The efficacy and safety of oral antibiotic treatment in patients with chronic low back pain and Modic changes: A systematic review and meta‐analysis

Abstract Background This systematic review and meta‐analysis aimed to summarize evidence regarding the effectiveness and safety of oral antibiotic intervention for chronic low back pain (CLBP) patients with/without type‐1 Modic changes (MC1). Methods AMED, CINAHL, Cochrane Library, Embase, and Medline were searched from inception to March 3, 2023. Randomized controlled trials (RCTs) or non‐RCTs that investigated the effectiveness or safety of oral antibiotics in treating CLBP patients were eligible for inclusion. Two independent reviewers screened abstracts, full‐text articles, and extracted data. The methodological quality of each included article were evaluated by RoB2 and NIH quality assessment tools. The quality of evidence was appraised by GRADE. Meta‐analyses were performed, where applicable. A subgroup analysis was conducted to evaluate the RCTs and case series separately, and to evaluate the effect of removing a low‐quality RCT. Results Three RCTs and four case series were included. All Amoxicillin‐clavulanate/Amoxicillin treatments lasted for approximately 3 months. Moderate‐ and low‐quality evidence suggested that antibiotic was significantly better than placebo in improving disability and quality of life in CLBP patients with MC1 at 12‐month follow‐up, respectively. Low‐quality evidence from meta‐analyses of RCTs showed that oral antibiotic was significantly better than placebo in improving pain and disability in CLBP patients with MC1 immediately post‐treatment. Very low‐quality evidence from the case series suggested that oral Amoxicillin‐clavulanate significantly improved LBP/leg pain, and LBP‐related disability. Conversely, low‐quality evidence found that oral Amoxicillin alone was not significantly better than placebo in improving global perceived health in patients with CLBP at the 12‐month follow‐up. Additionally, oral antibiotic users had significantly more adverse effects than placebo users. Conclusions Although oral antibiotics were statistically superior to placebo in reducing LBP‐related disability in patients with CLBP and concomitant MC1, its clinical significance remains uncertain. Future large‐scale high‐quality RCTs are warranted to validate the effectiveness of antibiotics in individuals with CLBP.

Low back pain (LBP) is the most disabling condition worldwide. 1Many efforts have been made to manage this growing global healthcare problem and associated socioeconomic consequences. 2Notably, chronic LBP (CLBP) and the treatment thereof have demonstrated a substantial challenge to modern healthcare providers. 3Pharmacological, conservative, and surgical interventions for LBP have their own disadvantages, although their clinical benefits are not fully satisfactory. 3For instance, the common use of opioid for treating patients with CLBP may lead to the opioid pandemic, which highlights the importance of better understanding the condition and developing alternative management strategies. 4ltiple factors can lead to LBP. [5][6][7] Knowledge of specific pain generators is needed to improve individualized treatments for patients with LBP.Of various spine phenotypes, Modic changes (MCs) could be a potential target for more personalized treatment for patients with CLBP.MCs are vertebral bone marrow changes visualized by magnetic resonance imaging (MRI). 8MCs were first described by de Roos et al. 9 on MRI in 1987.Modic et al. 8 then classified MCs into three categories based on T1-and T2-weighted MRI and histological analysis.MC type 1 (MC1) represents reactive/inflammatory changes in the marrow; MC2 consists of yellow lipid marrow replacement; and MC3 is a feature of calcification within the endplate and subchondral vertebral marrow. 8MC1 and to a lesser extent MC2 has been associated with both LBP and disability. 10,11s are common among patients with LBP.The prevalence of MCs in patients with lumbar disc herniation and CLBP was up to 45%, while that in the general adult population was 5%. 12,13Populationand patient-based studies have demonstrated a close association between MCs and LBP. 11,14Recently, the International Classification of Diseases 10th revision (ICD-10) and the ICD-11 have included a specific diagnosis for patients with LBP and concomitant endplate/ MCs phenotype-vertebrogenic LBP. 15 This underscores the importance of the endplate and marrow changes in LBP with vertebrogenic origins.Accordingly, advances in the treatment of LBP patients with MCs could have major impacts on back pain and the associated global health burden. 16e etiology of MCs is linked with pathophysiological changes in the endplate and discus causing biomechanical changes in the functional spinal unit (FSU). 58][19][20][21][22] Of various potential causes, the hypothesis of bacterial infection, especially by Cutibaterium acnes (C acnes), has been drawing a lot of attention.C acnes has been found in disc tissues and speculated as a culprit caused by hematogenous spread to the discus/vertebra complex. 23Such findings and microbiome analysis illustrate the potential cause of LBP by infection. 24Unlike bacterial disc infections (e.g., spondylodiscitis), which are characterized by an acute systemic infection, 25 an ongoing low virulent infection in the discus/vertebral complex without systemic manifestation may cause symptomatic disc degeneration. 26It may also partly explain the development of intervertebral edema with upregulation of a localized inflammatory response in the adjacent vertebra. 5,24,27[30] Findings from these studies have caused considerable debate regarding management of CLBP. 31While oral antibiotics may significantly improve LBP and disability in patients with CLBP and MCs, their improvements may not be clinically significant or safe.To date, no relevant systematic reviews have been conducted to summarize the evidence, and these important questions remain unanswered.Therefore, this systematic review and meta-analysis aimed to summarize the evidence regarding the effectiveness and safety of oral antibiotic use for patients with CLBP who exhibit MCs on MRI.

