Cortical function and sensorimotor plasticity are prognostic factors associated with future low back pain after an acute episode: the Understanding persistent Pain Where it ResiDes prospective cohort study : PAIN

Journal Logo

Research Paper

Cortical function and sensorimotor plasticity are prognostic factors associated with future low back pain after an acute episode: the Understanding persistent Pain Where it ResiDes prospective cohort study

Jenkins, Luke C.a,b; Chang, Wei-Jua,b; Buscemi, Valentinaa,b; Liston, Matthewa,b; Humburg, Petera,c; Nicholas, Michaeld; Graven-Nielsen, Thomase; Hodges, Paul W.f; McAuley, James H.a,g; Schabrun, Siobhan M.a,*

Author Information
PAIN 164(1):p 14-26, January 2023. | DOI: 10.1097/j.pain.0000000000002684

1. Introduction

Low back pain (LBP) is the leading cause of years lived with disability worldwide88 with approximately 40% of people experiencing pain for longer than 3 months after onset (termed “chronic LBP”).16 Clinical strategies designed to “treat” LBP once it has become chronic show at best, modest effect sizes regardless of the intervention type.1,62,64,71 An alternative approach for a condition with variable prognosis and treatment response, such as LBP, is stratification of individuals by outcome and targeted treatment.23,32,51,63,76 A key step in implementing this approach is the identification of relevant risk factors.69

Risk factors are often used as building blocks for prognostic models.69 Currently, the models used in clinical practice to determine an individual's risk of developing chronic LBP (eg, STarT Back Screening Tool31; and the short-form Orebro Musculoskeletal Pain Screening Questionnaire)50 rely on self-report psychosocial and symptom-related factors. Although these models allocate higher predicted risk scores to individuals who develop chronic pain, their ability to discriminate between those who will, and will not, develop chronic pain remains limited.38,39 Risk models that integrate psychological (eg, depression and coping strategies) and symptom-related factors (eg, baseline pain intensity and history of prior LBP) explain up to 46% of the variance in LBP outcome.41 Together these data suggest that although psychological and symptom-related risk factors are associated with the development of chronic LBP, a large proportion of variation in outcome is due to risk factors that are currently unmeasured or unknown.26,41

Emerging evidence suggests several neurobiological risk factors with a putative link to LBP outcome that are yet to be evaluated as risk factors in longitudinal studies. These include altered sensory and anterior cingulate cortex excitability,9,22 altered corticomotor excitability,9,74,82 brain-derived neurotrophic factor (BDNF) genotype,4,12,42,90 and BDNF serum concentration.20,48 Prior studies have shown altered excitability in the primary sensory (S1) and primary motor (M1) cortices that is associated with the development and maintenance of chronic pain.19,22,73 Furthermore, a single nucleotide polymorphism in the human BDNF gene is associated with decreased behavioural-driven changes in corticospinal output and cortical organization.3,10,20,25,42,65 As serum BDNF concentration is associated with BDNF genotype,46 both measures are considered markers of neuroplastic potential.4,18

The Understanding persistent Pain Where it ResiDes (UPWaRD) study aimed to recruit and follow a cohort of adults living in Australia who experienced an acute episode of LBP. The primary aim as reported “a priori” in the study protocol34 was to use this cohort to identify biological (with an emphasis on neurophysiological factors), psychological, and sociodemographic risk factors of worse LBP outcome at the 6-month follow-up. The neurobiological risk factors selected for investigation in the protocol were those with a putative link to the development of aberrant cortical and spinal neuroplasticity, hypothesized to explain why some individuals develop chronic pain after an acute episode.

2. Methods

2.1. Study population

Details of the participants, recruitment, and procedures for this study are reported in the study protocol.34 In brief, participants were eligible for inclusion if they had experienced acute LBP, reporting pain of at least 2/10 (Numerical Rating Scale [NRS], 0 = “no pain” and 10 = “worst pain imaginable”) at any time during the 7 days preceding initial screening.59 Pain must have been present for more than 24 hours and less than 6 weeks duration after a period of at least 1 month without pain.17,59,75,92 Acute LBP was defined as pain in the region of the lower back, superiorly bound by the thoracolumbar junction and inferiorly by the gluteal fold. Participants remained eligible if they had pain referred beyond this region that was not caused by lumbosacral radiculopathy. Radiculopathy was suspected if participants reported a history of pain with dermatome-associated distribution, leg pain worse than back pain, worsening leg pain during coughing, sneezing or straining, a positive straight leg raise test or if the participant presented with neurological signs such as dermatome-associated sensory loss, impaired motor function, or attenuated reflexes. Participants who presented with suspected lumbosacral radiculopathy during the clinical examination were excluded from the study and referred to their general practitioner for further assessment. Any individual who presented with suspected serious spine pathology (eg, fracture, tumour, and cauda equina syndrome), other major diseases or disorders (eg, schizophrenia, chronic renal disorder, and multiple sclerosis), a history of spine surgery, and presence of active (ie, being treated) chronic pain conditions or contraindications to the use of transcranial magnetic stimulation (TMS) were excluded.40 Four assessors performed all study-related procedures at laboratories located at Western Sydney University for Neuroscience Research Australia, Sydney, New South Wales, Australia. All procedures were approved by Western Sydney University (H10465) and Neuroscience Research Australia (SSA: 16/002) Human Research Ethics Committees and conducted in accordance with the Declaration of the World Medical Association.91 All participants gave written informed consent. Preplanned methodology was published34 and the study registered (ACTRN12619000002189; preresults), adhering to recommendations of the PROGRESS initiative and TRIPOD statement.69,57

Important demographic and clinical data were collected at baseline for the UPWaRD LBP participants and reported in Table 1. Participants reported their age, sex, height, and weight which were converted into their body mass index. Participants were considered to have no inciting event for their LBP episode if they selected “no obvious cause” from a list of statements including “other.” Participants were asked to report if they had experienced LBP in the past and completed the STarTBack Risk Screening Tool.30 Other clinical data reported by participants were the presence of comorbid health conditions selected from a list including “others” (eg, hypertension) and any current medications (eg, acetaminophen). Participants also reported how many times in the past 3 months they had visited their general practitioner, allied health practitioners, or completed diagnostic tests in relation to their pain. Participants completed the Brief Pain Inventory that was used to rate their average and worst pain intensity, and the degree pain was interfering with their life over the previous 7 days using an 11-point NRS. Low back pain–related disability during their acute LBP episode was measured using the Roland–Morris Disability Questionnaire (RMDQ), and measures of psychological function were obtained, including the 21-item Depression, Anxiety, and Stress Scale (DASS-21); Pain Catastrophizing Scale (PCS); and Pain Self-Efficacy Questionnaire (PSEQ), that are described in more detail below.

