Adjustments and Disc Pain?

The Within-Session Change in Low Back Pain Intensity Following Spinal Manipulative Therapy Is Related to Differences in Diffusion of Water in the Intervertebral Discs of the Upper Lumbar Spine and L5-S1


Study Design

Single-group, prospective, repeated-measures design with responder analysis.


To determine differences in the changes in diffusion of water within the lumbar intervertebral discs between participants with low back pain who reported a within-session reduction in pain intensity following a single treatment of spinal manipulative therapy and those who did not.


There is a paucity of research that describes the physiologic events associated with analgesia following intervention for low back pain. Postintervention increases in the diffusion of water within various soft tissues of the spine may be one of many potential mechanisms linked to pain reduction.


Nineteen adults between 20 and 45 years of age participated in this study. All participants reported low back pain of at least 2 on an 11-point (0–10) verbally administered numeric pain rating scale at the time of enrollment. Participants underwent T2- and diffusion-weighted lumbar magnetic resonance imaging scans immediately before and after receiving a single treatment of spinal manipulative therapy. Individuals who reported a decrease in current pain intensity of more than 2 following treatment were classified as “within-session responders,” and the remainder were classified as “not-within-session responders.” The apparent diffusion coefficient (ADC), representing the diffusion of water in the nucleus pulposus, was calculated from ADC maps derived from the midsagittal diffusion-weighted images.


Two-way, repeated-measures analyses of variance indicated significant group-by-time interactions. Participants in the within-session-responder group (n = 12) had a postintervention increase in ADC at L1-2 (P = .001), L2-3 (P = .002), and L5-S1 (P = .01) compared to those in the not-within-session-responder group (n = 7). Large effect sizes in ADC between responder groups were observed at L1-2 (d = 1.74), L2-3 (d = 1.83), and L5-S1 (d = 1.49). No significant group-by-time interactions were observed at the L3-4 and L4-5 levels.


Changes in the diffusion of water within the lumbar intervertebral discs at the L1-2, L2-3, and L5-S1 levels appear to be related to differences in within-session pain reports following a single treatment of spinal manipulative therapy. J Orthop Sports Phys Ther 2014;44(1):19–29. Epub 21 November 2013. doi:10.2519/jospt.2014.4967

Patients with low back pain (LBP) represent the cohort most commonly treated by physical therapists.44 Although recent refinements in treatment approaches are encouraging,15,17,30,34,63 LBP remains the most frequent cause of lost work time and disability among working-age adults in industrial countries.13,68 Unfortunately, the overall economic and societal impact of LBP is not improving and appears to be worsening.32,66 Numerous factors within the biopsychosocial model likely contribute to the difficulty of current treatment approaches to prevent the poor recovery and growing disability rates that affect a substantial subset of people with LBP.23,32,46,6466 One of the primary barriers to developing more effective interventions is a lack of knowledge regarding the physiologic mechanisms by which LBP is propagated and sustained.6,12,14,20,40,53 Specifically, there is a paucity of research that describes the physiologic events associated with analgesia following intervention for people with LBP; that is, it is unclear why some patients report symptom reduction following a given treatment and other clinically similar patients do not. This inability to clearly determine the mechanism by which a treatment yields a favorable outcome makes it impossible to provide a basis on which to develop new treatments and to refine current treatments.49

Recently, advances in brain and spine imaging have begun to yield encouraging findings of a number of central and peripheral mechanisms thought to be important components of the generation and propagation of LBP.12,20,23,41,5153,65 Among these hypothesized mechanisms is an increase in diffusion (rate of movement) of water within various soft tissues of the spine, occurring in response to treatment, that may be linked to pain reduction.79,24 Relative to LBP, the lumbar intervertebral disc (IVD) is a key soft tissue structure in which this phenomenon may occur.46,8,14,42,51,53,58 For example, histochemical and biomechanical analyses performed on animal models have suggested that increased diffusion of water within the lumbar IVD is a favorable event that may (1) enhance gas and nutrient transport,3,28,60 (2) aid in the removal of metabolic waste products that may be associated with pain,3,28,42 and (3) have a positive effect on internal and/or external pressure gradients acting on the disc.3,57 Establishing the extent to which increased diffusion within the IVD occurs in vivo in people who receive treatment for LBP and describing the extent to which this event may be associated with a reduction in reported pain intensity would provide important information regarding the mechanisms by which interventions influence symptoms.