| MATERIALS AND METHODS
The study was conducted and reported according to the Preferred Reporting Items of Systematic Reviews and Meta-analyses (PRISMA) guidelines.
The protocol was registered on PROSPERO (CRD42021219667).A5.The reference lists of all included studies were screened for relevant articles.The corresponding authors of the included studies were contacted by email to identify additional relevant publications or relevant data to facilitate meta-analyses.Forward citation tracing was conducted using Scopus.

| Selection criteria
Studies were included if they investigated the effectiveness of antibiotics in patients with CLBP (aged 18 years or older) with or without disc degeneration-related surgery.The definition of CLBP might vary slightly across studies.Both RCTs and non-RCTs studies were eligible for inclusion to obtain a comprehensive overview of this topic.MRIconfirmed lumbar-related MCs should be reported in the included studies.Studies were excluded if they involved acute hematogenous osteomyelitis, or spondylodiscitis related to Staphylococcus aureus (e.g., methicillin-susceptible Staphylococcus aureus bacteriemia, methicillin-resistant Staphylococcus aureus bacteriemia), or high virulent infection.Additionally, case reports, letters to the editor, conference proceedings, and commentaries were excluded.

| Study selection
Citations identified from databases were exported to EndNote X9.2 (Thomson Reuters).After removing duplicates, two independent reviewers (G.M. and A.W.) independently screened the titles and abstracts of potential citations according to the selection criteria.
Between-reviewers' disagreements were discussed to obtain consensus.Any remaining disagreement was resolved by a third reviewer (D.S.).Relevant systematic/literature reviews were included for fulltext reading to identify additional primary studies for screening.The full-text screening procedure was identical to the abstract screening.

| Data extraction
Relevant data from the included studies were independently extracted and counterchecked by the two reviewers (G.M. and A.W.).The extracted data included: authors' names, year of publication, study design, study location, settings, study objectives, response and/or attrition rates, the randomization method (if applicable), participants' selection criteria, definitions of CLBP, types of medical imaging, types of MCs, oral antibiotics and/or other control treatments, follow-up duration and numbers, primary and secondary outcomes, relevant statistics, reported clinical significance, and side effects.

| Risk of bias assessments
Two independent reviewers (S.P. and A.W.) evaluated the risk of bias in the included studies.Specifically, the risk of bias of the included RCTs was evaluated by the Risk of Bias 2 (RoB 2) tool.Because only relevant case series were identified, the included case series was evaluated by the National Institutes of Health quality assessment tool for before-after studies with no control group.The two reviewers' assessments were compared.Any discrepancies were resolved by consensus.