Table 1 - Reporting key clinical characteristics of participants at baseline, compared between those with (N = 52) or without (N = 44) low back pain at 6 months.
Characteristic at baseline Low back pain present at 6 months
Yes (N = 52) No (N = 44) P
Age, y (mean ± SD) 39 (15) 40 (17) 0.82
Sex: female, n (%) 30 (57.7) 21 (47.7) 0.3
Body mass index, kg/m2 (mean ± SD)* 26.5 (6.6) 25.3 (4.7) 0.34
Pain below gluteal fold: No, n (%) 31 (62.0) 36 (83.7) 0.02
Inciting event: No, n (%) 19 (36.5) 14 (32.6) 0.69
Compensable injury or sickness benefits: No, n (%) 48 (92.3) 44 (100.0) 0.06
Previous history of LBP: No, n (%) 13 (25.0) 9 (20.5) 0.87
STarT Back Risk Score, n (%)
 Low 32 (62.7) 34 (79.1) 0.06
 Medium 14 (27.5) 9 (20.9)
 High 5 (9.8) 0 (0.0)
Comorbid health conditions, n (%)
 None 33 (66.0) 32 (72.7) 0.48
 Heart disease or hypertension 7 (14.0) 3 (6.8) 0.26
 Lung disease 2 (4.0) 1 (2.3) 0.64
 Diabetes 2 (4.0) 0 (0.0) 0.18
 Ulcer or stomach disease 3 (6.0) 1 (2.3) 0.37
 Kidney disease 0 (0.0) 0 (0.0) NA
 Depression or anxiety 4 (8.0) 5 (11.4) 0.58
 Cancer 0 (0.0) 1 (2.3) 0.28
 Anaemia or other blood diseases 1 (2.0) 1 (2.3) 0.93
 Osteoarthritis 3 (6.0) 2 (4.5) 0.75
 Inflammatory arthropathy 1 (2.0) 0 (0.0) 0.35
 Stroke or other neurological conditions 0 (0.0) 0 (0.0) NA
 Other medical problems† 4 (8.0) 3 (6.8) 0.83
Medication use, n (%)
 None 26 (50.0) 20 (45.5) 0.66
 Not pain related 14 (26.9) 14 (31.8) 0.60
 NSAID 9 (17.3) 4 (9.1) 0.24
 Acetaminophen 9 (17.3) 5 (11.4) 0.41
 Opioid 4 (7.7) 2 (4.5) 0.53
 Benzodiazepine 3 (5.8) 0 (0.0) 0.11
 Antidepressant 3 (5.8) 4 (9.1) 0.53
 Anticonvulsant 2 (3.8) 1 (2.3) 0.66
Healthcare utilization, (mean ± SD)
 General practitioner* 0.8 (1.6) 0.2 (0.5) 0.02
 Allied health 1.3 (2.4) 1.4 (2.9) 0.83
 Diagnostic tests 0.2 (0.5) 0.1 (0.4) 0.39
DASS-21 (mean ± SD)
 Depression* 8.4 (9.6) 2.5 (3.2) <0.001
 Anxiety* 5.5 (5.4) 2.3 (3.0) <0.01
 Stress* 13.1 (10.2) 5.2 (5.0) <0.001
PCS (mean ± SD)
 Rumination* 4.4 (4.2) 2.2 (2.5) <0.01
 Magnification* 3.2 (3.0) 1.6 (2.0) <0.01
 Helplessness* 5.7 (5.3) 3.2 (3.3) <0.01
Pain Self-Efficacy Questionnaire (mean ± SD) 43.8 (13.4) 51.9 (10.0) <0.01
Roland–Morris Disability Questionnaire (mean ± SD) 7.0 (4.9) 4.2 (3.5) <0.01
Average pain intensity in the past wk, NRS (mean ± SD) 5.0 (1.9) 3.3 (1.8) <0.001
Worst pain intensity in the past wk, NRS (mean ± SD) 6.7 (2.1) 5.9 (1.8) 0.07
Pain interference, NRS (mean ± SD)
 General activity 5.0 (2.8) 3.7 (2.7) 0.03
 Mood 4.8 (2.8) 3.0 (2.7) <0.01
 Walking ability* 3.7 (3.2) 2.7 (2.5) 0.08
 Normal work 4.6 (3.0) 3.2 (2.7) 0.02
 Relations with other people* 2.8 (3.2) 1.3 (1.9) 0.01
 Sleep* 4.9 (3.2) 2.8 (2.4) <0.001
 Enjoyment of life 3.9 (2.8) 2.6 (2.7) 0.02
Variable means were compared between participants with or without LBP at 6 months using t tests (continuous variable) or χ2 tests (categorical variables).
Statistically significant values are in bold font.
*Welch t test was performed.
†Other comorbid health conditions include Meniere disease or vestibular migraine, T5 to T8 thoracic compression fracture, endometriosis, hypothyroidism, pituitary microadenoma or prolactinoma, repetitive strain injury wrists, and obstructive sleep apnoea.
DASS-21, 21-item Depression, Anxiety, and Stress Scale; LBP, low back pain; NRS, Numerical Rating Scale; NSAID, nonsteroidal anti-inflammatory drug; PCS, Pain Catastrophizing Scale.

2.2. Candidate predictors recorded at baseline

Fifteen candidate predictors were selected “a priori” based on a theoretical association with the development of chronic pain and supporting evidence from cross-sectional studies.4,22,52,74,82 Justification for each variable and specific methodology is provided in the study protocol.34 In brief:

2.2.1. Sensory and anterior cingulate cortex excitability

Participants were seated comfortably in a chair with feet on the floor and arms relaxed. Participants were asked to sit still, keep their eyes closed, and not to fall asleep for the duration of the test procedure. A bipolar electrode (silver–silver chloride disposable electrodes, interelectrode distance 2.0 cm; Noraxon USA, Scottsdale, AZ) was positioned 3 cm lateral to the L3 spinous process, ipsilateral to the side of the worst LBP, and a constant current stimulator (Digitimer Ltd, Hertfordshire, United Kingdom) delivered nonnoxious electrical stimuli (single stimulus; pulse width 1 ms). The testing intensity was set at 3 times the perceptual threshold. If this testing intensity evoked pain, it was decreased in 1 mA increments until the stimulus was reported as nonnoxious.

Sensory-evoked potentials (SEPs) were recorded in response to 2 blocks of 500 nonnoxious electrical stimuli (∼2 Hz, with random interval schedule of 20% to decrease accommodation), using gold-plated cup electrodes positioned over S1 (3 cm lateral and 2 cm posterior of Cz) on the hemisphere contralateral to the side of worst LBP and referenced to Fz using the International 10/20 system.33 Electrode impedance was maintained at <5 kΩ. EEG signals were amplified 50,000 times, band pass filtered between 5 and 500 Hz, and sampled at 1000 Hz using a Micro1401 data acquisition system and Signal software (CED Limited, Cambridge, United Kingdom).

Individual SEP traces were inspected, and those with eye movements, muscle artefacts, or electrical interference were rejected. Less than 15% of all SEP traces were excluded. Remaining traces from the 2 SEP blocks were averaged for each participant and used for analysis.9 The averaged wave form was full-wave rectified, and the area under the rectified curve (µV) was determined for the N80 (between the first major downward deflection of the curve after stimulus onset and the first peak, N80), N150 (between the first and second peak, N80 and N150, respectively), and P260 (between the second negative peak, N150, and the positive deflection of the curve starting around 150 ms after stimulus onset, P260) time epochs.19Figure 1 displays a rectified trace with time epochs used for analysis. The SEP area measurement was chosen for analysis as it is less susceptible to signal-to-noise ratio concerns14; and considered superior to peak-based measures for assessing event-related potentials.19,37,66,93

Figure 1.:
Example of a SEP recorded from the paraspinal muscles (average of 500 traces) from a single participant. A shows the N80, N150, and P260 SEP peaks. B shows the area under the rectified curve for each time epoch (N80, N150, and P260 area) that was used for analysis. SEP, sensory-evoked potential.

2.2.2. Corticomotor excitability

Participants underwent a standardised TMS mapping procedure as described in the study protocol and in previous studies.34,61,72,82 Single-pulse, monophasic stimuli (Magstim 200 stimulator, 7 cm figure-of-eight coil; Magstim Co. Ltd. Dyfed, United Kingdom) were delivered over the primary motor cortex (M1) contralateral to the side of the worst LBP. The coil was positioned tangential to the skull with the handle pointing posterior laterally at 45° from midline.5,36,56 Electromyography (EMG) was recorded from the paraspinal muscles 3 cm lateral to the spinous process of L3 and 1 cm lateral to the spinous process of L5 on the side of the worst LBP using disposable Ag/AgCl surface electrodes (Noraxon USA, Inc).47,61 Ground electrodes were placed over the anterior superior iliac spine bilaterally. Electromyography data were amplified 1000×, filtered between 20 and 1000 Hz, and sampled at 2000 Hz using a Micro1401 data acquisition system and Spike2 software (CED Limited). As 120% of active motor threshold often exceeds the maximum stimulator output82; all stimuli were delivered at 100% of stimulator output while participants activated the paraspinal muscles to 20 ± 5% of their EMG recorded during a maximum voluntary contraction (determined as 20% of the highest root mean square EMG averaged over 1 second during 3, 3-s maximal muscle contractions performed against manual resistance in sitting).74,72,82 Real-time feedback of time paraspinal muscle root mean square EMG and the target level was displayed on a monitor for the duration of the test procedure.81 All TMS procedures adhered to the TMS checklist for methodological quality.11

Transcranial magnetic stimulation motor–evoked potentials (MEPs) were analysed using MATLAB 2019a (The MathWorks, Portola Valley, CA). Onset and offset of the MEPs in each individual trace were visually identified then averaged at each scalp site. Paraspinal MEP amplitudes were normalized to the largest MEP amplitude across sites and superimposed over the respective scalp sites to generate a topographical map. A scalp site was considered active if the normalised MEP amplitude was equal to or greater than 25% of the peak response.9 Normalised values below 25% of the peak response were removed and the remaining values rescaled from 0% to 100%.72,82

Two parameters were calculated from the normalised motor cortical maps. First, L3 and L5 map volumes were calculated as the sum of normalized MEP amplitudes recorded at all active scalp sites.89 Second, the centre of gravity (CoG), defined as the amplitude weighted centre of the map, was calculated for the M1 cortical representation of L3 and L5 paraspinal muscles using the formula: CoG = ΣVi × Xi/ΣVi, ΣVi × Yi/ΣVi where: Vi = mean MEP response at each site with the coordinates Xi, Yi.83,89 Distance between the L3 and L5 CoG (L3/L5 CoG overlap) was calculated as the Euclidean distance (ED) using the formula: ED = ((YL3YL5)2+(XL3XL5)2), where Y = anterior–posterior coordinates; X = medial–lateral coordinates of L3 and L5.85,86

2.2.3. Brain-derived neurotrophic factor genotyping

Buccal swabs were taken on the day of baseline testing (Isohelix DNA Isolation Kit).12 Samples were immediately frozen and stored at −80°C. Genomic DNA samples were polymerase chain reaction amplified and sequenced by the Australian Genome Research Facility. Genotyping was performed as recommended by the manufacturer with reagents included in the iPLEX Gold SNP genotyping kit (Agena) and the software and equipment provided with the MassARRAY platform (Agena, San Diego, CA).13 Consistent with prior investigations,4,45,54 the BDNF gene was coded as a dichotomous variable (AA/AG or GG). The more common G allele encodes valine (Val), whereas the A allele encodes methionine (Met).

