Low-back pain (LBP) is the single leading cause for disability worldwide, affects all age groups and has increased from 58 million years lived with disability (YLDs) in 1990 to 83 million YLDs in 2010. The burden is accordingly substantially higher than previously assessed, causing activity limitation and work absence with subsequently enormous economic burden. Norwegian expenses reach at least NOK 24 billions annually whereof a substantial part is hospital costs. The research project responds to this challenge and aim to conduct a multicenter randomized placebo-controlled trial, complemented by a study of epigenetic and molecular biomarkers, to re-examine the finding of a recent randomized controlled trial that antibiotic treatment can cure patients with chronic low back pain (LBP), a former disc herniation and present Modic Changes (MCs). The hypothesis is that MCs is caused by low virulent anaerobic organisms in the disc. Investigators also want to add important new knowledge to the research field beyond the only former RCT by broadening the inclusion criteria to include both patients with type I and type II MCs, improving the MRI assessment of MCs, further clarifying the pathogenesis of MCs by studying genetic variability, gene and protein expression of inflammatory biomarkers, and conducting health economic analysis.

Pre-defined research questions / hypothesis and endpoints:

Main objective and endpoint: To evaluate the effect of Amoxicillin versus placebo on disease-specific disability evaluated by the Roland Morris Disability Questionnaire (RMDQ) at one year (12 months) follow-up in patients with chronic LBP and MCs type I or II adjacent to a previously herniated disc (Hypothesis A).

Thus, the projects Main objective is to re-examine the clinical effect of antibiotic treatment reported in the former Danish study at one year (12 months) follow-up. Investigators will use the same primary outcome measure (RMDQ), but the effect will be evaluated in patients with MCs type I or MCs type II since, as outlined above, investigators argue that some MCs type I patients may be classified as MCs type II patients and vice versa dependent on the magnet strength of MRI machines used, and MCs type I and type II most likely represent a common process (that can be influenced by a common treatment) (hypothesis A). As a secondary objective (SO 1) investigators will evaluate the effect of Amoxicillin versus placebo on RMDQ at one year (12 months) follow-up separately in patients with type I and type II MCs, respectively (hypotheses B and C).

Exploratory and key supportive objectives (KSOs) and endpoints):

• KSO 2. To evaluate the effect of Amoxicillin versus placebo on Oswestry Disability Index (ODI) at one year (12 months) follow-up in the whole cohort of included patients (hypothesis D).

• KSO 3. To evaluate the effect of Amoxicillin versus placebo on LBP intensity at one year (12 months) follow-up in the whole cohort of included patients (hypothesis E).

• KSO 4. To evaluate whether the short tau inversion recovery (STIR) signal (intensity and extent) of MCs on baseline MRI predicts change in RMDQ from baseline to one year (12 months) follow-up (hypothesis F).

• KSO 5. To assess whether change in STIR signal (intensity and extent) of MCs at one year (12 months) follow-up is related to change in RMDQ from baseline to one year (12 months) follow-up (hypothesis G).

• KSO 6. To evaluate the effect of Amoxicillin versus placebo on health-related quality of life (the EQ-5D) at one year (12 months) follow-up in the whole cohort of included patients (hypothesis H).

• To evaluate cost-effectiveness of Amoxicillin versus placebo at one year (12 months) follow-up in the whole cohort of included patients.

• To evaluate whether positive pain provocation tests at baseline predicts change in RMDQ at one year (12 months) follow-up.

• To evaluate the difference in incidence of AEs and SAEs between the two intervention groups from inclusion to one year (12 months) follow-up in the whole cohort of included patients.

• To evaluate, separately in the two intervention groups, whether lack of a clinically important improvement in RMDQ, ODI, and LBP intensity, respectively, from baseline to post-treatment (100-das after start of treatment) is associated with lack of a clinically important improvement in these outcomes from baseline to one-year (12 months) follow-up.

• Further clinical objectives and endpoints: To evaluate the effect of Amoxicillin versus placebo on:

  • RMDQ at one year (12 months) follow-up separately in patients previously undergoing back surgery for disc herniation and patients NOT previously undergoing back surgery for disc herniation.
  • other outcome measures not mentioned above (leg pain intensity, hours with LBP during the last 4 weeks, bothersomeness, days with sick leave, co-interventions, patients' satisfaction, global perceived effect) at one year (12 months) follow-up in the whole cohort of included patients
  • secondary outcome measures at one year (12 months) follow-up separately in patients with type I and type II MCs at baseline, respectively
  • primary and secondary outcomes measures posttreatment (100 days after start of treatment) in the whole cohort and separately in patients with type I and type II MCs at baseline, respectively.

