Radiofrequency Applications to Dorsal Root Ganglia: A Literature Review

The MEDLINE, EMBASE, and Cochrane databases were searched for key words radiofrequency and dorsal root ganglion , DRG and radiofrequency , RF-DRG , pulsed radiofrequency and dorsal root ganglion , DRG and pulsed radiofrequency , and PRF-DRG ; this search yielded 49 articles. To limit the review to peer-reviewed literature, conference proceedings and abstracts were not sought. Studies that involved application of radiofrequency currents, of any modality, to the spinal nerve roots or DRG, in the treatment of pain, were included in the review. Editorials, letters, and any duplicate articles were not included. Twenty-six of the retrieved articles met the above criteria and were further reviewed. An additional 6 pertinent articles were identified after the review of 26 initially identified articles. To understand the evolution of this technique, elucidate its mechanisms of action, and tabulate the adverse effects, all the identified clinical and laboratory studies were reviewed.1–32A meta-analysis of the clinical data obtained was not possible because of the small number of RCTs on diverse clinical conditions that studied heterogeneous patient populations. The best evidence synthesis method33was therefore used to assess the outcomes of the literature reviewed.

The best evidence synthesis method seeks to provide the best available evidence on the topic at hand, disregarding the lesser-quality evidence in favor of the better-quality one. This method of research review is especially suited when pooling across the published studies is not possible because of a small number of trials reporting on a large number of study categories. The standards for the evidence and the study selection criteria are predetermined and are objective, consistent, and germane to the topic being reviewed. After the study selection criteria are established, the study inclusion techniques are exhaustively inclusive, similar to a meta-analysis. Unlike a meta-analysis, however, detailed descriptions of the best evidence on the topic are provided, to give readers the opportunity to verify the original literature and formulate their own independent conclusions.

In determining the best evidence for the efficacy of RF-DRG, we considered the randomized controlled data as the best evidence available, and these studies were therefore further critically analyzed. The internal validity of the RCTs identified was based on the validated criteria in pain research proposed by Jadad et al.  34The four levels of best evidence (table 1) used in this review, and the 5-point Jadad scale used for qualitative study analysis (table 2), were similar to the ones used in the previous related reviews.1,2Identification of an item on the Jadad scale in a trial rendered it a positive score, whereas its absence was marked as zero; a trial scoring 3 or more points was graded as a high-quality trial, whereas a trial scoring 2 or fewer points was regarded as a low-quality trial.1 

Of the 32 articles reviewed (fig. 1), there were 3 review articles,1–324 clinical studies4–26(one publication9had 2 clinical studies), and 6 laboratory studies.27–32Fifteen clinical studies were of conventional or continuous mode radiofrequency application to the DRG (continuous RF-DRG),4–7,9–15,18,19,23,24and 9 were of pulsed mode RF-DRG.8,9,16,17,20–22,25,26There were 5 RCTs,4–84 of continuous RF-DRG4–7and 1 of pulsed RF-DRG.8Three RCT pertained to cervicobrachial pain,4,5,81 pertained to cervicogenic headache,6and 1 pertained to lumbosacral radicular pain.7There was 1 nonrandomized controlled trial of both pulsed and continuous RF-DRG use.9There were 9 uncontrolled studies of prospective design,9–17of which 6 were of continuous RF-DRG10–15and 3 were of pulsed RF-DRG9,16,17use. Of the 5 retrospective studies encountered,18–222 used continuous RF-DRG18,19and 3 used pulsed RF-DRG.20–22There were 4 case reports or case series, 2 each of continuous23,24and pulsed RF-DRG.25,26Two of the studies5,26had duplicate articles that were not included in this review.

Of the 6 prospective controlled trials (table 3), 2 were RCTs of continuous RF-DRG in the treatment of cervicobrachial pain. The trial by Van Kleef et al.  4included 20 patients with cervical radicular symptoms; 9 patients received the continuous RF-DRG, whereas 11 received the sham treatment: the electrode was placed in a manner identical to that used in the treatment group, but no radiofrequency current was applied. Patients were evaluated before and at 8 weeks after the treatment, using greater than 2-point reduction of pain on the visual analog scale (VAS) as the criterion for success. The mean VAS scores decreased from 6.4 to 3.3 in the treatment group and increased from 5.9 to 6.0 in the sham treatment group. Eight patients (88.8%) in the treatment group and 2 patients (18.1%) in the sham group were reported to have successful results, which were statistically significant (P = 0.0027, Fisher exact test). The treatment group also showed greater improvement on the Multidimensional Pain Inventory and McGill Pain Questionnaires, with respective changes of 1.7 and 11.2 in the treatment group compared with 0.1 and 0.4 in the control group. The authors concluded that continuous RF-DRG provided significant short-term pain relief in patients with cervicobrachial pain. This was a relatively small study of only 20 patients. Although this trial was randomized and double-blinded, the randomization techniques were not adequately described. The trial also lacked the description of any analgesic drugs used, and only short-term results at 8 weeks were reported.

