Corticosteroid Injections Into Lumbar Facet Joints

A Prospective, Randomized, Double-Blind Placebo-Controlled Trial

Abstract

Background 

Corticosteroid injections into the intra-articular zygapophysial (z-joints) are frequently used to treat this cause of low back pain. No studies have been done on the efficacy of intra-articular corticosteroids in those with z-joint pain confirmed by dual comparative medial branch blocks.

Objective 

The aim of the study was to determine whether an injection of a corticosteroid into lumbar z-joints is effective in reducing pain and the need for radiofrequency neurotomy.

Methods 

This is a double-blind, prospective, randomized, placebo-controlled trial. The study was conducted in Academic Medical Center. Twenty-eight subjects with z-joint pain confirmed by medial branch blocks were included in the study. Subjects with confirmed z-joint pain via dual comparative medial branch block were randomized to receive either intra-articular corticosteroid (triamcinolone 20 mg) or saline via fluoroscopic guided injection.

Results 

No statistically significant difference in the need for radiofrequency neurotomy (radiofrequency neurotomy) between the groups, with 75% (95% confidence interval = 50.5%–99.5%) of the saline group vs. 91% (95% confidence interval = 62.3%–100%) of the corticosteroid group receiving radiofrequency neurotomy. There is no difference in mean time to radiofrequency neurotomy between saline (6.1 wks) and corticosteroid (6.5 wks) groups. There is a need for radiofrequency neurotomy.

Conclusions 

Corticosteroid injections into the lumbar z-joints were not effective in reducing the need for radiofrequency neurotomy of the medial branches in those with z-joint pain confirmed by dual comparative medial branch blocks.

Attention has been directed toward utilization of interventional spine procedures as a contributor to the increasing health care costs of low back pain in the United States.^1,2 Spine interventions targeting the lumbar zygapophysial joints (z-joints) comprise roughly 10% of lumbosacral injections in the United States.²

As a synovial joint, the lumbar z-joints are commonly affected by degenerative changes and osteoarthritis.³ Studies have reported a prevalence between 3% and 45% of patients with axial low back pain attributed to the z-joint, depending on how the diagnosis was made.^4,5 Higher prevalence has been reported in studies using history, physical examination, and/or imaging findings, whereas studies using blocks have lower reported prevalence.⁶

Intra-articular (IA) corticosteroid injections are a common and accepted therapeutic option for the treatment of osteoarthritis pain in peripheral joints.⁷ Not surprisingly, IA corticosteroid injections into the z-joints have been theorized as a potential treatment for low back pain. Unfortunately, the current evidence supporting the effectiveness of z-joint steroid injections is scarce.^8–10 In addition, the conclusions of the few existing studies may be limited because of the utilization of overinclusive and nonspecific inclusion criteria.¹⁰ This is because there is no reliable way to diagnose lumbar z-joint pain using historical features and/or physical examination findings alone.¹¹ Therefore, studies that used history and/or physical examination findings as the inclusion criteria likely enrolled a heterogeneous patient population. Even single diagnostic blocks are somewhat limited in their ability to accurately diagnose z-joint pain, because they have a false-positives rate more than 30%.⁴ This has lead to the recommendation of dual comparative medial branch blocks (MBB) as the accepted method of clinically diagnosing z-joint mediated pain.¹²

To date, there are no studies that use the current accepted method for diagnosing z-joint pain (dual medial branch blocks) to enroll subjects and attempt to evaluate the efficacy of IA z-joint steroid injections.¹² The goal of this study was to determine the efficacy of corticosteroids injected into the IA z-joints, in a homogeneous group of patients as selected by a positive response to dual comparative medial branch blocks.