2.2.4. Brain-derived neurotrophic factor serum concentration

Peripheral venous blood was drawn into serum tubes (BD, SST II Advance) through venepuncture of the median cubital vein at baseline. The sample was clotted (30 minutes, room temperature) then separated by centrifugation (2500 rpm, 15 minutes). The samples were pipetted into 50 μL aliquots and stored at −80°C. After thawing, the Simple plex Ella platform was used to analyse the specific expression of BDNF. In brief, 10 μL samples were diluted with 90 μL of sample diluent then added to the cartridge, according to the standard procedure provided by the manufacturers (Protein Simple, San Jose, CA). All steps in the immunoassay procedure were performed automatically, and scans were processed with no user activity. Cartridges included built-in lot–specific standard curves. Single data (pg/mL) for each sample were automatically calculated and converted to ng/mL for statistical analysis. The limit of detection was 5.25 pg/mL. Fifteen randomly selected samples were analysed in duplicate and demonstrated near perfect correlation (r = 0.98, P = < 0.001).

2.2.5. Psychological status

Three questionnaires were used to assess specific aspects of psychological status known to be of relevance to the development of chronic LBP: depression and anxiety7 and pain catastrophising6 and pain beliefs.49 The DASS-21 was used to assess depression, anxiety, and stress. The 13-item PCS assesses distressing thoughts related to painful experiences.78 A total score between 0 and 52 was calculated, where higher scores represent more severe catastrophic thoughts about pain.78 The 10-item PSEQ was used to assess an individual's beliefs in their ability to perform a range of functional activities while in pain.60 The DASS-21 (Cronbach α = 0.88), PCS (Cronbach α = 0.87), and PSEQ (Cronbach α = 0.92) have all demonstrated high degrees of reliability (internal consistency).28,60,78

2.2.6. Symptom-related factors

Baseline pain intensity was drawn from the Brief Pain Inventory administered on the day of baseline testing. Participants rated their pain on average over the previous week using an 11-point NRS.15 Participants were considered to have a previous history of LBP if they answered “yes” to the question: “Have you experienced LBP in the past”?

2.2.7. Demographics

Age and sex were collected from all participants on the day of baseline testing.

2.3. Primary and secondary outcomes recorded at the 6-month follow-up

The primary outcome was average pain intensity over the past week, assessed using the NRS at the 6-month follow-up. The secondary outcome was disability assessed using the 24-point RMDQ at the 6-month follow-up.70 The primary and secondary outcomes were combined to distinguish between those who did and did not report LBP at 6 months, defined as an NRS score ≥2 on average over the previous 1 week or ≥7 on the RMDQ scale at the 6-month follow-up.30,80 This dichotomised outcome measure was used in the secondary analysis, described in detail below.

2.4. Sample size

Sample size was calculated “a priori” and is described in detail in the study protocol.34 In brief, we assumed that at least 5 variables would show no association with the outcome and be excluded from analysis, and 20% of participants would be lost to follow-up. Therefore, 120 participants were required to ensure at least 10 subjects per variable within linear regression models. Sample size for the logistic regression model was calculated using the rule of thumb that 5 events per candidate variable (EPV) are required for adequate statistical power.87

2.5. Statistical analyses

All analyses were conducted with the statistical programming language R, version 4.0.3 (R Development Core Team, Vienna, Austria).68 Statistical significance was accepted at P ≤ 0.05. Continuous data are presented as mean ± SD and categorical data as frequency and percent (%). Candidate predictors measured on a continuous scale were not dichotomised as this can increase risk of bias in regression models.57,58 Categorical candidate predictors were coded as follows: sex [male or female], previous history of LBP [yes or no], and BDNF genotype [AA/AG or GG].

All missing data were imputed using the multiple imputation by chained equations procedure, and 30 imputed data sets were generated.84 Missing data within the candidate predictors are described in Table 2. Comparisons were made between participants who did or did not complete follow-up using independent samples t test and Fisher exact test for continuous and categorical data, respectively. Spearman rank correlation coefficients were used to determine whether there was any evidence of collinearity between measures of cortical excitability.24

Table 2 - Missing data within candidate predictor variables.
Candidate predictor Number of missing data (%) Reason
SEP N80 component area (µV)
SEP N150 component area (µV)
SEP P260 component area (µV)
2 (2%)
2 (2%)
2 (2%)
Equipment failure
L3 map volume (cm2)
L5 map volume (cm2)
L3/L5 centre of gravity overlap (cm)
31 (25.8%)
31 (25.8%)
31 (25.8%)
Unresolvable noise to signal ratio (N = 5)
Consent not obtained (N = 13)
Participant unable to tolerate (N = 7)
Equipment failure (N = 6)
BDNF genotype (0 = AA/AG, 1 = GG)
BDNF serum concentration (ng/mL)
0 (0%)
30 (25%)
Researcher error during phlebotomy (N = 10)
Consent not obtained (N = 11)
Simple plex Ella machine error (N = 9)
PCS (0-52 scale)
DASS-21 (0-63 scale)
PSEQ (0-60 scale)
2 (2%)
7 (5.8%)
3 (2.5%)
Incorrect completion of the questionnaire
NRS score at T1 (0-10 scale)
Previous history of low back pain (0 = no, 1 = yes)
2 (2%)
4 (3%)
Incorrect completion of the questionnaire
Age (y)
Sex (0 = female, 1 = male)
0 (0%)
0 (0%)
All variables were treated as continuous, with the exception of sex, history of low back pain, and BDNF genotype.
BDNF, brain-derived neurotrophic factor; DASS-21, 21-item Depression, Anxiety, and Stress Scale; L3, electrode recording site 3 cm lateral to the L3 spinous process; L5, electrode recording site 1 cm lateral to the L5 spinous process; LBP, low back pain; NRS, 11-point Numerical Rating Scale; PCS, Pain Catastrophizing Scale; PSEQ, Pain Self-Efficacy Questionnaire; SEP, sensory-evoked potential; T1, within 6 weeks of acute LBP onset.

Next, the remaining candidate predictors underwent a variable selection procedure using the least absolute shrinkage and selection operator (lasso) technique.79 This is a variation from the variable selection procedures described in the published study protocol.34 For a detailed description about lasso variable selection procedure used in this study, please refer to Supplemental File 1 (available at For the primary analysis, outcome variables of pain and disability were treated as continuous data and the strongest risk factors were selected using the lasso variable selection procedure. The results of the primary analyses are presented using hierarchical linear regression. Goodness of fit for the linear models was assessed using the R2 and adjusted R2 value.

In the secondary analysis, the lasso variable selection procedure was repeated within a logistic regression model to determine the strongest risk factors of future LBP at 6-months (dichotomized outcome: NRS ≥ 2 or RMDQ ≥ 7). The goodness of fit for the final logistic regression model was estimated following ten-fold crossvalidation as described in Supplemental File 1 (available at and reported as the C-statistic (ie, area under the receiver operating characteristic curve).

3. Results

3.1. Participant characteristics

Between April 14, 2015, and January 23, 2019, 498 participants presented with an acute episode of LBP for screening and 120 were included in the study sample (Fig. 2). Two hundred and seven participants (41.5%) were ineligible because they had chronic LBP, 2 were ineligible because they had previous spinal surgery, and 3 were ineligible because physical examination by the study investigator suggested a diagnosis of lumbosacral radiculopathy. Of the 286 eligible participants, 94 (32.9%) failed to respond to contact attempts to organise baseline assessment and 72 (25.2%) declined participation after reviewing the study information sheet. The date of the final participant's 6-month follow up was July 25, 2019. Baseline data were obtained on average 2.4 weeks (SD 1.4, range 1 day-6 weeks) after the onset of an acute LBP episode. Seventy-two participants (60%) were coded GG, and 48 (40%) were coded AA/AG for BDNF genotype. According to the Hardy–Weinberg equilibrium, this observed distribution does not differ significantly from the expected rate (χ2 = 1.27, df = 1, P = 0.26). Table 1 shows baseline demographic and clinical characteristics. Participants with LBP at 6 months were more likely at baseline assessment to have pain referred below the level of the gluteal fold and consult more frequently with general practitioners. These participants also had higher levels of psychological distress, lower pain self-efficacy, higher average, and worst pain intensity and experience more disability and pain interference. There were no other statistically significant differences at baseline assessment between participants who did or did not experience LBP at 6 months.

Figure 2.:
Study flow chart. *Defined as LBP lasting for longer than 6 weeks or an LBP episode preceded by a period of less than 1 month without pain. LBP, low back pain.