A new application of lumbar magnetic resonance imaging (MRI) known as diffusion-weighted MRI allows diffusion of water within the IVD to be quantified by providing an estimate of the rate at which water moves within preselected tissue slices.5 This event is represented by the apparent diffusion coefficient (ADC).4,5,52 The ADC is obtained by averaging the signal intensity from several diffusion-weighted images of the same tissue slice obtained over time to generate an “ADC map,” from which estimates of the ADC may be calculated by specialized software (FIGURE 1).11 The authors of the present study have conducted a series of studies examining the relationship of the ADC within the nuclear region of the lumbar IVDs to intervention and pain reports.8,9,11 In 2 initial studies,9,11 we developed a procedure to obtain measures of the ADC from the nuclear region in the lumbar IVDs that have an acceptable balance of diffusion weighting and signal intensity, while providing excellent reliability of the measurements (FIGURE 1). As a result, we observed that the ADC of the L5-S1 IVD was significantly increased following a 10-minute application of posterior-to-anterior (PA) manual pressures applied to the lumbar spine of people with a prior history of LBP.9 Furthermore, when the same individuals had lain prone for 10 minutes during a separate session, this finding was not present. Based on this finding, we concluded that PA pressures may generate a stimulus that results in a rapid measurable increase in diffusion of water within the nuclear region of the L5-S1 IVD, and that changes in ADC were not simply due to prolonged recumbency. Many of the study participants were not symptomatic at the time of testing, thus the linkage between this finding and pain was unknown.

FIGURE 1. (A) Measurement of the ADC obtained from an ADC map that was calculated from midsagittal images (b-factor, 400 s/mm2). The circular area (arrow) represents the region of the L4 intervertebral disc from which the ADC was sampled. (B) Midsagittal, T2-weighted image that depicts the same anatomic “slice.” Abbreviation: ADC, apparent diffusion coefficient.

In a subsequent study8 of individuals who currently had LBP, we compared differences in changes in the ADC within the L5-S1 IVD between 10 participants who had a within-session positive response (reduction of pain of 2/10 or greater) and 10 participants who did not have a within-session positive response (reduction of pain of less than 2/10) following a single application of PA pressure and prone press-up exercises.8 Our results indicated that those participants with a within-session positive response had a significant increase in ADC, whereas those who did not have a within-session positive response demonstrated a decrease in the ADC at this level.

These findings from our previous research suggest that postintervention changes in the ADC of lumbar IVDs may be linked to patient reports of LBP. This has led to our central hypothesis, which is that individuals who have a reduction of pain intensity following intervention will also have increases in the rate at which water travels within the lumbar IVDs. Conversely, we believe that those patients who do not have a reduction in pain intensity following intervention will have no change in, or a reduction of, the rate at which water travels within the lumbar IVDs. To further test our hypotheses, the current study examined spinal manipulative therapy (SMT). We chose to investigate this treatment because SMT is frequently used by physical therapists and others to treat patients with LBP, it has a low risk of adverse events, and some (but not all) patients are likely to have favorable within-session responses to this treatment.12,16,25,31,36,37,39,45,47,62 SMT is performed by applying 1 or more short-amplitude, high-velocity thrusts at various angles to the spine and/or pelvis of a prepositioned patient.47 It is hypothesized, but not definitely proven, that these thrusts create a stimulus to spinal tissues that may result in analgesia.12,20,22,50 A large body of research has generally supported efficacy and effectiveness of SMT for the treatment of LBP, with overall effect-size improvements in pain following SMT that are typically modest when compared to placebo treatment or to no treatment at all.16,19,33,3638,42,55,59,62 These findings suggest a heterogeneous response between patients receiving SMT. One cause for this variation may be explained by differences in water diffusion within the lumbar IVD. There are currently no data that describe changes in diffusion within this structure that are associated with SMT.

The purpose of the current study was to determine differences in the changes of diffusion of water in the 5 lumbar IVDs between individuals with nonspecific LBP who reported a within-session reduction in pain intensity and those who did not report a within-session reduction in pain following a single treatment of SMT. An observation that only the within-session responders to SMT demonstrated increased diffusion would support the hypothesis that the postintervention change in the diffusion of water within the IVD is one mechanism by which manual therapy influences pain.