METHODS

This study was institutional review board approved (319-2010) and registered on clinicaltrilas.gov (NCT01382771). Written informed consent was obtained from all patients. It was a prospective, randomized, double-blind, placebo-controlled study from a single academic medical center. The Spine Intervention Society funded the study. Patients were clinically evaluated for chronic low back pain by either a board-certified physical medicine and rehabilitation attending or a board-certified orthopedic spine surgeon in a spine clinic. Enrollment occurred between October 2010 and September 2012. This study conforms to all CONSORT guidelines and reports the required information accordingly (see Checklist, Supplemental Digital Content 1, https://links.lww.com/PHM/A613).

Inclusion/Exclusion

Subjects who were clinically suspected of having z-joint mediated low back pain were selected for a single set of medial branch blocks if they had (1) pain rating of ≥4/10 on the 0–10 Numeric Pain Rating scale, (2) symptom duration of 3 mos or more, and (3) failed conservative therapy with minimum of physical therapy and medications (see Fig. 1 for patient flow diagram, Table 1 for full inclusion and exclusion criteria). The MBBs could involve a single or multiple levels and be unilateral or bilateral. The z-joints that were targeted were selected based on clinical assessment combined with review of magnetic resonance imaging or computed tomography findings. Unilateral or bilateral injection was based on whether the patient had unilateral or bilateral pain. In addition to location of pain and image findings, determining what level to inject was also made with consideration to prevalence data suggesting that L4-5 and L5-S1 are the most commonly affect segments for facet arthropathy.¹³

FIGURE 1: Patient flow diagram.

TABLE 1: Inclusion/exclusion criteria

Procedure/Technique of First Screening MBB

The MBB was done according to the Spine Intervention Society guidelines,¹² using fluoroscopy for needle placement and live injection of contrast dye to determine ideal placement and avoid intravascular flow.¹⁴ The needles were placed at the junction of the superior articular process and the transverse process and confirmed with injection of contrast dye. Approximately 0.2 ml of 0.5% bupivacaine was then injected at each level. Patients were assessed immediately after the injection for pain relief using patient specific provocation maneuvers identified in the preoperative screening examination and also given a pain diary. Subjects were considered to have a positive block if they had 80% or more relief of their low back pain by either perceived percentile improvement or by percentage improvement on pain numeric rating scale within the first 3 hrs after this procedure.

Enrolment at Time of Second MBB/IA Injection

Those with an initial positive MBB with the 0.5% bupivacaine were offered a second MBB with 2% lidocaine using the same approach as the first block. At the time of the second MBB, patients were also offered enrollment in the study. If enrolled, patients were then randomly assigned to receive either 0.5 ml of triamcinolone 40 mg/ml (Bristol-Myers Squibb Company, Princeton, NJ) or 0.5 ml of preservative free normal saline injection into the IA z-joints that were identified by the initial MBBs. No subjects had more than 4 z-joints targeted, and no subjects received intravenous or oral pain medications, anxiolytics, or any other form of conscious sedation before or immediately after the procedure. The IA injection was done using fluoroscopic guidance, and needle placement was confirmed by multiplanar fluoroscopy with the injection of approximately 0.1 to 0.2 ml of contrast dye also under live fluoroscopy.

Blinding

Subjects and treating physicians were blinded to the treatment. The treating physician was blinded by the assistance of an interventional physical medicine and rehabilitation spine fellow. The injection was done by the fellow; if needed, the attending would insert the needles and confirm IA placement by injection of contrast dye (Pictures 1–7) and then have the fellow inject the treatment solution. The use of an independent interventionalist allowed the attending physician to remain blinded. The juxtaposition of the second MBB with the IA injection was done to facilitate recruitment, avoid subsequent procedures, minimize additional radiation exposure, and decrease procedural and facility costs to the subject. All injections, both MBB and lumbar z-joint injections, were performed in accordance with Spine Intervention Society guidelines.¹²

Response to Second MBB

As with the first MBB, patients were assessed immediately after the injection for pain relief and given a pain diary. Only those with 80% or more relief either by perceived improvement or by percentage improvement on pain numeric rating scale were considered to be a positive block. Patients not achieving 80% or more pain relief with the second injection were not included in the final analysis (Fig. 1). Patients were also excluded if they demonstrated longer duration of relief with the second MBB than the first MBB, as the first block used 0.5% bupivicaine while the second block used 2% lidocaine.