Slappendel et al.  5published the second RCT of continuous RF-DRG in patients with cervicobrachial pain. It included 61 patients; 32 patients received continuous RF-DRG with the electrode tip heated to 67°C (group 1), and 29 received continuous RF-DRG with electrode tip temperature of 40°C (group 2). The patients were evaluated with VAS scores, and subjectively in terms of “better,”“equal,” or “worse” pain. In group 1, the mean VAS scores decreased from 6.7 to 4.8 at 6 weeks and increased to 5.0 at 3 months; in group 2, the mean VAS scores decreased from 6.3 to 4.9 at 6 weeks and to 4.4 at 3 months. Clinically significant pain relief (> 2-point reduction in VAS scores) was reported in 15 patients in each of the groups at 3 months—47% and 51% of the patients in groups 1 and 2, respectively. No statistically significant difference in the reduced pain scores was found between the two groups. This was a multicenter trial, and 63 patients were included after screening 314 patients, with dropout rate of only 2 patients at 3 months; it was randomized and double-blinded. However, the criteria for a successful outcome were not defined, and only short-term results at 3 months were reported. This trial lacked a placebo control group, and the results of the two treatment groups were compared only to historic controls, the placebo control group in the trial by Van Kleef et al.  4and the diagnostic nerve root blocks.

Haspeslagh et al.  6published an RCT of continuous RF-DRG in the treatment of cervicogenic headache; 30 patients with severe chronic cervicogenic headache were randomized to two equal groups, a radiofrequency group (group 1) and a local nerve injection group (group 2). The patients in group 1 received radiofrequency lesioning of the facet joints (C3–C6), and at 8 weeks, the patients with continued pain that responded positively to a series of diagnostic nerve root blocks received continuous RF-DRG at the affected level. The patients with continued pain at 8 weeks in this group who did not respond to the diagnostic nerve root blocks received transcutaneous electrical nerve stimulation therapy. The patients in group 2 received local anesthetic and steroid injection of the greater occipital nerve; this was repeated at 8 weeks in the unresponsive patients. At 16 weeks, the nonresponsive patients in both groups received transcutaneous electrical nerve stimulation therapy. The primary outcome measures used were VAS and global perceived effect (GPE), and success was defined as a greater than 2-point reduction on the VAS and/or GPE score of greater than +2. The 8-week, 16-week, and 1-yr success rates for group 1 were 80, 66.7, and 53.3%, respectively, compared with 66.7, 55.3, and 50% for group 2. There was no statistically significant difference in the success rates between the two groups, and the authors concluded that the sequential radiofrequency treatments of facet joints and DRG had similar efficacy to the local nerve injections in the treatment of cervicogenic headaches. Although this trial used multiple outcome measures, kept record of the medications used, and followed up patients for a year, this was a relatively small trial, with only 15 patients in each study group, it had a high dropout rate of almost 33%, and the randomization and double-blinding techniques were not adequately described.

Geurts et al.  7published an RCT of continuous RF-DRG application in patients with chronic lumbar radicular pain. Of the 83 patients included in the trial, 45 received the continuous RF-DRG, and 38 received the sham treatment. The outcome measures included VAS, Analgesic Rating Scale, and Short Form-36 quality-of-life questionnaire. The success was defined as either reduction in the VAS scores by 50% or reduction of the combined VAS and Analgesic Rating Scale scores by 25%, and increase of Short Form-36 scores by 25%. At 3 months, the mean VAS leg pain score decreased from 6.1 to 5.4 in the treatment group and from 6.2 to 4.2 in the sham-treated group. Seven patients (16%) in the continuous RF-DRG group and 9 patients (25%) in the sham-treated group were considered to have successful results (P = 0.43). The authors concluded that continuous RF-DRG was not an effective treatment for chronic lumbar radicular pain and recommended against its routine use. This was a large multicenter trial, with 1,001 patients screened over a period of 2½ yr. Of the 83 patients included in the trial, only 3 dropped out at 3 months. This trial was placebo controlled, and adequate conduct of randomization and double-blinding was described. Multiple outcome measures evaluating multiple domains, including pain, physical impairment, and analgesic use, were used and were included in the final outcome evaluation. This trial, however, reported short-term results at 3 months and excluded patients with sciatica who had neurologic deficits.