• Further radiological objectives and endpoints: To assess whether characteristics of MCs on baseline MRI predict change in ODI, or change in intensity of LBP from baseline to one year (12 months) follow-up. To compare change in characteristics of MCs from inclusion to one year (12 months) follow-up between treatment groups, and to assess whether this change in MCs is related to change in RMDQ, ODI, and pain intensity from baseline to one year (12 months) follow-up. To determine the reliability of different MCs characteristics by different MRI methods. To assess the relationships of these MC characteristics to each other and to clinical variables.

Genetic objectives and endpoints: To investigate the effect of Amoxicillin on epigenetic patterns, longitudinal gene- and protein expression, genetic variation, from baseline to post-treatment (100 days after start of treatment) and from baseline to one year (12 months) follow-up in patients with MCs type I or II, and to evaluate correlations with clinical data.

Randomization lists will be created using STATA 13 (StatCorp LP, College Station, TX, USA) and will be stratified by MODIC type (1/2) and previous surgery (yes/no) with a 1:1:1:1 allocation and random block sizes of 4 and 6. This will ensure similar numbers of patients receiving antibiotics or placebo within each stratum (1-MCs type I, no previous back surgery for disc herniation, 2-MCs type I, previous back surgery for disc herniation, 3-MCs type II, no previous back surgery for disc herniation, and 4-MCs type II, previous back surgery for disc herniation). Investigators intend to recruit two equally large patient groups, one with type I MCs and one with type II MCs, in order to evaluate treatment effect separately in each MC type group and not only in the total sample. Investigators will achieve this by stratified randomization as explained above, and investigators will stop the inclusion of the two MCs type II strata (with and without previous surgery) when investigators have enough MCs type II patients, and continue the inclusion of the two MCs type I strata (or vice versa). Patients are stratified for previous back surgery for disc herniation (Yes/No) since it is not clear how the low virulent anaerobic organisms gain access to the disc (during normal bacteraemia or as a result of intraoperative contamination). Hence, stratifying for previous back surgery for disc herniation will ensure balanced distribution of this potential source of infection between groups.

Investigators have designed the study with enough power to evaluate the treatment effect separately in the MC type I group and in the MC type II group (investigators will include equally many Modic I and Modic II patients in the trial), but the main analysis will be in the whole cohort of patients with MCs type I or II. In power calculations for each MC type group, investigators used a two-sided alfa of 0.05 and a power of 0.90 and wished to be able to identify a difference in mean RMDQ of 4 (SD 5) between the two treatment groups at 12 months follow-up. The minimal important difference in mean RMDQ between groups is not clear, but a change in RMDQ of 2-3 in individual patients over time is very unlikely to be important and may represent measurement error. Investigators therefore used a difference of 4 in mean RMDQ between groups in the power calculations. The SD of 5 is within the upper range of commonly reported SDs for RMDQ in patients with persistent LBP.

A to-sided alfa 0.05 (i.e. 95 % Confidence Interval), a power 90, ratio of sample size 1 (antibiotic group / placebo group) and a difference in RMDQ 4 (SD 5) results in a sample size of 33 in each treatment group, or 66 in both treatment groups; i.e. 132 in the total sample (both MC type groups). The Danish RCT by Albert et al 2013 had 11 % dropouts. Adding 20 % for dropouts (26 patients) investigators calculated that they need to include 158 patients in the study, rounded up to 160 patients: 80 patients with type I MCs and 80 patients with type II MCs. The study continue inclusion until 80 patients are included in the MC type group that is slowes to recruit, implying inclusion of at least 80 patients in the other MC type group and at least 160 patients in total.