| Data synthesis
If the primary or secondary outcome measures were comparable between two or more included studies, relevant data were pooled for meta-analyses using RevMan v.5.3 (The Cochrane Collaboration, Software Update, Oxford, UK).If the between-group differences in clinical outcomes of a primary study were reported as medians and interquartile ranges, these values were converted to means and standard deviations using the McGrath et al. method to enable meta-analyses. 32The pooled between-group differences in outcome measures were reported as mean differences (MDs) and 95% confidence intervals (95% CIs).The forest plot and I 2 statistics of each pooled outcome measure were reported.Random effects models were used for metaanalyses with I 2 > 50% to account for differences across heterogeneous studies. 33,34If I 2 < 50%, fixed effects models were used.If there were ≥10 studies in a given meta-analysis, the publication bias was assessed by a funnel plot. 35The alpha level was set at 0.05 for all meta-analyses.If a given clinical outcome could not be meta-analyzed, the finding was summarized narratively.Sensitivity analyses were conducted to evaluate the results after eliminating low-quality studies.

| The GRADE evaluation
Two authors (S.P. and A.W.) independently assessed the quality of evidence regarding the effect of oral antibiotics on each reported clinical outcome using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) http://gradepro.org.The quality of evidence was rated as high, moderate, low, or very low.

| RESULTS
The systematic searches and forward citations yielded 183 and 240 citations, respectively.After removing duplicates, 408 titles and abstracts were screened.Fourteen were included for full-text screening.Seven full-text articles (three RCTs 28,29,36 and four case series [37][38][39][40] ) involving 1508 people with CLBP and concomitant MCs were included (Figure 1).These studies were conducted in Canada, Denmark, France, Iraq, and Norway.Five included cohorts involved patients with CLBP and MC1, while two included cohorts involved patients with CLBP and MC1 and MC2, and/or mixed MCs. 29,37Amoxicillin (750 mg) three times per day for 90 days and Amoxicillin-clavulanate (500 mg/125 mg) three times per day for 100 days were used in two and five included studies, respectively, while calcium carbonate or maize starch was used in the placebo pills of the included RCTs (Table 1).All included studies were followed up immediately post-treatment, or at 11, 12, or 24 months after the treatment commencement.Because no meta-analysis involved 10 or more articles, no publication bias was analyzed.

| Risk of bias assessments
Two of the three included RCTs had a low risk of bias (Figure 2).
Although one RCT had a low risk of bias, its statistical analysis plan was not described in the research protocol on a clinical trial registry. 28e RCT with a high risk of bias had an unclear randomization procedure, no information regarding the blinding of clinicians or assessors, high attrition rate (>38%), no information about whether the statistical analyses followed the predetermined plan (Figure 2). 36The four case series had fair or poor quality (Table 2).Their common sources of bias were unclear eligibility criteria, no sample size calculation, unclear blinding of assessors, and unknown reasons for loss to follow-up.

| Between-group comparisons in pain reduction among patients with CLBP and/or leg pain
Two meta-analyses were conducted to evaluate the effectiveness of oral antibiotics in improving LBP in people with CLBP and MC1 immediately after treatment or at the 12-month follow-up (Figure 3).Low-quality evidence from two RCTs supported that Amoxicillinclavulanate was significantly better than placebo pills in reducing LBP immediately after 100 days of treatments in CLBP patients with MC1 (MD: À1.14; 95% CI: À1.58 to À0.70;I 2 = 0%; Figure 3), which was not clinically significant according to the reported minimal clinically important difference for LBP intensity. 41Low-quality evidence from another meta-analysis substantiated that oral antibiotic was not significantly better than placebo in reducing LBP in patients with CLBP at the 12-month follow-up (Figure 3).However, low-quality evidence from a RCT suggested that compared to CLBP patients with MC1 in the antibiotic group, a significantly greater proportion of patients in the placebo group had constant back pain, LBP-related sleep disturbance, positive cranial compression test, and pain during lumbar flexion or extension, or Valsalva maneuver at the 12-month follow-up (Table 3, Table A2). 28The same study found that patients in the antibiotic group had significantly less hours with LBP in the last 28 days than those in the placebo group at the 12-month follow-up. 28As expected, low-quality evidence showed that Amoxicillin was not significantly better than placebo in improving LBP in patients with MC2 alone, or mixed MCs immediately post-treatment or at the 12-month follow-up (Table A2).
Regarding leg pain, low-quality evidence supported that Amoxicillin-clavulanate was significantly better than placebo in improving leg pain in CLBP patients with MC1, or both MC1 and MC2 at the 12-month follow-up.However, only very low-quality evidence supported that patients with MC1 had clinically significantly better LBP reduction than placebo immediately post-treatment (Table 3, Table A2).
Likewise, there was moderate evidence that 90 or 100 days of oral antibiotic was significantly better than placebo in lowering RMDQ scores in CLBP patients with MC1 at the 12-month follow-up (MD: À0.71; 95% CI: À6.86 to À0.55;I 2 = 77%).Low-quality evidence from one included RCT also showed that patients with CLBP and MC1 taking antibiotics had significantly better improvements in ODI scores than placebo controls at the 12-month follow-up (Table 3, Table A2).Similar post-treatment findings were observed in a mixed cohort of patients with MC1 and MC2.Interestingly, one included RCT found that compared to the placebo group, a significantly higher percentage of a patient cohort with MC1 and MC2 in the intervention group displayed clinically significant improvement in disability (>30% reduction in baseline RMDQ scores) at the 12-month follow-up. 29