The only RCT of pulsed radiofrequency (PRF) application to the DRG was published by Van Zundert et al.  8It included 23 patients with chronic cervicobrachial pain. Eleven of these patients received pulsed RF-DRG, and 12 received sham treatment. Using GPE and VAS, success was defined as a greater than 2-point reduction in VAS and greater than 50% improvement in GPE scores. At 3 months, 9 patients (82%) in the treatment group and 3–4 patients (25–33%) in the ST group had statistically significant (P = 0.02–0.03) successful results; the two values indicate respective successes based on VAS and GPE criteria. Reduction of pain medication use was also noted in the PRF group, but no significance was reached at 3 months. This was a placebo-controlled trial, and the procedures for random allocation and double-blinding were well described. Although this was a multicenter study and 256 patients were originally screened, and only 1 patient dropped out during the study period, only 23 patients were recruited over a period of 2½ yr. The number of patients eventually studied was well short of the intended target of 42 patients, making the study results statistically less powerful. Although the study period was extended for 6 months and multiple outcome measures were used measuring, pain, general well-being, analgesic use, and physical disability, statistical significance in favor of PRF was reached only in pain and overall well being at 3 months. The two study groups also had dissimilar characteristics, and patients in the control group were older (52 vs.  42 yr) and had significantly higher pretreatment VAS scores (76.2 vs.  55.7). Based on the above weaknesses, the authors could only conclude that PRF of cervical DRG “might” provide pain relief at 3 months in patients with cervicobrachial pain.

The first publication of PRF use was published by Sluijter et al.  9It included a prospective nonrandomized controlled trial that compared pulsed RF-DRG with continuous RF-DRG with maximum electrode temperature of 42°C. Of the 60 patients with radicular pain of unspecified nature included in the trial, 36 received the pulsed RF-DRG, and 24 received the continuous RF-DRG at 42°C. At 6 weeks, 31 patients (86%) in the pulsed RF-DRG group and 3 patients (12%) in the continuous RF-DRG at 42°C group reported greater than 50% improvement in their GPE scores. The authors concluded efficacy of pulsed RF-DRG in comparison with continuous RF-DRG at 42°C at 6 weeks. Although this was a prospective controlled trial, it was not randomized, blinded, or placebo-controlled; only short-term results at 6 weeks were reported; and it compared pulsed RF-DRG with a form of continuous RF-DRG (continuous RF-DRG at 42°C) not used in routine clinical practice.

Of the 6 prospective uncontrolled trials of continuous RF-DRG application (table 4), 4 described its use in one spinal region. In 3 such studies, continuous RF-DRG was used in the treatment of cervicobrachial pain.10–12The study by Sluijter and Koetsveld-Baart10included 20 patients; 65% of these patients had greater than 70% pain relief at 3 and 9 months of follow-up. Vervest and Stolker11reported 24 patients; 80% of them had “excellent” to “good” pain relief at 2 months; these patients were followed up for 1.5 yr, along with a group of patients who received facet radiofrequency, and “excellent” to “good” results were reported for 84.6% of the patients in the group. Van Kleef et al.  12followed up 20 patients for 6 months and 17 patients for 9 months after continuous RF-DRG. Using a numeric rating scale, greater than 50% pain relief was reported in 50% of patients at 3 months, 30% at 6 months, and 22% at 9 months. In a study by Stolker et al. ,13continuous RF-DRG was used in the treatment of thoracic segmental pain. Of the 45 patients studied, 91% obtained greater than 50% pain relief at 2 months on a five-grade oral analog scale; 80% of the patients continued to experience the pain relief for 24 months. The authors reported long-term efficacy of continuous RF-DRG in the treatment of chronic thoracic segmental pain.

