatic patients,

    both of whom were in subgroup 2

    and one of whom had obvious con-

    genital anomalies. Comparison of the

    conventional and ratio methods for

    determining cervical spinal stenosis

    with use of ROC curves shows that

    the ratio method is more than 21/2

    times as sensitive as the conventional

  • Volume 164 Number 3 Radiology #{149}775

    method with use of a cutoff value of

    0.82. Also, the ratio method is more

    specific than the conventional meth-

    od for determining spinal stenosis.

    At a ratio of 0.82, the ROC curveyields a 92% accuracy with only a 6%

    false-alarm rate. With use of 6% as

    the maximal false-alarm rate, the con-

    ventional method (at 14 mm) yields

    only a 62% accuracy.

    CONCLUSION

    The ratio method of diagnosing

    cervical spinal stenosis is indepen-

    dent of magnification factors caused

    by differences in target distance, ob-

    ject-to-film distance, or body type,

    because the sagittal diameter of the

    spinal canal and that of the vertebral

    body are in the same anatomic plane

    and are similarly affected by magnifi-

    cation. There is normally a one-to-

    one relationship between the sagittal

    diameter of the spinal canal and that

    of the vertebral body, regardless of

    sex. A spinal canal/vertebral body ma-

    tio of less than 0.82 indicates signifi-

    cant cervical spinal stenosis. U

    References1. Torg JS, Pavlov H, Genuario SE, et al.

    Neurapraxia of the cervical spinal cordwith transient quadriplegia. J Bone JointSung [Am] 1986; 68A:1354-1370.

    2. Wilkinson HA, LeMay ML, Ferris EJ.

    Roentgenographic correlation in cervicalspondylolysis. AJR 1969; 105:370-374.

    3. Boijsen E. Cervical spinal canal in intra-spinal expansive processes. Acta Radiol1954; 42:101-115.

    4. Hinck VC, Hopkins CE, Savara BS. Sagit-tal diameter of the cervical spinal canal inchildren. Radiology 1962; 79:97-i08.

    5. Moiel RH, Raso E, Waltz TA. Centralcord syndrome resulting from congenitalnarrowing of the cervical spinal canal.Trauma i970; iO:502-5i0.

    6. Payne EE, Spillane JD. The cervicalspine: an anatomico-pathological study of

    70 specimens (using special technique)

    with particular reference to problems ofcervical spondylolysis. Brain 1957; 80:571-596.

    7. Wells CE, Spillane JD, Bligh AS. The cer-vical spinal canal in syningomyelia. Brain1959; 82:23-40.

    8. Wolf BS, Khilnani M, Malis L. Sagittal di-ameter of the bony cervical canal and itssignificance in cervical spondylosis.Mount Sinai Hosp 1956; 23:283-292.

    9. Epstein JA, Carras R, Epstein VS. LavineLS. Myelopathy in cervical spondylolysis

    with vertebral subluxation and hyperlor-dosis. J Neurosurg 1970; 32:42i.

    10. Epstein VS. Epstein JA, Jones MD. Cervi-cal spinal stenosis. Radiol Clin North Am

    1977; 15:215-226.1 1. Countee RW, Vijayanathan T. Congenital

    stenosis of the cervical spine: diagnosisand management. J Nat! Med Assoc 1979;71:257-264.

    12. Chrispin AR, Lees F. The spinal canal incervical spondylosis. J Neurol NeurosurgPsychiatry 1963; 26:i66-170.

    13. Ehni C. Cervical arthrosis: diseases ofcervical motion segments (spondylosis,disk rupture, radiculopathy, and myelopa-thy). Chicago: Year Book Medical, 1984;26-43.

    14. Kessler JT. Congenital narrowing of thecervical spinal canal. J Neurol NeurosurgPsychiatry 1975; 38:1218-1224.

    15. Alexander E, Davis CH, Field CH. Hy-perextension injuries of the cervical spine.Arch Neurol Psychiatry 1958; 79:146-150.

    16. Brain L, Wilkinson M, eds. Cervicalspondylolysis and other disorders of thecervical spine. Philadelphia: Saunders,1967.

    17. Burrows EH. The sagittal diameter of thespinal canal in cervical spondylolysis.

    Clin Radiol 1963; 17:77-86.18. Epstein NE, Epstein JA, Zilkha A. Trau-

    matic myelopathy in a seventeen-year-oldchild with cervical spinal stenosis (with-out fracture or dislocation) and a C2-C3Klippel-Feil fusion: a case report. Spine1984; 4:344-347.

    19. Hinck VC, Sachdev NS. Developmentstenosis of the cervical spinal canal. Brain1966; 89:27-36.

    20. Nugent GR. Clinico-pathologic correla-tions in cervical spondylo!ysis. Neurology1959; 9:273-281.

    21. Schneider RC, Cherry GR, Pantek H.Syndrome of acute central cervical spinalcord injury with special reference tomechanisms involved in hyperextensioninjuries of cervical spine. J Neurosurg1954; 11:546-577.

    22. Stoltmann HF, Blackward W. The role ofthe ligamentum flavum in the pathogene-sis of myelopathy in cervical spondyloly-sis. Brain 1974; 87:45.

    23. Taylor AR. The mechanism of injury tospinal cord in the neck without damage tothe vertebral column. J Bone Joint Surg1951; 33B:543-546.

    24. Grant TT, Puffer J. Cervical stenosis: adevelopmental anomaly with quadnipare-sis during football. Am J Sports Med 1976;4:219-221.

    25. Stratford J. Congenital cervical stenosis:a factor in myelopathy. Acta Neurochir1978; 41:101-106.

    26. Funk FJ, Wells RE. Injuries of the cervicalspine in football. Clin Orthop Re! Res1975; 109:50-58.

    27. Maroon JC. “Burning hands” in footballspinal cord injuries. JAMA 1977; 238:2049-2051.

    28. Pallis C, Jones AM, Spillane JD. Cervicalspondylosis: incidence and implications.Brain 1954; 77:274-289.

    29. Payne EE. The cervical spine and spon-dy!o!ysis. Neurochirurgia 1959; 1:178-196.

    30. Staltman HF, Blackward W. The role ofthe !igamentum flavum in the pathogene-sis of myelopathy in cervical spondyloly-sis. Brain 1974; 87:45-50.

    3i. Metz CE. ROC methodology in radiolog-ic imaging. Invest Radio! 1986; 21:720-

    733.32. Sheridan TB, Farrell WB. Man-machine

    systems: information, control and deci-sion models of human performance. Cam-bridge, Mass.: MIT Press, 1981; 355-382.

    33. Di Chiro C, Fisher RL. Contrast radiogra-phy of the spinal cord. Arch Neurol 1964;11:125-143.

    34. Lowman RM, Finkelstein A. Air myelog-raphy for demonstration of cervical spinalcord. Radiology 1942; 39:700-706.

    35. Burrows EH. Sagittal diameter of the spi-nal canal and cervical spine stenosis. ClinRadio! 1963; 14:77-86.

    36. Kehilnani MT. Wolf BS. Transverse di-ameter of cervical spinal cord on pantopa-

    que myelography. J Neurosurg 1963;20:660-664.

    37. Paul LW, Chandler A. Myelography inexpanding lesions of the cervical spinalcord. Presented at the 45th Scientific As-

    sembly and Annual Meeting of the Radio-logical Society of North America, Chica-

    go, November 15, 1959.38. Gonsalves CC, Hudson AR, Horsey WJ,

    Tucker WS. Computed tomography of

    the cervical spine and spinal cord. Com-put Tomogr 1978; 2:279-293.