Study participants were recruited from the local community and were eligible for enrollment if they were between 20 and 50 years of age and reported an LBP intensity of at least 2/10 on the 11-point (0–10) numeric pain rating scale at the time of testing.26,43 A reported pain intensity of 2/10 was used as the lower bound for entry criteria because this value is equal to the minimal detectable change on an 11-point numeric pain rating scale and would prevent a floor effect when using this scale for postintervention classification.26 Potential participants were excluded if they had any contraindications to undergoing MRI and/or SMT.47 In addition, individuals were excluded for signs of nerve root compression, visual evidence of a lateral shift of the spine, possible pregnancy, or a history of inflammatory joint disease, osteoporosis, discitis, or neoplastic disorders of the spine. Additional exclusion criteria included a history of invasive procedure to the lumbar spine or evidence of any of the following abnormalities visible on T2-weighted imaging: lumbar disc extrusion, severe nerve compression,10 spondylolisthesis of greater than 4 mm, or sacralization of a lumbar vertebra. Prior to enrollment in the study, participants underwent standard safety screening for MRI and provided written informed consent as approved by the Institutional Review Board at the University of South Carolina. The procedure used in this study is illustrated in FIGURE 2.

FIGURE 2. Flow chart of participants through the study. Abbreviation: MRI, magnetic resonance imaging.

Intake Measures and Patient Classification

Following screening and informed consent, participants completed a pain diagram, the Roland-Morris Disability Questionnaire,61 and a questionnaire that sampled the effects of sitting, walking, standing, and bending on their current symptoms. At this time, participants provided a verbal estimate of their pretreatment current pain intensity using the 11-point numeric rating scale, with 0 as no pain and 10 as the worst pain imaginable. Participants then underwent a physical examination performed by the first author (P.F.B.) in a room adjacent to the scanner. This examination began with visual assessment of standing posture to exclude individuals who presented with a lateral shift. This was followed by a visual assessment of 1 repetition of active lumbar flexion and lumbar extension. Participants were then positioned supine, where active range of motion of hip flexion, hip internal and external rotation, and a straight leg raise were determined by visual assessment. Passive overpressure was applied at the end of each of these motions. Participants who had distal lower extremity pain during the passive straight leg raise at less than 45° were excluded. Following this, participants were positioned in prone, and 2 passive PA-directed pressures were applied to the spinous processes of the L1 through L5 vertebrae to the end of the available range of motion, as perceived by the examiner. The examiner classified the mobility of each motion segment as normal, hypermobile, or hypomobile.35 Participants who were classified as hypermobile at any segment or who reported a peripheralization of symptoms during these procedures were excluded. The remainder of the participants were classified as candidates for SMT and were enrolled in the study.

Imaging Procedure

Participants were imaged in supine, with their hips and knees maintained at 30° of flexion by a bolster positioned under the knees. Brief “scout series” including sagittal, axial, and coronal images were obtained prior to each scanning session to ensure consistent positioning of the participant within the scanner. Spin-echo techniques, using multielement spine coils, were used to obtain T2-weighted sagittal views. These images were used to assist in ruling out contraindications for SMT and to classify the L1-2 to L5-S1 IVDs based on the presence or absence of degeneration.54 Immediately following this procedure, participants underwent a diffusion-weighted MRI scan using a single-shot, dual spin-echo, echo planar imaging acquisition with multi-element spine coils and abdominal coils. Automatic shimming was used for all image acquisition. Based on previous work,7,9,11 we used a diffusion-weighting b-factor of 400 s/mm2 as the best combination of diffusion weighting and signal intensity. The specific parameters used for imaging are listed in TABLE 1. Images were obtained using a MAGNETOM Trio (Siemens AG, Munich, Germany) 3.0-T MRI scanner at The McCausland Center for Brain Imaging, Palmetto Health Richland Heart Hospital, University of South Carolina, Columbia, SC.