| Between-group comparisons of bothersome by LBP and global perceived health
Low-quality evidence suggested that CLBP patients with MCI in the antibiotic group experienced significantly less LBP-related bothersome and significantly better perceived general improvement than placebo controls at the 12-month follow-up (Table 3, Table A2).Low-quality evidence supported no significant between-group difference in global post-treatment improvement in a mixed cohort of CLBP patients with MC1 and MC2.
F I G U R E 2 Risk of bias assessments of the included randomized controlled trials."+" = low risk of bias; "-" = high risk of bias; "?" = some concerns.
T A B L E 2 NIH quality assessment for before-after (pre-post) studies with no control group.Note: Q1: Was the study question or objective clearly stated?Q2: Were eligibility/selection criteria for the study population prespecified and clearly described?Q3: Were the participants in the study representative of those who would be eligible for the test/service/intervention in the general or clinical population of interest?Q4: Were all eligible participants that met the prespecified entry criteria enrolled?Q5: Was the sample size sufficiently large to provide confidence in the findings?Q6: Was the test/service/intervention clearly described and delivered consistently across the study population?Q7: Were the outcome measures prespecified, clearly defined, valid, reliable, and assessed consistently across all study participants?Q8: Were the people assessing the outcomes blinded to the participants' exposures/interventions? Q9: Was the loss to follow-up after baseline 20% or less?Were those lost to follow-up accounted for in the analysis?Q10: Did the statistical methods examine changes in outcome measures from before to after the intervention?Were statistical tests done that provided p values for the pre-to-post changes?Q11: Were outcome measures of interest taken multiple times before the intervention and multiple times after the intervention (i.e., did they use an interrupted time-series design)?Q12: If the intervention was conducted at a group level (e.g., a whole hospital, a community, etc.) did the statistical analysis take into account the use of individual-level data to determine effects at the group level?Total Score: Number of yes; CD, cannot be determined; NA, not applicable; NR, not reported; N, no; Y, yes.

| Between-group comparisons in posttreatment health-related quality of life and patient satisfaction
Low-quality evidence supported that patients with MC1 or a mixed cohort of patients with MC1 and MC2 in the intervention group displayed significantly better quality of life (as measured by EQ5D-5L or EQ-5D thermometer) than placebo controls at the 12-month followup.However, low-quality evidence suggested no such significant between-group difference in patients with MC2 (Table A2).
Additionally, low-quality evidence from one included RCT substantiated that patients with MC1 or a mixed cohort of patients with MC1 and MC2 in the antibiotic group showed significantly better but small improvement in the quality-adjusted life-year (QALY) than their placebo controls at the 12-month follow-up.However, the same study found no significant between-group difference in patient satisfaction in a mixed cohort of patients with MC1 and MC2 at the 12-month follow-up (Table A2).

| Between-group comparisons of posttreatment changes in MC1 size or serum content
Low-quality evidence suggested that there was no significant between-group difference in the size of MC1 volume or serum analysis results at the 12-month follow-up (Table 3, Table A2).