Follow-up at 6 months was completed in 96 participants (80%). Missing follow-up cases were due to the participant failing to respond to multiple contact attempts. The intention was to determine the presence of LBP at 6 months (183 days); in practice, follow-up occurred in a mean of 194 (SD = 20) days after entering the study with an acute episode of LBP. Baseline candidate predictors and their univariable association with chronic LBP are provided in Table 3. Individual participant data for normalised motor cortical maps and SEPs are displayed in Figures 3 and 4.

Table 3 - Candidate predictors selected “a priori” and compared between participants, with (N = 52) or without (N = 44) low back pain at 6 months.
Low back pain present at 6 months
Characteristic at baseline Yes (N = 52) No (N = 44) OR (95% CI) P
Gender: female (%) 30 (58) 21 (48) 0.73 (0.34-1.58) 0.42
History of LBP: No (%) 13 (25) 9 (20) 0.71 (0.27-1.86) 0.48
BDNF genotype: AA/AG (%) 23 (44) 12 (27) 0.42 (0.18-0.99) 0.05
L3 map volume (cm2) 7.0 (3.0) 9.9 (4.0) 0.88 (0.78-1.00) 0.05
L5 map volume (cm2) 7.5 (3.3) 8.3 (3.2) 0.96 (0.85-1.08) 0.49
L3/L5 CoG overlap (cm) 0.5 (0.5) 0.8 (0.6) 0.54 (0.26-1.14) 0.10
Log-transformed N80 area (µV) −3.4 (1.4) −1.9 (1.3) 0.49 (0.35-0.68) <0.001
Log-transformed N150 area (µV) −3.2 (1.3) −1.9 (1.3) 0.50 (0.36-0.70) <0.001
Log-transformed P260 area (µV) −3.1 (1.2) −1.8 (1.4) 0.50 (0.35-0.71) <0.001
BDNF serum concentration (ng/mL) 52.1 (12.2) 50.7 (12.5) 1.00 (1.00-1.00) 0.39
PCS (0-52 scale) 13.6 (11.7) 7.0 (7.0) 1.07 (1.02-1.13) 0.01
DASS-21 (0-63 scale) 26.9 (22.0) 10.1 (8.2) 1.06 (1.02-1.10) 0.01
PSEQ (0-60 scale) 43.5 (13.5) 51.0 (9.1) 0.95 (0.91-0.99) 0.01
Average baseline pain intensity (0-10 scale) 5.0 (1.9) 3.3 (1.8) 1.57 (1.22-2.01) <0.001
Age (y) 39 (15) 40 (17) 1.00 (0.97-1.02) 0.92
Statistically significant values in bold font.
Values are numbers (%), means (SD), unadjusted odds ratios (ORs) with corresponding 95% confidence intervals (95% CIs), and P values from univariable logistic regression models. Values for each baseline characteristic are calculated from raw data. Odds ratio and P values are pooled after multiple imputation. The odds ratio is the increase in odds per unit increase in the predictor.
AA/AG, G allele encodes Val, A allele encodes Met; BDNF, brain-derived neurotrophic factor; CoG, centre of gravity; DASS-21, 21-item Depression, Anxiety, and Stress Scale; NRS, Numerical Rating Scale; PCS, Pain Catastrophizing Scale; PSEQ, Pain Self-Efficacy Questionnaire.

Figure 3.:
Motor cortical maps for 2 representative participants at the L3 recording site normalized to peak MEP amplitude. A displays large volume corticomotor excitability during acute LBP in a participant who was recovered at 6 months. B displays small volume corticomotor excitability during acute LBP in another participant who experienced chronic or recurrent LBP at 6 months. The dashed lines indicate the location of the vertex (coordinate 0,0). The coloured scale represents the proportion of the maximum MEP amplitude. Warmer colours represent higher excitability. LBP, low back pain; MEP, motor-evoked potential.
Figure 4.:
SEP recorded in response to stimuli to the paraspinal muscles at baseline (average of approx. 500 traces) for 2 representative participants. The black trace displays high SEP excitability in the acute stage of LBP in a participant who was recovered at 6 months. The red trace displays low SEP excitability in the acute stage of LBP in a participant who reported LBP at 6 months. LBP, low back pain; SEP, sensory-evoked potential.

3.2. Incidence of chronic low back pain

Analysis of complete cases revealed that 54% of participants reported LBP at the 6-month follow-up (average pain intensity 3.9 [SD = 1.7] and average RMDQ score 4.8 [SD = 5.4]) and could be considered to have persistent or recurring LBP. This is comparable with the incidence of chronicity in previous Australian estimates.29 Of the 52 participants deemed to have LBP, only 12 participants (23%) had an RMDQ score ≥ 7. The remaining 46% of participants were classified as recovered (average pain intensity 0.3 [SD = 0.5] and average RMDQ score 0.9 [SD = 1.6]).

3.3. Continuous data distribution and collinearity

All variables were normally distributed except the N80, N150, and P260 SEP area, and these variables were log transformed. Baseline characteristics were then compared between participants who did (N = 96), and did not (N = 24), complete follow-up at 6 months (Table 4). Apart from N150 and P260 SEP area measures, no statistically significant differences at baseline were identified between participants who did, or did not, complete follow-up; however, the N80 SEP area did demonstrate a strong tendency (P = 0.06). Next, Spearman correlation coefficients were calculated between all measures of cortical excitability (Table 5). No strong correlation was identified between SEP and TMS measures. The N80 and N150 (rs = 0.84, P = <0.001) and N80 and P260 (rs = 0.85, P = <0.001) SEP area values were strongly correlated. As the N150 and P260 SEP area measurements may have impacted the missing at random assumption and demonstrated collinearity with the N80 SEP area, they were excluded from further analyses. The remaining 13 candidate predictors were subjected to the λ-1se variable selection procedure to identify relevant risk factors.

Table 4 - Comparison of candidate predictors selected “a priori” and compared between participants who did and did not complete 6-month follow-up.
Characteristic Completed follow-up (N = 96) Did not complete follow-up (N = 24) P
Gender: female (%) 53.1 33.3 0.08
History of LBP: No (%) 22.9 12.5 0.32
BDNF genotype: AA/AG (%) 36.5 54.2 0.11
L3 map volume (cm2) 8.4 (4.0) 10.0 (5.5) 0.13
L5 map volume (cm2) 7.8 (3.5) 8.5 (4.2) 0.43
L3/L5 CoG overlap (cm) 0.7 (0.6) 0.5 (0.4) 0.21
Log-transformed N80 area (µV) −1.2 (0.7) −1.5 (0.6) 0.06
Log-transformed N150 area (µV) −1.1 (0.6) −1.4 (0.6) 0.04
Log-transformed P260 area (µV) −1.1 (0.6) −1.4 (0.6) 0.03
BDNF serum concentration (ng/mL) 52.2 (13.7) 50.9 (13.7) 0.71
PCS (0-52 scale) 10.6 (10.3) 12.6 (8.9) 0.35
DASS-21 (0-63 scale) 19.1 (19.3) 27.2 (24.7) 0.17
PSEQ (0-60 scale) 47.1 (12.1) 42.5 (11.5) 0.10
Average baseline pain intensity (0-10 scale) 4.2 (2.0) 4.6 (1.3) 0.37
Age (y) 40 (16) 40 (12) 0.91
Statistically significant values in bold font.
Values are numbers (%), means (SD) compared between LBP participants who did and did not follow-up at 6 months follow-up using t tests (continuous data, normally distributed), or Fisher exact test (categorical data). All values are calculated from raw data.
AA/AG, G allele encodes Val, A allele encodes Met; BDNF, brain-derived neurotrophic factor; CoG, centre of gravity; DASS-21, 21-item Depression, Anxiety, and Stress Scale; LBP, low back pain; NRS, Numerical Rating Scale; PCS, Pain Catastrophizing Scale; PSEQ, Pain Self-Efficacy Questionnaire.

Table 5 - Spearman correlation coefficients between measures of cortical excitability during acute low back pain.
L3 map volume (cm2) L5 map volume (cm2) Log-transformed N80 area (µV) Log-transformed N150 area (µV) Log-transformed P260 area (µV) L3/L5 CoG overlap (cm)
L3 map volume (cm2) 0.82* 0.02 0.07 0.07 −0.15
L5 map volume (cm2) −0.05 −0.06 −0.02 −0.29*
Log-transformed N80 area (µV) 0.84* 0.85* 0.20
Log-transformed N150 area (µV) 0.90* 0.21
Log-transformed P260 area (µV) 0.20
L3/L5 CoG overlap (cm)
All values are calculated from raw data.
*Correlation is significant at the 0.01 level (2-tailed).
Correlation is significant at the 0.05 level (2-tailed).
CoG, centre of gravity.