Outcomes/Statistical Analysis

The initial primary outcome was the categorical percentage of patients achieving 80% or more pain reduction at 6 wks on the 0–10 Numeric Pain Rating scale. Secondary planned outcomes included Numeric Pain Rating scale pain reduction and Oswestry Disability Index (ODI) at 3, 6, and 12 mos. Secondary outcome measures were initially described as categorical outcomes on the ODI, work status, analgesic usage, ancillary treatment, and the categorical need for a radiofrequency neurotomy (RFN) due to failure of care as determined by both the subject and physician (both blinded).

Categorical outcomes were analyzed by calculating 95% confidence intervals (CIs), with non-overlapping confidence intervals considered statistically significant. Group data were analyzed by calculating means. t test was used with P values of less than 0.05 considered significant when evaluating for differences in baseline characteristics.

RESULTS

This study had 28 subjects enrolled, although 4 patients failed to have 80% or more relief with the second confirmatory block and were thus not used in the final results (Fig. 1). In total, 24 of 28 patients (85.7%, 95% CI = 72.7%–98.7%) had a positive response to the second block. Subjects not responding to the second block were evenly distributed between the saline and corticosteroid groups.

Mean age at enrollment was 63.15 yrs (Table 2). Twenty subjects were female and eight were male. Age and gender distribution were similar between groups, as was the anatomic level of the z-joints targeted (Table 3). This group was representative of a chronic pain population with the average duration of symptoms being 5.4 yrs, with no statistical difference between groups (5.0 yrs vs. 5.8 yrs, P = 0.75).

TABLE 2: Demographics of subjects at enrollment

TABLE 3: Joint levels targeted at baseline^a

In the placebo group, 9 of 12 subjects (75%, 95% CI = 50.5%–99.5%) went on to receive RFN (Table 4). In those receiving corticosteroids, 10 of 12 subjects (83.3%, 95% CI = 62.3%–100%) went on to receive RFN. The mean time to RFN was 6.1 wks in saline group and 6.5 wks in the steroid group (P = 0.89). There was no statistically significant difference in either the need to progress to or time to RFN between groups.

TABLE 4: Group results

More than 75% of enrolled subjects underwent RFN before the first scheduled follow-up at 6 wks. As such, data pertaining to ODI and Numeric Pain Rating scale were irreparably corrupted by a confounding intervention and subsequently is not presented, because it does not accurately reflect difference in outcomes between the two groups. Decision to proceed to RFN was made based on perceived inadequate relief after IA z-joint injection of either steroid or saline.

DISCUSSION

This is the first study that the authors are aware of to use dual controlled MBBs as diagnostic criteria for enrollment of patients into a study comparing IA z-joint steroid injections to placebo for the treatment of lumbar z-joint pain. This study was initially designed to determine pain relief as a primary outcome; however, after enrolling 28 subjects, it became clear that the study had a fundamental flaw in its primary outcome and the associated follow-up timeframe. Specifically, the average time to RFN in this study was 6 wks, which unfortunately corresponded with the first scheduled research outcome timeframe. Therefore, the researchers quickly realized that the initially planned primary and secondary outcome measures of 0–10 Numeric Pain Rating and ODI may be obtained immediately before, after, or even during the RFN. Because more than 75% of enrolled subjects underwent RFN at the scheduled follow-up time, outcomes such as visual analog scale and ODI would be rendered invalid, because both treatment groups were largely corrupted by being subjected to a proven treatment method. Because other studies have demonstrated lumbar RFN to be an effective treatment in select patients,¹⁵ it was determined that the RFN must be offered if indicated, as withholding it was felt to be unethical. Therefore, it was realized that progression to RFN must be the primary outcome in and of itself, as any other alternative would involve not offering RFN as a rescue treatment. In addition, because the indication for an RFN is continued pain in the setting of positive diagnostic MBBs, and an RFN is known to induce postprocedure pain in a large percentage of patients,¹⁶ the authors realized that the primary outcome timeframe of 6 wks was also fundamentally flawed. At this point, no further subjects were enrolled and follow-up ceased.