| Between-group comparisons of drug use, sick leave, or medical consultations
Low-quality evidence from one RCT supported no significant between-group difference in the usage of analgesics, non-steroid anti-inflammatory drugs, or opioids among patients with MC1 or MC2 over the 12-month study period.Likewise, there was low-quality evidence that the Amoxicillin-clavulanate and placebo groups had a similar number of sick leave days over 1 year.Conversely, low-quality evidence supported that patients with MC1 in the intervention group had significantly less LBP-related medical consultation than those in the placebo group over 1 year (Table A2). 28

| Between-group comparisons of adverse events
There was low-quality evidence that oral antibiotics caused significantly more adverse events (low-grade gastroenterological complaints) than placebo pills.
3.9 | Within-group improvements in pain, disability, global perceived health, and quality of life following oral antibiotic treatments Very low-quality evidence from case series showed that oral antibiotics yielded both clinically and statistically significant decreases in LBP among patients with MC1, MC2, or mixed MCs immediately post-treatment, and at the 12-or 24-month follow-ups (Table 4, Table A3).Very low-quality evidence also supported that Amoxicillinclavulanate significantly reduced the number of days with LBP or the number of LBP-related sleep disturbances.Likewise, there was very low-quality evidence that Amoxicillin-clavulanate significantly reduced leg pain in a mixed cohort of patients with MC1 and MC2 immediately post-treatment, and at 12-and 24-month follow-ups, although these changes were not clinically significant.
Very low-quality evidence supported that 100 days of Amoxicillinclavulanate treatment significantly improved RMDQ scores or patientspecific function scale scores in CLBP patients with MC1 immediately post-treatment and at the 12-month follow-up (Table 4, Table A3).

Strength of evidence
Between-group comparisons at the 1-year follow-up for patients with MC1 only  The calculated percentages reported in the article were wrong based on the number of reported participants.
Likewise, very low-quality evidence substantiated that Amoxicillinclavulanate significantly improved RMDQ scores immediately after treatment, and at the 12-and 24-month follow-ups in CLBP patients with MC1, MC2, and mixed MCs.Very low-quality evidence supported that up to 76% of patients with MC1 reported moderately better, much better, or even cured LBP after 100 days of Amoxicillin-clavulanate.
However, very low-quality evidence suggested that patients did not show significant changes in white blood cell test results after taking a course of Amoxicillin-clavulanate (Table 4, Table A3).

| Sensitivity analyses
After removing the RCT with high risk of bias, 36 the quality of evidence regarding the conclusion regarding the effects of Amoxicillin-clavulanate against placebo in improving LBP intensity and RMDQ in people with CLBP remained unchanged.However, there was no more evidence to support that Amoxicillin-clavulanate yielded clinically significantly better LBP reduction than placebo.Likewise, the removal of a poor-quality case series did not alter our conclusion on the within-group improvements in clinical outcomes following antibiotic treatments.[52][53][54][55] Although current CLBP treatment guidelines recommend avoiding the use of opioids when possible, both opioid prescription rates and longterm use patterns remain to be a concern in recent years. 56,57The complex etiology of CLBP is further evidenced by the presence of potential low-grade infection of intervertebral disc, suggesting microbiome-targeted therapies as new potential treatment avenues. 12,23,28Our findings regarding patients receiving antibiotics experiencing greater improvements in pain and LBP-related disability  appear to support this notion.This concept is particularly interesting as it relates to MCs, an underlying inflammation and bone edema that may be mediated by bacterial cytokine and propionic acid production. 27While MCs and associated endplate abnormalities are known to play a significant role in CLBP, meriting their own specific code in the updated ICD-10 coding system to designate "vertebrogenic LBP," clinical effects of antibiotic treatment for LBP depend on MC types. 170][61] Past research has illustrated a direct correlation between antibiotic consumption and resistance at a country-wide level. 61Two meta-analyses reported that greater antibiotic consumption was associated with more antibiotic resistance [odds ratios (ORs) ranging from 2.3 to 2.5], although this resistance appeared to diminish within 12 months (OR 1.33, 95% CI 1.2-1.5). 59,60Similar research also suggests a positive association between dosages of antibiotics and adverse effects, which concurred with our analyses. 62That said, antibiotics (including amoxicillin) are often used for 3-4 months before the re-evaluation of the acne treatment. 63These durations exceed 100 days of antibiotic treatment for CLBP.However, prior research showed that oral amoxicillin was not more cost-effective than placebo in treating CLBP, further high-quality RCTs are warranted to compare the cost-effectiveness between antibiotics and other active CLBP treatments in different healthcare settings.