3.4. Risk factors associated with pain intensity at 6 months

Higher baseline pain intensity (P = <0.001), higher depression and anxiety (DASS-21: P = 0.31), higher pain catastrophizing (PCS: P = 0.39), lower N80 SEP area (P = 0.01), and lower L3 map volume (P = <0.01) were the risk factors demonstrating the strongest association with pain intensity over the previous week (continuous outcome) at 6 months and, thus, were entered into a hierarchical regression model. The results of hierarchical regression modelling are presented in Table 6. The baseline pain intensity explained 22% of the variance in pain intensity at 6 months. Addition of the DASS-21 and PCS instruments explained a further 10% of the variance (F[2,82] = 4.99, P = 0.01), and the addition of novel neurobiological risk factors (N80 SEP area and L3 map volume) explained a further 15% of the variance in the 6-month pain intensity (F[2,74] = 8.61, P = <0.001). In combination, these 5 variables explained 47% (R2 = 0.47 95% confidence interval [CI] = 0.31-0.60) of the total variance in the 6-month pain intensity.

Table 6 - Hierarchical linear regression model predicting the 6-month pain intensity.
Variables entered R 2 Adjusted R 2 Significance of adjusted R 2 change B (95% CI) (final model) Significance of B
Step 1 Average baseline pain intensity (0-10 scale) 0.22 0.22 0.33 (0.17-0.48) <0.001
Step 2 DASS-21 (0-63 scale)
PCS (0-52 scale)
0.32 0.30 0.01 0.14 (−0.13 to 0.41)
0.10 (−0.13 to 0.35)
Step 3 L3 map volume (cm2)
Log-transformed N80 area (µV)
0.47 0.44 <0.01 −0.29 (−0.48 to −0.10)
−0.25 (−0.42 to −0.07)
Statistically significant P values in bold font.
Hierarchical linear regression was performed with the 6-month pain intensity as the dependent variable. Predictors identified from the λ-1se variable selection procedure were entered into the model. Step 1 included average baseline pain intensity, step 2 included psychological risk factors, and step 3 included measures of sensorimotor cortical excitability. The significance of the adjusted R2 change was calculated using a one-way ANOVA. All results reported in this table are pooled across the imputed datasets.
ANOVA, analysis of variance; B, unstandardized beta coefficient; CI, confidence interval; DASS-21, 21 item Depression, Anxiety, and Stress Subscale; PCS, Pain Catastrophizing Scale.

3.5. Risk factors associated with disability at 6 months

Higher pain catastrophizing (PCS: B = 0.47, 95% CI = 0.30-0.64, P = <0.001) and older age (B = 0.24, 95% CI = 0.08-0.42, P = <0.01) were the strongest risk factors associated with disability over the previous week (continuous outcome) at 6 months. These 2 factors explained 30% of the variance in disability outcome at 6 months (R2 = 0.30, 95% CI = 0.14-0.46).

3.6. Risk factors associated with the presence or not of low back pain at 6 months

When the dichotomous variable of LBP at 6 months was designated on the basis of pain or disability above a threshold value (NRS ≥ 2 or RMDQ ≥ 7), 6 risk factors remained in the multivariable logistic regression model after λ-1se variable selection and 10-fold crossvalidation (Table 7). Lower primary sensory cortex excitability (N80 SEP area: P = <0.01), lower corticomotor excitability (L3 map volume: P = 0.07), BDNF genotype MET allele carriers (P = 0.18), higher depression and anxiety (DASS-21 score: P = 0.01), no history of LBP (P = 0.11), and higher baseline pain intensity (P = <0.01). The C-statistic for the multivariable model was 0.91 (95% CI = 0.84-0.95).

Table 7 - Multivariate logistic prediction model of risk of low back pain at 6 months.
Predictor B Odds ratio (95% CI) Significance of B
Log-transformed N80 area (µV) −0.79 0.45 (0.27-0.75) <0.01
L3 map volume (cm2) −0.21 0.81 (0.65-1.01) 0.07
BDNF genotype (0 = AA/AG, 1 = GG) −0.87 0.41 (0.10-1.69) 0.22
DASS-21 (0-63 scale) 0.07 1.07 (1.02-1.23) 0.01
Average baseline pain intensity (0-10 scale) 0.54 1.71 (1.17-2.50) 0.01
History of LBP (0 = no and 1 = yes) −1.29 0.27 (0.05-1.61) 0.15
Statistically significant values in bold font.
All results reported in this table are pooled across the imputed data sets. The odds ratio is the increase in odds per unit increase in the predictor.
AA/AG, G allele encodes Val, A allele encodes Met; B, unstandardized beta coefficient; BDNF, brain-derived neurotrophic factor; CI, confidence interval; DASS-21, 21-item Depression, Anxiety, and Stress Subscale; GG, G allele encodes Val; LBP, low back pain.

3.7. Sensitivity analysis

A sensitivity analysis was conducted to explore the effect of imputing missing 6-month outcome data on study results. Participants who did not return for follow-up at 6 months were removed from the data set and the analysis repeated. Risk factors included in the multivariable regression models after λ-1se variable selection were identical for all outcomes. The multivariable linear model predicting 6-month pain intensity had a R2 value of 0.45 (95% CI = 0.29-0.60), and the linear model predicting 6-month disability score had a R2 value of 0.32 (95% CI = 0.17-0.47). Goodness of fit for the multivariable logistic regression model was also comparable, demonstrating a C-statistic of 0.88 (95% CI = 0.81-0.95).

4. Discussion

This prospective longitudinal cohort study is the first to investigate biological measures of sensorimotor cortical function and neuroplasticity as risk factors of the 6-month outcome after an acute episode of LBP. A novel finding was the identification of lower primary sensory cortex excitability (N80 SEP area) and lower corticomotor excitability (L3 map volume) in the acute stage of LBP as risk factors for higher pain intensity reported at 6 months. When these variables were combined with psychological factors (higher emotional distress) and symptom-related factors (no LBP history and higher baseline pain intensity), they explained a similar percentage of the variance in the 6-month pain intensity (47%) as earlier models that integrate psychological and symptom-related factors (46%)41; but uniquely, addition of the novel neurobiological risk factors explained a further 15% of the variance in the 6-month pain intensity. These findings provide further support for the importance of assessing diverse phenotypic traits, across a range of neurobiological, psychological, symptom-related, and demographic domains, when attempting to predict LBP outcome. For example, in a longitudinal study using cluster analyses, individuals with the worst recovery from acute LBP at the 6-month follow-up displayed higher depression-like symptoms in conjunction with higher serum concentrations of tumor necrosis factor.43 We caution readers not to infer causal relationships between any of the risk factors identified in this study and the 6-month LBP outcome. No attempt was made to identify, or control for, relevant confounders as the aim of this study was to identify novel risk factors of the 6-month LBP outcome, with a clear emphasis on neurobiological variables. As stated within the PROGRESS Initiative, prognostic factor research has many uses in health care and clinical research.69 The study data we report here identifies for the first time novel neurobiological risk factors of future LBP outcome. Future studies should assess their predictive value over established prognostic factors.

The primary analysis aimed to the candidate predictors most strongly associated with pain intensity and disability at 6 months as continuous outcomes. These analyses revealed largely discrete risk factors for the outcomes of pain (strongest risk factors of lower sensory cortex excitability [N80 SEP area], lower corticomotor excitability [L3 map volume], and higher baseline pain intensity with a lesser contribution from higher emotional distress and higher pain catastrophising) and disability (older age and higher pain catastrophizing), with no neurobiological risk factors identified for disability following the lasso variable selection procedure. Patient-reported outcomes of pain (eg, NRS) and disability (eg, RMDQ) are known to be weakly correlated with each other when LBP is chronic,44 and previous research has shown that disability is more closely aligned with psychological risk factors than is pain intensity. For example, in a large cohort study of people with chronic LBP, psychological questionnaires believed to assess “pain-related distress” explained 51% of the variance in disability, but only 35% of the variance in pain intensity.7 The absence of a relationship between sensorimotor cortex excitability and disability suggests other factors predict this outcome. This is an important consideration when interpreting the current findings. People who develop pain-related disability account for a significant proportion of the total healthcare costs associated with LBP, thus identifying risk factors of LBP-related disability is of critical importance.21,26 Previous studies have suggested a link between pain-related neurobiology (eg, neuroendocrine responses to pain55 and pain-specific neurophysiological and psychological processes that may in turn drive pain-related disability). Future research should seek to identify and validate neurobiological risk factors for LBP-related disability.

Our findings on the psychological and symptom-related risk factors for chronic LBP are consistent with prior work.27,52,67 We confirm previous findings that show higher emotional distress and higher pain in the acute stage of LBP are risk factors for the development of persistent or recurring LBP. The discovery that lower sensory cortex excitability and lower corticomotor excitability are risk factors for future LBP is novel. When N80 SEP area, L3 map volume, and BDNF genotype were combined with psychological (higher emotional distress) and symptom-related (no LBP history and higher baseline pain intensity) factors, the multivariable logistic regression model could discriminate between those with and without LBP at 6 months follow-up (C-statistic 0.91 [0.84-0.95]).