However, despite these limitations, it is important to note the strengths of the study and the additional insights it offers, especially when compared with the current available literature.

Many of the currently available studies used the presence of low back pain as enrollment criteria.¹⁰ However, this resulted in very heterogeneous study populations due to the fact that z-joint pathology accounts for low back pain in no more than 45% of patients even by using the most generous estimates.^5,17 This study is unique because it is the first to use the most validated means of diagnosing z-joint mediated pain, dual comparative MBBs, to enroll a more homogenous group of patients with confirmed z-joint pain. If lumbar z-joint steroid injections do have a positive effect on z-joint pain, finding the true effect should be more likely in such a group.

As a double-blind, randomized, placebo-controlled study, this study adds some additional clarity to the questions that remain despite previous studies. First, information can be gleaned regarding the utilization of RFN. Given the practice patterns of the treating physicians who were blinded, only patients with a return of pain would be offered an RFN. Because 75% or more of both groups went onto receive an RFN, without a significant difference in the time to RFN, it seems that neither corticosteroids nor saline into the IA z-joints offer enough pain relief to avoid a subsequent RFN. The same can also be said of the concurrent MBBs that all patients received. Though not specifically designed to evaluate the effectiveness of a MBB, this study adds further data showing that MBBs are not effective therapeutic procedures.¹⁸

Although this study did not show a difference in the categorical need for an RFN, there are other studies with divergent results on the amount of pain relief obtained with corticosteroids. Most studies selected patients based on history, physical examination, or radiographic findings. Even the best available studies on IA z-joint steroid injections use only single-photon emission computed tomography (SPECT) imaging to identify patients.^19,20 Given the variability in patient selection, it is not surprising that the results from observational studies of lumbar z-joint steroid injections vary greatly. The reported percentage of patients who have complete relief in observational studies varies significantly from 10% to 82%, with most studies falling in the 15% to 38% range^21–27 in the 1- to 12-mo follow-up period.

Two pragmatic studies have shown equivocal results.^28,29 A study by Lilius et al.²⁸ compared IA anesthetic plus steroid, extra-articular anesthetic plus steroid, or IA saline in patients with 3 mos of unilateral low back pain, without performing diagnostic MBBs. There were no differences between the groups noted. However, this study used volumes of injectate that exceed the physiologic capacity of the lumbar z-joints, raising significant doubt about the accuracy of the IA needle placement. Carette et al.²⁹ compared IA z-joint injection of either methylprednisolone or saline. Only patients who achieved 50% or more pain relief 30 mins after a single lumbar IA z-joint anesthetic injection were included. There was no difference between groups at 1 and 3 mos. However, at 6 mos, 22 of 49 (46%, 95% CI = 31.8–58.6) of patients in the steroid group had “marked” or “very marked” improvement compared with only 15% in the saline group (P = 0.002).²⁹ While appearing more favorable at this one time point, the results are in question given a lack of physiologic plausibility that significant results did not appear until 6-mo postinjection.