Data on main outcome will be analysed by a statistician who is blinded to group status. The primary analyses (main outcome) will be by intention-to-treat in the whole cohort of patients with MCs type I or II using an ANCOVA analysis adjusted for baseline RMDQ-score. The significance level will be 0.05. Investigators will report the p value with 95 % CI. ANCOVA analysis, adjusted for baseline RMDQ-score, will also be used to examine main outcome separately in the sub-samples of type I and II MSc. The significance level will still be 0.05. Investigators will report the exact p value with 95 % CI for each of the two sub-samples. For the secondary outcome measures also ANCOVA analysis will be used with adjustment for baseline values and in accordance to intention to treat (ITT). The analysis will be carried out in the whole cohort of patients with MCs type I or II with a significance level of 0.05, and separately in the two sub-samples of type I and II MSc. Investigators will report the p value with 95 % CI. Analyses of MRI results will include observer agreement analyses (kappa, McNemars test, Bland Altmann plots) and multiple regression analyses to assess relationships with clinical variables.

As secondary analysis, investigators will perform responder analyses of RMDQ to supplement the interpretability of the main analysis. They will analyze if a higher proportion of patients with MCs type I or II at baseline reports a clinically relevant improvement of their RMDQ-score from baseline to one year (12 months) follow-up in the antibiotic treatment group than in the placebo group. Analysis (chi-square test) will be performed for three different cut-offs (>75%, >50% and >30% improvement of RMDQ score from baseline to one year follow-up, respectively). Intention-to-treat (ITT) principles will be used and the significance level will be 0.05. Investigators will report the exact p value and the number needed to treat (NNT) with 95 % CI.

Analyses of MRI results will include observer agreement analyses (kappa, McNemars test, Bland Altmann plots) and multiple regression analyses to assess relationships with clinical variables. Cost-effectiveness will be analyzed as the difference in costs between the 2 treatment groups divided by their difference in QALYs gained. The results will be presented as an incremental cost-effectiveness ratio (ICER).

In addition to outcome measured listed, the following will be evaluated:

• To check Blinding: Patients will be asked which study medicine (Amoxicillin / placebo / unsure) they think they received. Points of time: post-treatment (100 days after start of intervention), and one year (12 months) after start of intervention.
• To evaluate Compliance to the medicine protocol; Patients will be asked every week during the intervention period how many days the last week they took the study medication (0-7)). At post-treament (100 days after start of treatment); Capsule count will be performed by a pharmacist.
• Adverse events will be Registered and CTC-coded at every study visit (=monthly) during the intervention period by the principal investigators.

EudraCT no: 2013-004505-14

Patients from all health regions in Norway referred to the participating hospitals will be screened for eligibility. Both conservatively and surgically treated patients (i.e. operated on for disc herniation > 12 months prior to inclusion) will be included. In addition, patients registered in the Norwegian Registry for Spine Surgery operated on for disc herniation and reporting severe LBP pain at one-year follow-up in the registry, will be invited.

Inclusion Criteria

• Age between 18 and 65 years
• LBP of > 6 months duration in the area below the 12th rib and above the gluteal folds with a Numerical Rating Scale (NRS) pain intensity score of ³ 5 (mean of three NRS scales; current LBP, the worst LBP within the last 2 weeks, and usual/mean LBP within the last 2 weeks).
• MRI-confirmed lumbar disc herniation within the preceding 2 years.
• MC type I and/or type II in the vertebral body marrow at the same level as the previously herniated disc. For patients with former surgery for disc herniation, the MC has to be located at an operated level.
• Written informed consent

Exclusion Criteria:

• Allergy to penicillin or cefalosporins
• Allergy/hypersensitivity to any of the excipients of the study drug
• Current pregnancy or lactation
• Elevated kidney (creatinine) or hepatic (ALAT/ASAT) values outside normal range
• Phenylketonuria (Følling disease)
• Mononucleosis or leukaemia
• Any specific diagnosis that may explain patient's low back symptoms (e.g. tumor, fracture, spondyloarthritis, infection, spinal stenosis).
• Former low back surgery (L1 - S1) for other reasons than disc herniation (e.g fusion, decompression, disc prosthesis).
• Former surgery for disc herniation, but < 12 months have elapsed since surgery.
• Former surgery for disc herniation, but MC located at non-operated level(s) only.
• Reservation against intake of gelatine (the capsules contains gelatine, which among other things is produced by ingredients from pigs)
• Regular use of glucocorticoids
• Regular use of opioids with the exception of codeine and tramadol
• Not understanding Norwegian language
• Unlikely to adhere to treatment and/ or complete follow-up (e.g ongoing serious psychiatric disease, drug abuse, plans to move)
• Antibiotic treatment within the preceding one month before treatment start
• Contraindications to MRI (e.g. cardiac pacemaker electrodes, metal implant in eye or brain, claustrophobia).
• Unwilling to participate