The discovery that low primary sensory cortex excitability and low corticomotor excitability in the acute stage of LBP are risk factors for worse LBP outcome at 6 months builds on a growing body of evidence that suggests measures of brain structure and function are important risk factors of LBP outcome. For example, in subacute LBP, greater functional connectivity of corticostriatal circuitry is associated with chronic LBP at 1 year.2,53 Using causal inference analyses, we have shown that low sensory cortex excitability (N80 SEP area) in acute LBP is a cause, and not an epiphenomenon, of chronic pain.35 Further research is required to determine if these neurobiological risk factors are modifiable.

Despite a robust and rigorous approach to data collection and analyses, this study has some limitations. First, MEPs used to quantify corticomotor excitability were obtained by delivering TMS at 100% of stimulator output over the M1 during a submaximal paraspinal muscle contraction. Although this methodology has been reported previously,74,72,82 recent methodological developments have shown that pseudorandomly delivering 90 stimuli over a 5 × 7 cm grid with a high-intensity coil reliably maps the M1 paraspinal muscle representation and minimises acquisition times (ICC = 0.82 [95% CI 0.66-0.91]).8 This approach should be considered in future studies and may decrease participant attrition and missing data. Second, despite some evidence, suggesting as few as 5 EPV may provide adequate statistical power for logistic regression87; the most commonly accepted rule for minimising overfitting encourages EPV of 20 or more, and our sample size does not achieve this.58 As recommended by the TRIPOD statement, penalized regression was applied to address this limitation of the study, and results of the logistic model underwent 10-fold crossvalidation to minimise optimism.57,58 However, additional, large-scale studies are needed to confirm and validate our findings. Third, missing data were present within some candidate predictors, and no outcome data were available for 24 participants. To limit the impact of missing data on our findings, missing values were imputed using the multiple imputation by chained equations procedure and a sensitivity analysis identified no differences between models developed with or without imputed outcome data. Fourth, the outcome measures for pain and disability asked participants to report these features over the past week, and we cannot determine whether LBP had persisted because the acute episode (ie, defined as chronic) or had reoccurred as a new discrete episode (ie, defined as recurrent). For this reason, we refer to our dichotomised results as risk factors for future LBP and cannot determine whether risk factors differ between chronic or recurring LBP. Finally, the dichotomous outcome of future LBP was based on a threshold value of ≥2 on the NRS or ≥7 on the RMDQ scale, as used in prior studies predicting LBP outcome.30,80 There remains no widely accepted cut-off for classifying LBP outcome based on subjective pain and disability scores, and transforming a continuous subjective measure into discrete categories remains challenging.17 The arbitrary cut-off chosen in this study to define recurrent or future LBP may not optimally reflect the experience of individuals with LBP, and continuous outcomes remain preferable, consistent with our primary analysis.75 Finally, 0.8% of individuals who presented for screening in the UPWaRD study were deemed ineligible because of the presence of radiculopathy. This figure is similar to that reported by a large-scale inception cohort of 973 individuals with acute, nonspecific LBP presenting to Australian primary care (1.2% of screened participants excluded due to radiculopathy).29 The relatively low prevalence of radiculopathy in these studies is likely explained by the recruitment process. Referring healthcare practitioners were provided with information on study inclusion criteria, and thus, participants with suspected radiculopathy were unlikely to have been referred for screening.

4.1. Conclusion

This study identified novel risk factors relating to cortical function and neuroplasticity for the development of future LBP. Neurobiological risk factors, when added to a multivariable linear regression model, explained a further 15% of the variance in the 6-month pain intensity. Future research should seek to determine whether the neurobiological risk factors identified in this study are modifiable causal mechanisms.

Conflict of interest statement

The authors have no conflicts of interest to declare.

Appendix A. Supplemental digital content

Supplemental digital content associated with this article can be found online at


The authors thank the patients who participated in the study. The authors thank Dr Valerie Wasinger, Biosciences Precinct Laboratory, Level 2, Biosciences South (E26), UNSW Sydney, NSW 2052, for her expertise and assistance with serum analyses.

This work was supported by grant 1059116 from the National Health and Medical Research Council (NHMRC) of Australia. S.M. Schabrun and P.W Hodges received salary support from the National Health and Medical Research Council of Australia (1105040 and 1102905, respectively). T. Graven-Nielsen is a part of Center for Neuroplasticity and Pain (CNAP) that is supported by the Danish National Research Foundation (DNRF121).