Some postulates that the previous research is sufficient to conclude that IA z-joint steroid injections are no better than sham injections.⁹ However, these results contrast to the more favorable results from a number of studies that used positive z-joint uptake on SPECT imaging to identify patients with z-joint pain.^19,20,30 Single-photon emission computed tomography imaging is a known diagnostic technique that can be used to detect bone areas with increased osteoblastic activity or to detect synovial changes caused by inflammation or hyperemia. Thus, though not validated to diagnose z-joint pain to the same rigor as MBBs, SPECT imaging may have a role in identifying patients that might respond positively to a corticosteroid injection in the IA z-joint. Dolan et al.¹⁹ found that in patients who received corticosteroids into the z-joint, those with positive findings on SPECT showed a statistically significant lower mean visual analog scale (3.3) compared with the SPECT negative group (5.5) (P < 0.05). Pneumaticos later corroborated these findings in a similar study.³⁰ Ackerman and Ahmed²⁰ have since used SPECT as a means to enroll patients in a pragmatic study comparing z-joint steroid injections to perineural medial branch steroid administration. At 12th weeks, 14 of 23 (61%, 95% CI = 41%–81%, P < 0.05) patients in the IA group reported 50% or more relief compared with only 6 of 23 (26%, 95% CI = 8%–44%) of those in the perineural group. Although skeptics will note that there is a small overlap in the 95% CIs (41%–81% vs. 8%–44%), these studies suggest that z-joint corticosteroid injections may benefit patients selected by these diagnostic criteria. This study also demonstrated that perineural medial branch steroid injection is not an effective treatment. The previous studies may point to using SPECT imaging, a noninvasive means of diagnosing patients with z-joint pain, to determine patients who may respond to corticosteroid injections. Again, this beckons for further research, which this study attempts to provide. Lastly, to the authors' knowledge, no prospective studies have been performed to evaluate the efficacy of IA z-joint steroid injections on low back pain patients with increased perifacet signal seen on fat saturation magnetic resonance imaging sequences. This is another area that beckons for further research, as this imaging finding may represent a true synovitis that may be amenable to steroid treatment.

Several additional weaknesses of our study design are also apparent. First, patient weight was not evaluated as part of the baseline demographics, which may have resulted in one group to have a disproportionate number of normal weight or overweight patients. There is no evidence of the effectiveness of facet steroid injections being affected by weight in other literature, however. Secondly, the duration of pain at time of enrollment must be noted. Enrollment in this study essentially reflects a chronic pain population. Issues related to chronic pain including central sensitization, catastrophizing, secondary gain, and poor coping mechanisms are barriers to achieving significant pain relief. Given that the study was terminated early because of the perceived flaws, it is not clear whether this patient population actually went on to obtain relief from the RFN. Specifically, this study population may represent a group of patients that do not achieve long-term relief with any interventional procedures. Although this would be more different than the published literature,¹⁵ long-term outcomes are necessary to show this group's response to the subsequent RFN. The authors of this study are currently enrolling subjects into a large study protocol based on modifications to this protocol to obtain these data. In addition, this study had a smaller percentage of nonresponders to the second block than previously reported,^5,15 which may represent a different patient population or a robust placebo response to the diagnostic MBB. Some may also question the use of IA saline as a placebo. However, given that neither treatment offered enough relief to avoid an RFN, it does cast doubt on saline being a highly effective treatment.

The use of RFN as a rescue treatment also has additional concerns. First, it is unclear what degree of pain relief may be needed to avoid proceeding to RFN. If subjects only received partial pain relief several weeks after the injection, they may opt for a RFN to achieve even greater relief. It is possible that the corticosteroid offered some relief yet failed to prevent the subsequent RFN. Although this is difficult to ascertain, it is unlikely that a patient receiving significant pain relief from a corticosteroid injection would undergo a subsequent RFN. In addition, given the double-blinded nature of the study, it is unlikely that a major difference existed between groups.

Although there is theoretical plausibility that z-joint steroid injections may have a role in treating low back pain in select patients, the current literature is lacking in quality studies. A well-defined homogeneous patient population, including only those with confirmed lumbar z-joint pain, is the best way to detect such an effect. Moreover, to ensure results are not due to sham, such a study also needs to adhere to the rigors of a double-blind placebo-controlled study with longer-term follow-up.

CONCLUSIONS

In this study, corticosteroids injected into the IA z-joints were not effective in reducing the need for RFN of the medial branches in those with dual medial branch block confirmed z-joint pain. A larger, more adequately powered study is needed to further clarify the hypothesis at hand. Given the current available research, pursuing such a study is warranted.