Spinal Manipulative Therapy

Upon completion of the pretreatment scan, the second author (R.B.) underwent an MRI safety screen and entered the scan room, where he removed the participant from the scanner and assisted the participant to roll into the sidelying position. This author, a certified manual therapist, was blinded to all subject information. The therapist performed the intervention by placing the participant in the left sidelying position, with the MRI table adjusted to the height of the therapist’s knees. This allowed the therapist to provide the typical clinical positioning for SMT. The therapist passively flexed the participant’s right hip and knee to approximately 90°, then placed the participant’s right foot over the popliteal fossa of the left knee. To achieve upper-trunk flexion and rotation, the therapist then gently pulled the participant’s left arm in an anterior and caudal direction. The participant’s left shoulder was then abducted and placed under a pillow for support of the participant’s head. The therapist then applied slight anterior and cephalad-directed pressure, with the therapist’s right hand contacting the participant’s right shoulder. From a diagonal stance, the therapist transferred his weight from his posterior (left) leg to his anterior (right) leg, before rolling the participant under him and bringing the participant’s superiormost (right) thigh in contact with the therapist’s left thigh. The therapist’s left hand was then positioned 2 finger breadths from the L5-S1 interspinous space, with his fingers pointing in the cephalad direction. The target contact for the therapist’s hypothenar eminence was the superior articulating process of S1. A high-velocity, short-amplitude thrust was then applied by the therapist, who used the forces applied by his left and right hand, respectively, that were generated by flexing his front (right leg) and leaning back onto the rear (left) heel (FIGURE 3). Following this procedure, the participant was asked to roll from the left sidelying position back to the right sidelying position, where the procedure was repeated. Upon completion of SMT, the participant was returned to the supine position and re-entered into the MRI scanner for the repeat scans. The time from the completion of SMT to the start of acquisition of the postintervention diffusion-weighted images was approximately 5 minutes. Upon completion of the repeat scans, the participant left the scan room and reported a posttreatment estimate of current pain intensity on the 11-point numeric pain rating scale, which was collected by the first author (P.F.B.). The time between this report and the completion of SMT was approximately 30 minutes.

FIGURE 3. Subject and therapist positioning for the performance of the spinal manipulative therapy on the magnetic resonance imaging table.

Evaluation of Images

Classification of T2-Weighted Signal A modification of the rating scale developed by Pfirrmann et al54 was used to identify the presence and extent of IVD degeneration, based on the intensity (brightness) and homogeneity of the T2 signal in the nuclear region. The criterion for normal IVD is the appearance of a homogeneous, bright-white nucleus, with a clear distinction between the annulus and nucleus, and that of degenerative IVD is the appearance of a nonhomogeneous and gray or black nucleus. Each of the T2-weighted, midsagittal images obtained during the initial scanning of all participants was evaluated independently by 2 of the authors (P.F.B. and D.M.L.) to classify the L5-S1 IVD as normal or degenerative. Consensus between the 2 examiners was used to address any disagreements in classification.

Determination of ADC Values Maps of the mean ADC were calculated by the main computing system with an imaging-analysis program known as USCLEO (Siemens AG). After the images were obtained, the coded files were saved and transferred to a remote work station for analysis. The midsagittal ADC maps were used to obtain measures of the ADC from the central nuclear region of L1-2 to L5-S1 IVDs for all scans. This image slice has been shown to provide reliable measures and allowed direct comparison to our previous studies. Prior to analysis, the ADC maps were compared to the diffusion-weighted images to rule out the presence of the “T2 shine effect,” which is a false ADC value that may occur in aging IVDs due to elevated T2 decay time rather than diffusion.56

The ADC values were calculated using standard software available on the work station that assessed signal intensity within the pixels selected by examiners using a circular region of interest (FIGURE 1). Care was taken to restrict the region of interest to the exact center of each IVD and to avoid the “partial-volume effect,” that is, the heterogeneity of tissue that may occur when the vertebral bodies or end plates are included in the region of interest. Measures obtained using this technique have been shown to be reliable, with intraclass correlation coefficients ranging from 0.95 to 0.99 and the standard error of measurement ranging from 0.006 to 0.026 × 10−3 mm2/s (0.1%–5.5%).9

To reduce measurement bias, the fourth author (D.M.L.) obtained all measures of ADC, while blinded to all participant information (participant code, date, test condition, and T2 findings). Consistency of slice location between preintervention and postintervention images was ensured by careful participant placement within the scanner and by making sure that the evaluated image represented the true midsagittal slice of the lumbar spine by including the spinous processes of all 5 lumbar vertebrae.