There are 2 prospective uncontrolled studies where the use of continuous RF-DRG is described in two spinal regions (table 4). Sluijter14published a study of 105 patients with cervicobrachial and lumbar radicular pain; 60 patients had lumbar and 45 had cervical continuous RF-DRG. Using patients’ opinions as “good,”“fair,” and “poor,”“good” results were reported in almost 40% of patients, at follow-up periods that varied from 3 to 21 months. Niv and Chayen15studied 50 patients with lumbar and thoracic segmental pain of malignant origin. Continuous RF-DRG was applied at the affected segmental levels, followed by injection of 40 mg methylprednisolone. At 3 months, 31 patients (62%) were “virtually pain free” and 14 patients (28%) had “fair” pain relief; 48% of the patients continued to be “virtually pain free” at 12 months.

There are 3 prospective uncontrolled trials of pulsed RF-DRG application (table 4).9,16,17Sluijter et al.  9reported 15 failed back surgery patients with chronic unilateral lumbar radicular pain who received pulsed RF-DRG at the affected segmental levels. At 6 months, greater than 2-point reduction in VAS scores was reported in 8 patients (53%); 6 of these patients (40%) continued to have the pain relief at 12 months. Pevzner et al.  16reported 28 patients with lumbar and cervical radicular pain who received pulsed RF-DRG. At 3 months, 2 patients had “excellent,” 12 had “good,” 9 had “fair,” and 5 had no pain relief. The study by Shabat et al.  17evaluated 28 patients with chronic neuropathic spinal pain who had pulsed radiofrequency of the suspected DRG. No diagnostic blocks were performed, and the involved vertebral level was determined solely by the clinical and the imaging findings. At 3 months, 82% of the patients reported reduction of VAS scores by more than 30%, a trend that continued for 1 yr (68%). All of the patients had concurrent treatments that included injection of 80 mg methylprednisolone before the pulsed RF-DRG, oral antiinflammatory medications, and physical therapy after the procedure. The nature and etiology of the spinal neuropathic pain was not elucidated.

Of the 5 studies of retrospective design (table 5), 2 used continuous RF-DRG, 1 for lumbar radicular pain and 1 for thoracic segmental pain. The study by Van Wijk et al.  18was a retrospective data analysis of 279 patients who received continuous RF-DRG for chronic lumbar radicular pain. A 4-point pain perception scale—pain free, moderate pain relief (> 50% pain relief), no change in pain, and increased pain—was used to monitor patients’ outcome. At 2 months, 164 patients (59%) experienced greater than 50% pain relief, which continued in 96 patients (58%) for a variable period of 2–70 months (mean, 22.9 months). The authors concluded that continuous RF-DRG provided long-term pain relief. The study by Van Kleef et al.  19analyzed 43 patients with chronic thoracic segmental pain of variable etiology who received continuous RF-DRG at the affected segment. Twenty-seven of the patients had pain limited to one or two segments, whereas 16 patients had multisegmental pain. At 8 weeks, 14 patients (52%) with one- or two-segment pain had greater than 50% pain relief, which continued at 9 months in 10 patients (37%); only 3 patients (18%) with pain in more than two segments obtained similar pain relief at both 8 weeks and 9 months.

There were 3 retrospective studies of pulsed RF-DRG use (table 5), 1 each in the treatment of cervicobrachial, thoracic segmental, and lumbar radicular pain. Van Zundert et al.  20performed a retrospective review of 18 patients who underwent pulsed RF-DRG for chronic cervicobrachial pain. Using GPE scores, 72% of the patients reported greater than 50% pain relief at 2 months, 56% of the patients maintained this pain relief for 3–11 months, and in 33% of the patients the pain relief lasted for more than a year. The study by Cohen et al.  21retrospectively analyzed 49 patients with chronic postsurgical thoracic segmental pain. Twenty-eight patients who received PRF of either the intercostal nerves (n = 15) or the DRG (n = 13) were compared with 21 patients who were treated pharmacologically. At 6 weeks, 62% of the patients who received pulsed RF-DRG reported greater than 50% pain relief, compared with 21% in the intercostal pulsed radiofrequency group and 27% in the medically managed group; at 3 months, these percentages were 54, 7, and 20%, respectively. Teixeira et al.  22analyzed 13 patients with lumbar radicular pain who were possible candidates for disk surgery; 9 of these patients exhibited motor and sensory deficits of the involved dermatomes. Significant improvements in the numeric rating scale scores (> 5 points) were reported in 12 patients (92%) at 1 yr, and the planned surgery was avoided; resolution of the neurologic deficits was reported in all of the patients. The authors recommended pulsed RF-DRG as an alternative to epidural steroid injections for the treatment of herniated disk. No diagnostic blocks were performed in this study, and the level of DRG lesioning was based only on clinical and imaging findings.