| Strengths and limitations
The strengths of the present review lie in its comprehensive literature search, pre-registered research protocol, as well as summary and analyses regarding the role of antibiotics in treating CLBP.Specifically, the strengths stem from standardized screening and data extraction procedures, the use of validated risk of bias assessment tools, and the GRADE method to synthesize evidence, as well as reporting according to the PRISMA guidelines.
The present review is not without limitations.First, two included studies (one RCT and one case series) had 29% to 41% attrition rates, which could have confounded the results. 36,37Second, one included RCT had a high risk of bias that might have affected our results. 36However, our sensitivity analysis yielded the same conclusion after discarding its findings.Third, the favorable findings in the included case series might be related to natural history rather than the effects of antibiotics.Fourth, because our review only included two high-quality small-scale Danish and Norwegian RCTs, our findings may not be generalized to other patient populations (e.g., older patients) or healthcare settings. 64 significance remains uncertain.Future large-scale high-quality RCTs are warranted to validate the effectiveness of antibiotics in individuals with CLBP.K E Y W O R D S low back, meta-analysis, Modic change, nucleus pulposus, pain, pain antibiotics, systematic review 1 | INTRODUCTION MEDLINE, EMBASE, CINAHL, Allied and Complementary Medicine, and Cochrane Library were searched from inception to March 3, 2023.There was no language restriction in the search.The search string included the combinations of keywords and medical subject headings (MeSH) related to (1) patients with CLBP aged 18 years or older; (2) disc degeneration, MCs, or endplate signal changes; (3) RCTs or non-RCTs; and (4) antibiotics.The search strategies/definitions are illustrated in Tables A1-

F I G U R E 3
Forest plots of effects of antibiotics versus placebo on low back pain (LBP) intensity in patients with chronic low back pain and type 1 Modic changes immediately post-treatment or at the 12-month follow-up (9-month post-treatment).T A B L E 3 The effects of antibiotics versus placebo on clinical outcomes among people with chronic low back pain in the included randomized controlled trials.Study (initial sample size)

4 |
DISCUSSIONThis is the first systematic review and meta-analysis to summarize the efficacy, and safety of antibiotic use in improving clinical outcomes in patients with CLBP who exhibit MCs on MRI.Results from three RCTs indicated that patients with MC1 receiving oral antibiotics (i.e., amoxicillin or amoxicillin/clavulanate) experienced significantly better pain reduction immediately post-treatment than placebo controls.While no significant between-group differences in LBP intensity were noted at the 12-month follow-up, the prevalence of constant back pain, LBP-related sleep disturbance, positive cranial compression testing, and pain during lumbar flexion/extension among the antibiotic cohort showed significantly better improvements than the placebo group.Compared to placebo, antibiotics yielded significantly better leg pain reduction in patients with MC1, or mixed MC1 and MC2 immediately post-treatment and at the 12-month follow-up.However, only patients with MC1 displayed clinically significant between-group differences.Similarly, oral antibiotics achieved significantly greater reductions in LBP-related disability among patients with MC1 immediately and 12 months following treatments.Additionally, a larger proportion of patients with MC1 and MC2 who took antibiotics experienced clinically significant improvement than the placebo group.Compared to patients with MC1 taking placebo pills, those receiving oral antibiotics perceived significantly better quality-of-life at the 12-month follow-up; however, no such difference was observed among patients with MC2.Despite various observed benefits of antibiotics, their evidence was very low to low.Additionally, antibiotics were not better than placebo in improving MC1 size or inflammatory biomarker profiles at the 12-month followup.Therefore, the current findings should be interpreted with caution.