Classification of Within-Session and Not-Within-Session Responders

Participants whose posttreatment pain intensity showed a pain reduction of at least 2/10 when subtracted from the pretreatment pain intensity were classified as “within-session responders”; all others were classified as “not-within-session responders.” The change score of at least 2/10 was chosen because previous research has suggested that this represents a likely minimal detectable change on this scale.26,43 Considering this, all participants enrolled in the study had a preintervention pain intensity measure equal to or exceeding the minimal detectable change on this scale.

Data Analysis

Characteristics of Within-Session Compared to Not-Within-Session Responders Differences between the characteristics of the participants classified as within-session responders and those classified as not-within-session responders were assessed using an independent t test for continuous variables (age; body mass index [BMI]; Roland-Morris Disability Questionnaire score; pretreatment pain intensity; as well as average, high, and low pain on a typical day). A Pearson chi-square test was used to determine between-group differences for frequencies of categorical variables (duration of current symptoms, history of prior back problems, anatomic locations of symptoms, and the presence or absence of decreased T2 signal at 1 or more IVDs between L1-2 and L5-S1).

ADC Values of Within-Session Compared to Not-Within-Session Responders Pretreatment and posttreatment ADC values of the nuclear regions in the L1-2 to L5-S1 IVDs for participants in both groups were summarized and evaluated after their distributions were tested for assumptions of normality using the Shapiro-Wilk test. The presence of significant differences in the ADC values within the central nuclear portion of the IVD as a function of group assignment (within-session and not-within-session responders) over time was investigated using a general-linear-model, repeated-measures, 2-by-2 analysis of variance. The preintervention and postintervention ADCs at each level were entered as the within-subject factor and the group assignment (within-session responder or not-within-session responder) was entered as the between-subject factor. This statistical approach allowed us to examine the main effect of treatment and the presence or absence of significant interaction between group assignment and the preintervention-to-postintervention change in ADC. Separate analyses were performed for each spinal level from L1-2 to L5-S1. To control for the potential of experimentwise error that may result from the application of multiple tests, we arbitrarily chose an alpha value of .01 (.05/5).

To provide an estimate of the strength of the differences in the within- and between-group comparisons in ADC values, we calculated effect size using Cohen d: [mean ADC postintervention – mean ADC preintervention]/pooled standard deviation. The magnitude of effect size, as calculated with this equation, was classified as follows: 0.2 to 0.5, small; greater than 0.5 to 0.8, medium; and greater than 0.8, large.48 All analyses were performed with SPSS Version 20.0 (SPSS Inc, Chicago, IL).



A total of 19 participants were enrolled in this study between January 2012 and March 2013. Thirteen were women and 6 were men. At the time of the study, all participants were working full-time or were full-time students. Twelve participants (3 men, 9 women) had a reduction in pain intensity of at least 2/10 following treatment and were classified as within-session responders. The remaining 7 participants (3 men, 4 women) were classified as not-within-session responders. Participants in the within-session-responder group had a mean ± SD BMI of 21.0 ± 2.3 kg/m2, whereas not-within-session responders had a significantly higher BMI of 23.9 ± 2.6 kg/m2 (P = .02) (TABLE 2).

Between-group variation in the frequencies of T2 signals was observed in the IVDs prior to treatment. Those participants classified as within-session responders had a greater than expected frequency of normal-appearing IVDs, whereas those participants classified as not-within-session responders had a greater frequency of at least 1 IVD with a decreased T2 signal (χ2 = 5.4, df = 1, P = .02). There were no other pretreatment differences in self-report measures or examination findings between these groups (TABLE 3).

Following treatment, a significant increase in the mean ADC was observed in the within-session-responder group at the L1-2 level (mean increase in ADC, 0.10 × 10−3 mm2/s; 95% confidence interval: 0.04, 0.16 × 10−3 mm2/s; effect size, d = 0.41; P = .004). The mean ADCs for each of the 4 remaining lumbar segments (L2-3 to L5-S1) in the within-session-responder group were higher following treatment; however, the differences in those values were not significant at the .01 level. In the not-within-session-responder group, the mean ADC following intervention was lower at all lumbar levels except L3-4; however, the differences in those values were not significant at the .01 level.

