A total of 65 individuals between the ages of 18 and 42 participated in this study. The asymptomatic group was comprised of 20 healthy individuals, with no history of low back pain lasting more than 3 days or pain reported within the previous year (Table 1). The symptomatic subjects were recruited from a very busy University based general practice. Forty-five patients with non-specific central low back pain that worsened with extension, constituted the symptomatic group (Table 1). Based on the appearance of the fifth lumbar vertebrae, suggesting developmental variance one subject was excluded from further participation. An additional symptomatic subject was recruited to meet the expected subject population in this study. This analysis is a sub-protocol of a larger study thus the discrepancies in the size of the two study populations. On the day of testing, these subjects had perceived pain with standing spinal extension averaging 4.3 ± 1.9 on the Visual Analog Scale and their Oswestry Disability Scores range between minimal and moderate. The reported average duration of symptoms of the current episode of low back pain was 28 days. None of them were off work. Patients for whom the current episode of back pain lasted longer than 12 weeks were not included in this study.

The primary exclusion criteria for both groups were: 1) spinal malignancy, 2) cardiovascular disease, 3) evidence of cord compression, 4) aortic aneurysm, 5) hiatal hernia, 6) uncontrolled hypertension, 7) spinal infection, 8) severe respiratory disease, 9) pregnancy, 10) abdominal hernia, 11) prior low back surgery, 12) gross spinal deformity, 13) spondylolisthesis, 14) known rheumatic joint disease, and 15) implanted biological devices that could interact with the magnetic field (i.e., pacemakers, cochlear implants, or ferromagnetic cerebral aneurysm clips). In addition to the above exclusion criteria, subjects in either group could not have any signs or symptoms related to nerve root or cord pathology. Therefore, subjects who demonstrated any of the following also were excluded: 1) radiating pain below the level of the buttock(s), 2) sensation changes in the lower extremities, 3) diminished reflexes, 4) lower extremity weakness, 5) urinary or fecal incontinence, and 6) increased peripheral pain with repeated lumbar extension.

As described in a previous publication, 27 dynamic imaging of the lumbar spine was performed using a vertically opened (double donut design) MRI system (0.5 Tesla, Signa SP, General Electric Medical Systems, Milwaukee, WI) with a 56 cm opening that allowed the examiner access to the subject during testing. This system was equipped with a pulse sequence programming environment and real-time interactive MRI capability.

Midline sagittal plane imaging of the spine was performed using a receive-only surface coil and an ultrafast spoiled GRASS (gradient recalled acquisition in the steady state) pulse sequence. Images were obtained at a rate of 1 per second using the following parameters: repetition time (TR): 200 ms; echo time (TE): 18 ms; number of excitations: 1.0; matrix: 256 × 256; field of view (FOV): 28 × 21 cm; and a 7 mm section thickness with an interslice spacing of 1 mm. 27 The surface coil was flexible and designed to expose the low back such that the examiner had direct access to palpate the lumbar spinous processes.

Prior to participation, informed consent was obtained as approved by the Institutional Review Boards of the University of Southern California and Stanford University. Two sets of images were acquired for each subject: 1) during the manual PA mobilization procedure and 2) during a PU maneuver. The order of testing was randomized for each subject.

For the PA assessments, subjects were placed in the prone position on a sliding table within the MRI system, such that the spine and torso were within the opening between the vertical magnets (Figure 1a). For each of the two procedures, the surface coil was secured to the lumbar region using cloth tape. Careful care was taken to place the coil in the same region of the subject's low back. A small pillow was positioned under the subject's abdomen, thereby mimicking the clinical procedure.

To gain an insight into the posture of the lumbar spine in the starting prone position, we measured the angular position of the lumbar curvature with an inclinometer (Universal Inclinometer, OPTP, Minneapolis, MN). To do so, we placed an inclinometer at the following anatomical locations S2, L3 and L1.

After subjects were positioned within the MRI system, a series of sagittal plane "localizers" were obtained to ensure that the image plane captured the vertebral bodies and spinous processes of all lumbar vertebrae. Once the image plane was determined, continuous imaging (1 Hz) commenced.