Of the 4 case reports or case series encountered (table 6), 2 pertained to continuous23–24and 2 pertained to pulsed RF-DRG.25–26Uematsu et al.  23used continuous radiofrequency for pain and for other symptoms. Of the 17 patients reported, 13 had chronic pain symptoms, 3 had spastic paraplegia, and 1 had Raynaud disease. At 3–48 months, “excellent” to “good” pain relief was reported in 5 patients (38%), relief of spasticity was reported in 2 of 3 patients, and temporary improvement in symptoms was reported in the patient with Raynaud disease. Nash24reported a series of 26 patients with chronic radicular pain of diverse etiology who received continuous RF-DRG at the involved sacral, lumbar, thoracic, or cervical levels. Using “excellent,”“good,” and “no improvement” criteria, “excellent” to “good” results were reported in 15 of 26 patients. Munglani25reported the use of pulsed RF-DRG in 3 patients with lumbar radicular pain and 1 patient with thoracic segmental pain. Marked reduction in pain was reported in these patients for 1–7 months. Rozen and Parvez26reported 5 patients with chronic inguinal pain after herniorrhaphy. Pulsed RF-DRG of the affected segments resulted in 75–100% pain relief, which lasted for 6–9 months.

We encountered 6 laboratory studies that pertained to RF-DRG (table 7). In a goat model, De Louw et al.  27studied the effects of continuous radiofrequency on the DRG by observing the morphologic changes and by measuring the monoclonal antibody MIB-1, with enhanced MIB-1 activity indicating a microglial proliferative and cellular injury response.35DRG lesions were created in the left lumbar region by applying continuous radiofrequency at 67°C for 60 s (treatment group). Proximity of the electrode to the DRG was determined by motor stimulation thresholds; a motor response at less than 0.2 V was considered intraganglionic, and a response at greater than 0.6 V indicated an extraganglionic electrode placement. Motor response was sought at less than 0.2 V (average, 0.1 V) at the L5 level and at greater than 0.6 V (range, 0.9–1.6 V) at the L1–L4 levels. The electrodes were placed similarly on the right side, but no radiofrequency current was applied (sham group). The control group comprised DRG that received no intervention and were procured from goats killed for unrelated experiments. Light microscopic observations, 2 weeks later, showed that in the treatment group, lesions made at the L5 level were small (1.8–2.0 mm) and intraganglionic, with complete loss of myelin of all fiber sizes, whereas the lesions made at L1–L4 levels were larger (2–2.8 mm) and extra ganglionic, and no abnormal morphology was seen in the treated DRG. The DRG MIB-1 antibody activity was significantly increased in the treatment group; it was higher in the sham-treated group and was insignificant in the control group. The authors concluded that the intraganglionic continuous radiofrequency lesions destroyed the large myelinated nerve fibers, whereas the extraganglionic continuous radiofrequency lesions affected the microglial cells without directly affecting the large nerve fibers.

Three studies measured markers of cellular stress (table 7).28–30In a rat model, Higuchi et al.  28compared the effects of pulsed RF-DRG, continuous RF-DRG at 38°C, and sham treatment, by measuring expression of immediate early gene c-fos in the dorsal spinal horn neurons; presence of the c-fos immunoreactive neurons indicated neuronal activation.36Three hours after application of these radiofrequency modalities, a significant increase in c-fos immunoreactivity was observed in the pulsed RF-DRG group, compared with the continuous RF-DRG at 38°C and the sham-treated group. The authors concluded that the pulsed RF-DRG activated pain-processing neurons in the dorsal horn; this effect was attributed to higher voltages and electromagnetic force delivered during PRF, and was independent of any tissue heating. Van Zundert et al. ,29in a similar study in a rat model, measured c-fos immunoreactivity at 7 days. The study groups included pulsed RF-DRG for 120 s, pulsed RF-DRG for 8 min, continuous RF-DRG at 67°C for 60 s, and a sham-treated group. An equal increase in c-fos immunoreactive neurons was observed in the various treatment groups, irrespective of the radiofrequency modality, whereas no such increase was seen in the sham-treated group. The authors concluded that the results of this study showed late or sustained effect of PRF on the DRG. Also in a rat model, Hamann et al.  30measured activating transcription factor 3 in the DRG; an increase in the activating transcription factor 3–positive neurons indicated cellular stress.37The study groups included a pulsed RF-DRG group (PRF to the L4 anterior ramus), a sciatic nerve PRF group, an axotomy group (the L4 anterior ramus was transected), and a sham-treated group. No increase in activating transcription factor 3–positive neurons was observed in the sham-treated group or when the PRF was applied distally to the sciatic nerve, a moderate increase was seen in the pulsed RF-DRG group, and a marked increase was observed in the axotomy group. The authors concluded that PRF caused cell stress without overt thermal injury; however, they did not rule out the possibility that individual PRF pulses might generate enough heat to cause the cellular stress.