[1]. Artus M, van der Windt DA, Jordan KP, Hay EM. Low back pain symptoms show a similar pattern of improvement following a wide range of primary care treatments: a systematic review of randomized clinical trials. Rheumatology (Oxford, England) 2010;49:2346–56.
[2]. Baliki MN, Petre B, Torbey S, Herrmann KM, Huang L, Schnitzer TJ, Fields HL, Apkarian AV. Corticostriatal functional connectivity predicts transition to chronic back pain. Nat Neurosci 2012;15:1117–19.
[3]. Bath KG, Lee FS. Variant BDNF (Val66Met) impact on brain structure and function. Cogn Aff Behav Neurosci 2006;6:79–85.
[4]. Baumbauer KM, Ramesh D, Perry M, Carney KB, Julian T, Glidden N, Dorsey SG, Starkweather AR, Young EE. Contribution of COMT and BDNF genotype and expression to the risk of transition from acute to chronic low back pain. Clin J Pain 2020;36:430–439.
[5]. Brasil-Neto JP, McShane LM, Fuhr P, Hallett M, Cohen LG. Topographic mapping of the human motor cortex with magnetic stimulation: factors affecting accuracy and reproducibility. Electroencephalogr Clin Neurophysiol 1992;85:9–16.
[6]. Burton AK, Tillotson KM, Main CJ, Hollis S. Psychosocial predictors of outcome in acute and subchronic low back trouble. Spine (Phila PA 1976) 1995;20:722–8.
[7]. Campbell P, Bishop A, Dunn KM, Main CJ, Thomas E, Foster NE. Conceptual overlap of psychological constructs in low back pain. PAIN 2013;154:1783–91.
[8]. Cavaleri R, Chipchase LS, Massé‐Alarie H, Schabrun SM, Shraim MA, Hodges PW. Corticomotor reorganization during short‐term visuomotor training in the lower back: a randomized controlled study. Brain Behav 2020;10:e01702.
[9]. Chang WJ, Buscemi V, Liston MB, McAuley JH, Hodges PW, Schabrun SM. Sensorimotor cortical activity in acute low back pain: a cross-sectional study. J Pain 2019;20:819–29.
[10]. Cheeran B, Talelli P, Mori F, Koch G, Suppa A, Edwards M, Houlden H, Bhatia K, Greenwood R, Rothwell JC. A common polymorphism in the brain‐derived neurotrophic factor gene (BDNF) modulates human cortical plasticity and the response to rTMS. J Physiol 2008;586:5717–25.
[11]. Chipchase L, Schabrun S, Cohen L, Hodges P, Ridding M, Rothwell J, Taylor J, Ziemann U. A checklist for assessing the methodological quality of studies using transcranial magnetic stimulation to study the motor system: an international consensus study. Clin Neurophysiol 2012;123:1698–704.
[12]. Cirillo J, Hughes J, Ridding M, Thomas PQ, Semmler JG. Differential modulation of motor cortex excitability in BDNF Met allele carriers following experimentally induced and use‐dependent plasticity. Eur J Neurosci 2012;36:2640–9.
[13]. Clarke SM, Henry HM, Dodds KG, Jowett TWD, Manley TR, Anderson RM, McEwan JC. A high throughput single nucleotide polymorphism multiplex assay for parentage assignment in New Zealand sheep. PLoS One 2014;9:e93392.
[14]. Clayson PE, Baldwin SA, Larson MJ. How does noise affect amplitude and latency measurement of event‐related potentials (ERPs)? A methodological critique and simulation study. Psychophysiology 2013;50:174–86.
[15]. Cleeland CS, Ryan K. Pain assessment: global use of the Brief Pain Inventory. Singapore: Annals, Academy of Medicine, 1994.
[16]. Costa LdCM, Maher CG, McAuley JH, Hancock MJ, Herbert RD, Refshauge KM, Henschke N. Prognosis for patients with chronic low back pain: inception cohort study. Bmj 2009;339:b3829.
[17]. de Vet HCW, Heymans MW, Dunn KM, Pope DP, van der Beek AJ, Macfarlane GJ, Bouter LM, Croft PR. Episodes of low back pain: a proposal for uniform definitions to Be used in research. Spine 2002;27:2409–16.
[18]. Deitos A, Dussán-Sarria JA, Souza Ad, Medeiros L, Tarragô MdG, Sehn F, Chassot M, Zanette S, Schwertner A, Fregni F, Torres ILS, Caumo W. Clinical value of serum neuroplasticity mediators in identifying the central sensitivity syndrome in patients with chronic pain with and without structural pathology. Clin J Pain 2015;31:959–67.
[19]. Diers M, Koeppe C, Diesch E, Stolle AM, Hölzl R, Schiltenwolf MM, van Ackern K, Flor H. Central processing of acute muscle pain in chronic low back pain patients: an EEG mapping study. J Clin Neurophysiol 2007;24:76–83.
[20]. Egan MF, Kojima M, Callicott JH, Goldberg TE, Kolachana BS, Bertolino A, Zaitsev E, Gold B, Goldman D, Dean M, Lu B, Weinberger DR. The BDNF val66met polymorphism affects activity-dependent secretion of BDNF and human memory and hippocampal function. Cell 2003;112:257–69.
[21]. Engel CC, Von Korff M, Katon WJJP. Back pain in primary care: predictors of high health-care costs. PAIN 1996;65(2-3):197–204.
[22]. Flor H, Braun C, Elbert T, Birbaumer N. Extensive reorganization of primary somatosensory cortex in chronic back pain patients. Neurosci Lett 1997;224:5–8.
[23]. Foster NE, Mullis R, Hill JC, Lewis M, Whitehurst DGT, Doyle C, Konstantinou K, Main C, Somerville S, Sowden G, Wathall S, Young J, Hay EM. Effect of stratified care for low back pain in family practice (IMPaCT Back): a prospective population-based sequential comparison. Ann Fam Med 2014;12:102–11.
[24]. Glantz S, Slinker B. Primer of Applied Regression & Analysis of Variance. New York: McGraw-Hill, Inc., 2001.
[25]. Hariri AR, Goldberg TE, Mattay VS, Kolachana BS, Callicott JH, Egan MF, Weinberger DR. Brain-derived neurotrophic factor val66met polymorphism affects human memory-related hippocampal activity and predicts memory performance. J Neurosci 2003;23:6690–4.
[26]. Hartvigsen J, Hancock MJ, Kongsted A, Louw Q, Ferreira ML, Genevay S, Hoy D, Karppinen J, Pransky G, Sieper J, Smeets RJ, Underwood M. What low back pain is and why we need to pay attention. Lancet 2018;391:2356–67.
[27]. Hayden JA, Chou R, Hogg-Johnson S, Bombardier C. Systematic reviews of low back pain prognosis had variable methods and results: guidance for future prognosis reviews. J Clin Epidemiol 2009;62:781–96.e78187.
[28]. Henry JD, Crawford JR. The short‐form version of the Depression Anxiety Stress Scales (DASS‐21): construct validity and normative data in a large non‐clinical sample. Br J Clin Psychol 2005;44:227–39.
[29]. Henschke N, Maher CG, Refshauge KM, Herbert RD, Cumming RG, Bleasel J, York J, Das A, McAuley JH. Prognosis in patients with recent onset low back pain in Australian primary care: inception cohort study. BMJ 2008;337:a171.
[30]. Hill JC, Dunn KM, Lewis M, Mullis R, Main CJ, Foster NE, Hay EM. A primary care back pain screening tool: identifying patient subgroups for initial treatment. Arthritis Rheum 2008;59:632–41.
[31]. Hill JC, Vohora K, Dunn KM, Main CJ, Hay EM. Comparing the STarT back screening Toolʼs subgroup Allocation of individual patients with that of independent clinical experts. Clin J Pain 2010;26:783–7.
[32]. Hingorani AD, Windt DAvd, Riley RD, Abrams K, Moons KGM, Steyerberg EW, Schroter S, Sauerbrei W, Altman DG, Hemingway HJB. Prognosis research strategy (PROGRESS) 4: stratified medicine research. BMJ 2013;346:e5793.
[33]. Homan RW, Herman J, Purdy P. Cerebral location of international 10–20 system electrode placement. Electroencephalogr Clin Neurophysiol 1987;66:376–82.
[34]. Jenkins LC, Chang WJ, Buscemi V, Liston M, Toson B, Nicholas M, Graven-Nielsen T, Ridding M, Hodges PW, McAuley JH, Schabrun SM. Do sensorimotor cortex activity, an individual's capacity for neuroplasticity, and psychological features during an episode of acute low back pain predict outcome at 6 months: a protocol for an Australian, multisite prospective, longitudinal cohort study. BMJ Open 2019;9:e029027.
[35]. Jenkins LC, Chang WJ, Buscemi V, Liston M, Skippen P, Cashin AG, McAuley JH, Schabrun SM. Low somatosensory cortex excitability in the acute stage of low back pain causes chronic pain. J Pain 2022;23:289–304.
[36]. Kaneko K, Kawai S, Fuchigami Y, Morita H, Ofuji A. The effect of current direction induced by transcranial magnetic stimulation on the corticospinal excitability in human brain. Electroencephalogr Clin Neurophysiol 1996;101:478–82.
[37]. Kappenman ES, Luck SJ. Best practices for event-related potential research in clinical populations. Biol Psychiatry Cogn Neurosci Neuroimaging 2016;1:110–15.
[38]. Karran EL, McAuley JH, Traeger AC, Hillier SL, Grabherr L, Russek LN, Moseley GL. Can screening instruments accurately determine poor outcome risk in adults with recent onset low back pain? A systematic review and meta-analysis. BMC Med 2017;15:1–15.
[39]. Karran EL, Traeger AC, McAuley JH, Hillier SL, Yau YH, Moseley GL. The value of prognostic screening for patients with low back pain in secondary care. J Pain 2017;18:673–86.
[40]. Keel JC, Smith MJ, Wassermann EM. A safety screening questionnaire for transcranial magnetic stimulation. Clin Neurophysiol 2001;112:720.
[41]. Kent PM, Keating JL. Can we predict poor recovery from recent-onset nonspecific low back pain? A systematic review. Man Ther 2008;13:12–28.
[42]. Kleim JA, Chan S, Pringle E, Schallert K, Procaccio V, Jimenez R, Cramer SC. BDNF val66met polymorphism is associated with modified experience-dependent plasticity in human motor cortex. Nat Neurosci 2006;9:735–7.
[43]. Klyne D, Barbe M, Hoorn W, Hodges P. ISSLS PRIZE IN CLINICAL SCIENCE 2018: longitudinal analysis of inflammatory, psychological, and sleep-related factors following an acute low back pain episode --the good, the bad, and the ugly. Eur Spine J 2018;27: 763–77.
[44]. Kovacs FM, Abraira V, Zamora J, Teresa Gil del Real M, Llobera J, Fernández C, Bauza JR, Bauza K, Coll J, Cuadri M, Duro E, Gili J, Gestoso M, Gomez M, Gonzalez J, Ibanez P, Jover A, Lazaro P, Llinas M, Mateu C, Mufraggi N, Muriel A, Nicolau C, Olivera MA, Pascual P, Perello L, Pozo F, Revuelta T, Reyes V, Ribot S, Ripoll J, Ripoll J, Rodriguez E. Correlation between pain, disability, and quality of life in patients with common low back pain. Spine 2004;29:206–10.