F I G U R E 4
Forest plot of effects of antibiotics versus placebo on Roland Morris Disability Questionnaire (RMDQ) scores of patients with chronic low back pain and type 1 Modic changes immediately post-treatment, and at the 12-month follow-up.T A B L E 4The effects of antibiotics on clinical outcomes among people with chronic low back pain in before-and after-studies.
Fifth, although our findings revealed consistent evidence for the benefits of oral amoxicillin/amoxicillin-clavulanate in treating CLBP patients with MC1, different imaging strengths, or varying MC infiltration may yield differential patients' responses to antibiotics.Sixth, since antibiotics-related side effects might cause participants to recognize the use of antibiotics, the existing findings should be interpreted with caution.5 | CONCLUSIONS This systematic review and meta-analysis are the first of such robust study design, to our knowledge, to assess the efficacy and safety of antibiotic use in improving pain, disability, and quality of life in patients with CLBP who exhibit MCs on MRI.Using antibiotics to treat CLBP remains a topic of vehement debate in the literature.Although existing limited evidence suggested that antibiotic treatment might significantly improve back/leg pain, pain-related disability, and quality of life among patients with MC1, most improvements were only statistically significant and their effects on patients with MC2 or mixed-type MCs were less prominent.Importantly, the benefits of antibiotic treatment are offset by a higher rate of drug-related adverse events and increased cost; thus, it remains too early to recommend modification of clinical treatment guidelines to promote their systematic use for treatment of CLBP.Further large-scale RCTs are warranted to compare the clinical-and cost-effectiveness of oral antibiotics and other active treatments in patients with CLBP with concomitant MC1 on imaging.APP E NDIX 1 : SEARCH STRATEGIES T A B L E A 1 Medline.
Characteristics of the included studies.
ITT analysis and per protocol analysesFor primary analysis, analysis of covariance (ANCOVA) with adjustment of baseline RMDQ score and the stratification variables; Missing RMDQ data was imputed by a multiple imputation model For secondary analyses, separate ANCOVA for RMDQ of two MC types.Other secondary outcomes were also analyzed.T A B L E 1 (Continued) Counting the number of returned study drug capsules at the end of treatment visit Band blinding Index Clinically significant difference: reduction of baseline Norwegian RMDQ by 4 points, ODI by 13-20 points; NPRS by 2-3 points; EQ5D-5L by 0.11-0.30Wilcoxon signed-rank tests Outcomes: LBP intensity, no. of days with pain, disease RMDQ, global perceived effect, Selfperceived function scale, serum analysis Limitations: No post-treatment MRI scans; no biopsy observational study with a highly selected group of patients T A B L E 1 (Continued) No. of night-time awakening Analgesics consumption Repeated MRI between 6 and 12 months for 12 participants Abbreviations: EQ5D-5L, EuroQol's health-related quality of life, version 2; NPRS, 11-point numeric pain rating scale.

1
Low* back pain OR LBP OR back pain OR lumbago OR lumbar spine pain OR Sciatica 2 Lumbar OR lower back OR back OR low back OR vertebral marrow OR intravertebral edema OR vertebral edema OR intravertebral oedema OR vertebral oedem* OR endplate chang* OR endplate oedema OR end-plate oedema OR endplate edema OR end-plate edema OR endplate NEAR/3 lesion* OR endplate NEAR/3 change* OR vertebral marrow NEAR/3 lesion Low* back pain OR LBP OR back pain OR lumbago OR lumbar spine pain 2 Lumbar OR lower back OR back OR low back OR vertebral marrow OR intravertebral edema OR vertebral edema OR intravertebral oedema OR vertebral oedem* OR endplate chang* OR endplate oedema OR end-plate oedema OR endplate edema OR end-plate edema OR endplate NEAR/3 lesion* OR endplate NEAR/3 change* OR vertebral marrow NEAR/3 lesion Allied and complementary medicine (OVID) 1 Low* back pain OR LBP OR back pain OR lumbago OR lumbar spine pain 2 Lumbar OR lower back OR back OR low back T A B L E A 2 EmbaseNo.QueryT A B L E A 3 CINAHL (EBSCO) 1 T A B L E A 4