Between-Group Comparisons of ADC Values

Two-way, repeated-measures analyses of variance indicated significant group-by-time interactions, with participants in the within-session-responder group having a postintervention increase in the ADC at L1-2 (F = 16.36, df = 17, P = .001), L2-3 (F = 13.91, df = 15, P = .002), and L5-S1 (F = 7.82, df = 15, P = .01) compared to participants in the not-within-session-responder group. The point estimates of the percentage of change in the ADC indicated that at the L1-2 level the mean value of the ADC for the within-session-responder group increased by 5.9%, whereas the mean value of the ADC for the not-within-session-responder group decreased by 7.0%. At the L2-3 level, the mean value of the ADC for the within-session-responder group increased by 4.3%, whereas the mean value of the ADC for the not-within-session-responder group decreased by 6.7%. At the L5-S1 level, the mean value of the ADC for the within-session-responder group increased by 7.3%, whereas the mean value of the ADC for the not-within-session-responder group decreased by 3.4% (FIGURES 4 and 5). These values exceed our previously described degree of measurement error in obtaining the ADC. Using the classification described by Cohen, large effect sizes were observed at L1-2 (1.74), L2-3 (1.83), and L5-S1 (1.49). No significant group-by-time interactions were observed at the L3-4 and L4-5 levels. Moderate effect sizes were observed at L3-4 (0.70) and L4-5 (0.67) (TABLE 4).

FIGURE 4. Midsagittal apparent diffusion coefficient maps calculated from images obtained from a 27-year-old woman with recurrent low back pain who was classified a within-session responder. (A) Pretreatment image and (B) posttreatment image. Following treatment, this participant’s apparent diffusion coefficient values increased by 8.2% at L1-2, 10.5% at L2-3, and 23.8% at L5-S1.
FIGURE 5. Midsagittal apparent diffusion coefficient maps calculated from images obtained from a 22-year-old woman who was classified a not-within-session responder. (A) Pretreatment image and (B) posttreatment image. Following treatment, this participant’s apparent diffusion coefficient values decreased by 15.7% at L1-2, 7.7% at L2-3, and 0.01% at L5-S1.


The findings from this study suggest that differences in an individual’s change in reported pain intensity following a single treatment of SMT are related, in part, to changes in the rate of diffusion of water (ADC) in the IVDs of the upper lumbar spine and the lumbosacral junction. The results of the current study are consistent with our previous observation of a significant interaction between subject responses to a single treatment of PA-directed manual pressures followed by prone press-ups administered to patients with flexion-sensitive LBP.8 In that previous study, subjects who reported a within-session reduction of pain intensity of at least 2/10 demonstrated an increase in ADC at the L5-S1 level, whereas those subjects who did not report a within-session reduction of pain intensity of at least 2/10 had a reduction in ADC. Interestingly, in our current study, we observed nearly identical effect-size (d) differences in ADC at the L5-S1 level between responder groups following a single treatment of high-velocity lumbar manipulation, compared to that observed in our previous study using a 10-minute session of mobilization and prone press-ups. The effect size at L5-S1 was 1.46 in our previous study and 1.49 in our current study. Because our previous study analysis was limited to the L5-S1 level, we cannot make between-study comparisons of ADCs for other lumbar segmental levels.

An interesting finding of this study was that the significant interactions with large between-group postintervention effect sizes occurred in the upper lumbar spine and at the lumbosacral junction. No significant interactions, however, were noted in the mid-lumbar spine at L3-4 and L4-5. To guard against the possibility of a type II error, we performed a power analysis that revealed a statistical power of 0.97 to detect significant between-group differences (effect size, F = 0.6; α = .01; n = 18). Considering this, it is likely that within our sample there was significant regional variation within the lumbar spine, rather than in the entire lumbar spine, in the change in ADC. The reason for this is unknown, but it may be that the “junctional” motion segments of the lumbar spine (ie, those regions that border more rigid portions of the osteocartilaginous spine, such as the thoracolumbar junction and the lumbosacral junction) are affected differently by SMT than are those segments in the middle portion of the lumbar spine.14 Another possible explanation is that increases or decreases in “muscle guarding” of the paravertebral muscles that may occur within session following SMT may influence external forces acting on the lumbar IVD, which may in turn affect diffusion gradients and pathways within the IVD.29,57

There are many theoretical explanations for our findings. Diffusion of water within the IVD is influenced by pressure gradients and chemical forces acting on it, as well as by structural barriers such as dense regions of collagen fibers within the nucleus (ie, a nuclear “cleft”).1,2,28,29,51,57,58,60 Internal pressure gradients are likely to be influenced by externally applied forces,2,29 such as those generated by SMT, that are believed to act on the disc; however, all participants received similar applied forces during the intervention, and only those in the within-session-responder group had increases in diffusion. This suggests an additional interplay involving biochemical events that might have occurred within the IVD in response to SMT. Because increased diffusion was associated with analgesia, it is possible that this increase led to, or was triggered by, some combination of central and/or peripheral chemical activity that influenced pain-regulating neurotransmitters and/or inflammatory mediators.41,42,69 We hope to address these issues in future studies.