[45]. Lang UE, Hellweg R, Kalus P, Bajbouj M, Lenzen KP, Sander T, Kunz D, Gallinat J. Association of a functional BDNF polymorphism and anxiety-related personality traits. Psychopharmacology 2005;180:95–9.
[46]. Lang UE, Hellweg R, Sander T, Gallinat J. The Met allele of the BDNF Val66Met polymorphism is associated with increased BDNF serum concentrations. Mol Psychiatry 2009;14:120–2.
[47]. Lariviere C, Arsenault AB, Gravel D, Gagnon D, Loisel P. Surface electromyography assessment of back muscle intrinsic properties. J Electromyogr Kinesiol 2003;13:305–18.
[48]. Laske C, Stransky E, Eschweiler GW, Klein R, Wittorf A, Leyhe T, Richartz E, Köhler N, Bartels M, Buchkremer G, Schott K. Increased BDNF serum concentration in fibromyalgia with or without depression or antidepressants. J Psychiatr Res 2007;41:600–5.
[49]. Lee H, Hübscher M, Moseley GL, Kamper SJ, Traeger AC, Mansell G, McAuley JH. How does pain lead to disability? A systematic review and meta-analysis of mediation studies in people with back and neck pain. PAIN 2015;156:988–97.
[50]. Linton SJ, Boersma K. Early identification of patients at risk of developing a persistent back problem: the predictive validity of the Örebro Musculoskeletal Pain Questionnaire. Clin J Pain 2003;19:80–6.
[51]. Linton SJ, Nicholas M, Shaw WJP. Why wait to address high-risk cases of acute low back pain? PAIN 2018;159:2437–41.
[52]. Linton SJ. A review of psychological risk factors in back and neck pain. Spine 2000;25:1148–56.
[53]. Mansour AR, Baliki MN, Huang L, Torbey S, Herrmann KM, Schnitzer TJ, Apkarian AV. Brain white matter structural properties predict transition to chronic pain. PAIN 2013;154:2160–8.
[54]. Mata J, Thompson RJ, Gotlib IH. BDNF genotype moderates the relation between physical activity and depressive symptoms. Health Psychol 2010;29:130–3.
[55]. McBeth J, Silman AJ, Gupta A, Chiu YH, Ray D, Morriss R, Dickens C, King Y, Macfarlane GJA. Moderation of psychosocial risk factors through dysfunction of the hypothalamic–pituitary–adrenal stress axis in the onset of chronic widespread musculoskeletal pain: findings of a population‐based prospective cohort study. Arthritis Rheum, 2007;56:360–71.
[56]. Mills KR, Boniface SJ, Schubert M. Magnetic brain stimulation with a double coil: the importance of coil orientation. Electroencephalogr Clin Neurophysiol 1992;85:17–21.
[57]. Moons KGM, Altman DG, Reitsma JB, Ioannidis JPA, Macaskill P, Steyerberg EW, Vickers AJ, Ransohoff DF, Collins GS. Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD): explanation and elaboration. Ann Intern Med 2015;162:W1–W73.
[58]. Moons KGM, Wolff RF, Riley RD, Whiting PF, Westwood M, Collins GS, Reitsma JB, Kleijnen J, Mallett S. PROBAST: a tool to assess risk of bias and applicability of prediction model studies: explanation and elaboration. Ann Intern Med 2019;170:1–33.
[59]. Müller M, Curatolo M, Limacher A, Neziri AY, Treichel F, Battaglia M, Arendt-Nielsen L, Jüni P. Predicting transition from acute to chronic low back pain with quantitative sensory tests—a prospective cohort study in the primary care setting. Eur J Pain 2019;23:894–907.
[60]. Nicholas MK. The pain self-efficacy questionnaire: taking pain into account. Eur J Pain 2007;11:153–63.
[61]. O'Connell NE, Maskill DW, Cossar J, Nowicky AV. Mapping the cortical representation of the lumbar paravertebral muscles. Clin Neurophysiol 2007;118:2451–5.
[62]. Oliveira VC, Ferreira PH, Maher CG, Pinto RZ, Refshauge KM, Ferreira ML. Effectiveness of self‐management of low back pain: systematic review with meta‐analysis. Arthritis Care Res 2012;64:1739–48.
[63]. Patel S, Friede T, Froud R, Evans DW, Underwood MJS. Systematic review of randomized controlled trials of clinical prediction rules for physical therapy in low back pain. Spine, 2013;38:762–9.
[64]. Pereira LM, Obara K, Dias JM, Menacho MO, Guariglia DA, Schiavoni D, Pereira HM, Cardoso JR. Comparing the Pilates method with no exercise or lumbar stabilization for pain and functionality in patients with chronic low back pain: systematic review and meta-analysis. Clin Rehabil 2012;26:10–20.
[65]. Pezawas L, Verchinski BA, Mattay VS, Callicott JH, Kolachana BS, Straub RE, Egan MF, Meyer-Lindenberg A, Weinberger DR. The brain-derived neurotrophic factor val66met polymorphism and variation in human cortical morphology. J Neurosci 2004;24:10099–102.
[66]. Picton TW, Bentin S, Berg P, Donchin E, Hillyard SA, Johnson R Jr, Miller GA, Ritter W, Ruchkin DS, Rugg MD, Taylor MJ. Guidelines for using human event‐related potentials to study cognition: recording standards and publication criteria. Psychophysiology 2000;37:127–52.
[67]. Pincus T, Burton AK, Vogel S, Field AP. A systematic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain. Spine 2002;27:E109–20.
[68]. R Core TeamR: A language and environment for statistical computing. Vienna: R Foundation for statistical computing, 2013.
[69]. Riley RD, Hayden JA, Steyerberg EW, Moons KGM, Abrams K, Kyzas PA, Malats N, Briggs A, Schroter S, Altman DG, Hemingway H. Prognosis Research Strategy (PROGRESS) 2: prognostic factor research. PLoS Med 2013;10:e1001380.
[70]. Roland M, Morris R. A study of the natural history of back pain: Part I: development of a reliable and sensitive measure of disability in low-back pain. Spine 1983;8:141–4.
[71]. Rubinstein SM, van Middelkoop M, Assendelft WJJ, de Boer MR, van Tulder MW. Spinal manipulative therapy for chronic low-back pain: an update of a Cochrane review. Spine 2011;36:E825–46.
[72]. Schabrun SM, Jones E, Elgueta Cancino EL, Hodges PW. Targeting chronic recurrent low back pain from the top-down and the bottom-up: a combined transcranial direct current stimulation and peripheral electrical stimulation intervention. Brain Stimul 2014;7:451–9.
[73]. Schabrun SM, Burns E, Hodges PW. New insight into the time-course of motor and sensory system changes in pain. PLoS One 2015;10, e0142857.
[74]. Schabrun SM, Elgueta-Cancino EL, Hodges PW. Smudging of the motor cortex is related to the severity of low back pain. Spine 2017;42:1172–8.
[75]. Stanton TR, Henschke N, Maher CG, Refshauge KM, Latimer J, McAuley JH. After an episode of acute low back pain, recurrence is unpredictable and not as common as previously thought. Spine 2008;33:2923–8.
[76]. Stanton TR, Hancock MJ, Maher CG, Koes BWJPT. Critical appraisal of clinical prediction rules that aim to optimize treatment selection for musculoskeletal conditions. Phys Ther 2010;90:843–54.
[77]. Steyerberg EW, Moons KGM, van der Windt DA, Hayden JA, Perel P, Schroter S, Riley RD, Hemingway H, Altman DG, Group P. Prognosis Research Strategy (PROGRESS) 3: prognostic model research. PLoS Med 2013;10:e1001381.
[78]. Sullivan M, Bishop S, Pivik J. The pain catastrophizing scale: development and validation. Psychol Assess 1995;7:524.
[79]. Tibshirani R. Regression shrinkage and selection via the lasso. J R Stat Soc Ser B 1996;58:267–88.
[80]. Traeger AC, Henschke N, Hübscher M, Williams CM, Kamper SJ, Maher CG, Moseley GL, McAuley JH. Estimating the risk of chronic pain: development and validation of a prognostic model (PICKUP) for patients with acute low back pain. PLoS Med 2016;13:e1002019.
[81]. Tsao H, Galea MP, Hodges PW. Driving plasticity in the motor cortex in recurrent low back pain. Eur J Pain 2010;14:832–9.
[82]. Tsao H, Danneels LA, Hodges PW. ISSLS prize winner: smudging the motor brain in young adults with recurrent low back pain. Spine 2011;36:1721–7.
[83]. Uy J, Ridding MC, Miles TS. Stability of maps of human motor cortex made with transcranial magnetic stimulation. Brain Topogr 2002;14:293–7.
[84]. van Buuren S, Groothuis-Oudshoorn K. Mice: multivariate imputation by chained equations in R. J Stat Softw 2011;45:67.
[85]. van de Ruit M, Grey MJ. Interindividual variability in use-dependent plasticity following visuomotor learning: the effect of handedness and muscle trained. J Mot Behav 2019;51:171–84.
[86]. Van de Ruit ML. Rapid assessment of corticospinal excitability using transcranial magnetic stimulation. University of Birmingham, 2016.
[87]. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007;165:710–18.
[88]. Vos T, Abajobir AA, Abate KH, Abbafati C, Abbas KM, Abd-Allah F, Abdulkader RS, Abdulle AM, Abebo TA, Abera SF. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet 2017;390:1211–59.
[89]. Wassermann EM, McShane LM, Hallett M, Cohen LG. Noninvasive mapping of muscle representations in human motor cortex. Electroencephalogr Clin Neurophysiol 1992;85:1–8.
[90]. Wei SY, Chao HT, Tu CH, Lin MW, Wei-Chi L, Low I, Intan L, Shen HD, Horng-Der S, Chen LF, Li-Fen C, Hsieh JC, Jen-Chuen H. The BDNF Val66Met polymorphism is associated with the functional connectivity dynamics of pain modulatory systems in primary dysmenorrhea. Sci Rep 2016;6, 23639.
[91]. William & Mary Alumni Association. WMA Declaration of Helsinki-Ethical principles for medical research involving human subjects. Brazil: Fortaleza, 2013.
[92]. Williams CM, Maher CG, Latimer J, McLachlan AJ, Hancock MJ, Day RO, Lin C-WC. Efficacy of paracetamol for acute low-back pain: a double-blind, randomised controlled trial. Lancet 2014;384:1586–96.
[93]. Woodman GF. A brief introduction to the use of event-related potentials in studies of perception and attention. Atten Percept Psychophys 2010;72:2031–46.

Low back pain, Prognostic factors, Sensorimotor cortex excitability, Plasticity, Brain-derived neurotrophic factor, Psychological, Sensory-evoked potential, Transcranial magnetic stimulation

Supplemental Digital Content

© 2022 International Association for the Study of Pain