Following a static sagittal view of the lumbar spine in the resting position, a PA force was applied to the spinous processes at each vertebrae starting caudally at L5 and moving cranially to L1. The force was directed anteriorly in a direction perpendicular to the tangent of the arc of the lumbar lordosis. Sufficient force was applied to induce movement to the end of the available range of segmental motion and was comparable in magnitude to that of a "grade IV" as defined by Maitland. 16 The force was applied slowly (over approximately a 1–2 second period) and held at end range for at least five seconds. Release of the force also occurred slowly (1–2 seconds) before applying force to the next vertebral level. Once the examiner released the force on L1, and a clear resting position was observed, imaging was terminated.

All PA mobilizations were performed by a physical therapist with 15 years of manual therapy experience. A second investigator viewed the images on the MRI console to assure that the forces were being applied to the correct vertebral level. If an error was observed, imaging was repeated (this has occurred in three instances).

For the PU assessment, subjects were placed in the prone position on the sliding table, such that the spine and torso were within the opening between the vertical magnets and the head and arms were outside of the MRI system (Figure 1b). For this procedure, the patient's positioning within the magnet was perpendicular to the position during the PA mobilization testing. The surface coil was again secured to the lumbar region using cloth tape. A small pillow was positioned under the subject's abdomen for comfort. Following the subject's new positioning within the MR system, the sagittal plane "localizers" were repeated to ensure that the image plane captured the vertebral bodies and spinous processes of all lumbar vertebrae. Once the image plane was determined, continuous imaging (1 Hz) commenced. After the acquisition of a static sagittal view of the lumbar spine in the resting position, subjects placed their hands in front of them and arched their back as far as possible within pain tolerance (Figure 1b). Subjects were instructed to hold the end press-up position for a count of 10 and slowly lower the trunk to the resting position. Upon returning to the prone position, imaging was terminated.

Prior to analysis, all images were transferred from the MRI system console to a Macintosh G3 computer (Apple Computer, Cupertino, CA). For purposes of this study, only the images at rest and at the end-range of segmental motion were analyzed. All data sets were again screened for quality of images and for pathology. Consequently, all images were deemed appropriate for further analysis. During screening, we noticed motion artifacts in the images where motion was expected. This helped in choosing the images for analysis. As the beginning of motion was readily observable, we chose the resting image for analysis at lest 5 images before motion was observed. The end of motion image was also chosen based clarity of the images, suggesting no motion. To assure ourselves that this assumption was correct, we have randomly selected 5 studies and have digitized three consecutive images deemed to be at the end of the range. This experiment assured us that our prior choice of the final image was correct.

Lumbar segmental motion in the sagittal plane was quantified by measuring the intervertebral angle which was defined as the angle formed by lines delineating adjacent vertebral endplates (Figure 2) 27. Segmental lumbar motion was defined as the difference between the intervertebral angles measured from the resting and the end-range images. An increase in intervertebral motion from the resting to the end-range position was indicative of spinal extension. Conversely a decrease in the intervertebral motion was indicative of spinal flexion. The superior vertebra was used to define the target functional spinal unit. For example, a PA force applied to L4 was identifying the function spinal unit of L4-L5.

The same investigator, blinded to the group assignment, made all measurements using the National Institutes of Health (NIH) software (Bethesda, MD). To establish the intra-tester reliability of angular measurements, dynamic MR images obtained from 5 healthy volunteers were measured on two separate occasions. Intraclass Correlation Coefficients (ICC) were found to be excellent, ranging from 0.95 to .99 for all subjects. The Standard Error of Measurement was determined to be between 0.4 and 0.66 degrees.

For the PA analyses only the data from the motion-segment associated with the PA was used for the analysis. The data for both groups was screened for normality of distribution using Wilks-Shapiro W statistics. The symptomatic group did not exhibit normal distribution; therefore non-parametric statistics were used for further analyses.

To determine the distribution of subjects outside of the normal range of segmental mobility, the norms were established based on the asymptomatic group's mean value ± 2 SD (see Tables 2 and 3). Individuals demonstrating segmental motion greater than 2 standard deviations range will be considered hypermobile while those demonstrating less than 2 standard deviations range were considered to be hypomobile. Overall frequency count will determine number of segments outside of normal range in each group of subjects. A Chi-square analyses of data at each motion- segment, comparing groups (control, symptomatic) and mobility (normal, non-normal) will be performed using SPSS statistical software (SPSS Inc., Chicago, IL) with a significance level of p < 0.05.