Another possible explanation for our findings is that the increased diffusion observed in this study was coincidental, that is, an epiphenomenon similar to the cavitation or “pop” often associated with SMT.21 However, we believe that this is unlikely, because one would expect to observe increased (or decreased) diffusion independent of changes in pain intensity.

Limitations to the external validity of our findings should be acknowledged. Our sample was one of convenience and was composed primarily of young adults with low levels of disability and pain intensity. It is not known if our findings would be reproduced in an older sample of individuals and/or those with nerve root entrapment, medical comorbidities, obesity, or biobehavioral factors.46,66 Although there were small differences in BMI between responder groups, both groups had BMIs within the “ideal range.” It is not known whether the between-group difference in BMI is a meaningful predictor of treatment response. An additional concern is that obesity may influence the rate at which energy from the MRI signal is absorbed by the body (ie, specific absorption rate); therefore, it is unknown if our results would be reproduced in a group of obese individuals seeking care for LBP.

Previous work has suggested that there may be important gender differences in pain reporting and response to care in individuals with LBP.18 We performed a post hoc analysis to investigate this and found that, although women had higher intensities of pain than men for reports of pretreatment pain, as well as reports of high, low, and average pain on a typical day, there were no gender-based differences in the changes in pain or in the relationship between change in pain and change in diffusion.

The longitudinal validity of our findings is unknown. Our results are limited to within-session responses to treatment.38 We do not know how each participant’s perceived pain level changed over time and, therefore, we can make no judgments regarding the long-term relationships between changes in ADC and changes in reported pain intensity following intervention. Finally, it is important to note that our measures of ADC are limited to the central nuclear portion of the IVD. We are currently unable to make judgments regarding the relationships of SMT and pain reports to events occurring in the other portions of the IVD and/or within other structures that are associated with pain perception.12,14 It should be noted that the current use of diffusion-weighted imaging of the IVD is most valuable as a research measure and is not likely to have immediate impact on clinical decision making; that is, we do not recommend its routine use in the management of patients receiving physical therapy care.

Although our findings are preliminary, the similarity of the results of the current study to our previous studies supports the hypothesis that variations in water diffusion occurring in the lumbar IVDs may be one of the mechanisms associated with differences in pain reports following manual therapy and/or exercise interventions for people with LBP. Further investigation is needed in populations with a wider range of pain and disability. Diffusion-weighted imaging of the spine will be of great value to provide dependent measures that help clarify pain pathways, map the geography of the internal disc environment, and assess physiologic changes in response to a wide array of interventions, including exercise approaches, injection treatments, and regenerative medicine procedures.27,67


In a group of participants with LBP who were considered to be candidates for SMT, there were significant differences in the postintervention changes in diffusion of water within the IVDs of the upper lumbar spine and at the L5-S1 level between those who did and did not report a reduction of pain intensity within the treatment session. At these spinal levels, within-session responders demonstrated increased diffusion, whereas not-within-session responders had either a reduction in diffusion or no change. This finding is consistent with previous studies using joint mobilization and suggests linkages between the application of manual therapies and physiologic events within the lumbar IVD and back pain intensity.

Key Points


Participants who reported decreased LBP intensity of at least 2/10 within the same session following a single treatment of SMT also had increases in the diffusion of water within the lumbar IVDs at the L1-2, L2-3, and L5-S1 levels. Participants who did not report a within-session decrease in LBP of at least 2/10 had a reduction or no change in the diffusion of water within the lumbar IVDs at the L1-2, L2-3, and L5-S1 levels.


The results of this study support previous research that suggests a linkage between changes in the diffusion of water within the lumbar IVD and changes in pain following manual therapy treatment to the low back.


The participants in this study had low levels of pain intensity and back pain-related disability. These data may not be generalized to populations of people with high pain intensity or high levels of disability.

Volume 44, Issue 1January 2014

Pages: 1-A191