In 2 studies (table 7), DRG morphology was observed after exposing them to various radiofrequency and heat modalities. In a rat model, Podhajsky et al.  31applied four modalities to the DRG that included PRF, continuous radiofrequency (CRF) at 42°C, CRF at 80°C, and conductive heat at 42°C (conductive heat probe heated to 42°C). Light microscopic observations made at 2, 7, and 21 days showed that the tissue response to the temperature increased to 42°C was irrespective of the heating modality; each caused edema at 2 days that persisted through 7 days and was resolved by 21 days. In contrast, lesions made at 80°C consistently caused thermal lesions, characterized by wallerian degeneration. In a similar study, Erdine et al.  32exposed rabbit DRG to PRF and CRF at 67°C; the study also included a sham-treated group (electrode placement on the DRG but no radiofrequency applied) and a control group (no intervention). At 2 weeks, there was no difference in the light microscopic appearances of the DRG in all of the above groups, and the electron microscopic appearance showed no pathologic changes in the DRG of the sham-treated and control groups. However, the electron microscopic appearance in both the CRF at 67°C and PRF groups showed damage to the cellular substructure, which was greater in the CRF at 67°C group. The authors suggested that PRF was more destructive than CRF at 67°C.

In the first publication of percutaneous radiofrequency application to the sensory spinal nerve roots, Uematsu et al.  23described the technique of cervical, thoracic, and lumbar dorsal rhizotomy. They placed a 17-gauge radiofrequency electrode in the respective intervertebral foramen (IVF) under fluoroscopic guidance; proximity to the target spinal nerve root was further facilitated by the electrical stimulation—1-ms electrical pulses at 2 Hz, between 0.5 and 2.0 V, were used to elicit pain and muscle contractions in the appropriate dermatomes. Subsequently, Sluijter, Koetsveld-Baart, and Mehta10,38described their technique of electrode placement, which varied considerably from the one introduced by Uematsu et al.  23To improve patient tolerance, thinner electrodes in the range of 22–23 gauge were used. For precise electrode tip positioning under fluoroscopic guidance, detailed location of the DRG in relation to the x-ray imaging was described: On anteroposterior x-ray projection, the DRG was described to lie immediately behind the lateral aspect of the facet column at all spinal levels; on lateral x-ray projection, it was localized to the dorsocranial quadrant of the IVF in the lumbar and thoracic region, and dorsocaudal quadrant of the IVF in the cervical region. The criteria used for electrical stimulation by these authors were also different: High-frequency electrical currents between 50 and 100 Hz were used to elicit paresthesia, whereas low-frequency currents between 2 and 5 Hz were used to elicit the motor response. To ensure proximity to the DRG, the sensory threshold was kept less than 0.6 V, whereas proximity to the motor nerve root was avoided by keeping the motor threshold greater than 1.5–2 times the sensory threshold. To delineate the DRG position in relation to the electrode tip and to exclude the intradural electrode positioning, a radiculogram before the radiofrequency lesioning was also obtained. Later, Van Kleef et al.  19recommended using greater than 0.4 V of electrical currents during sensory stimulation to avoid intraganglionic electrode placement. The majority of the subsequent studies of both continuous and pulsed RF-DRG adapted the technique of electrode placement described by these authors with little variation.

Except Uematsu et al. ,23who performed lumbar RF-DRG in the lateral recumbent position, the prone position was used for both lumbar and thoracic RF-DRG by most authors (fig. 2). An anterolateral approach, with the patient in the supine position, was used in all of the studies of cervical RF-DRG (fig. 3). Using anteroposterior and lateral fluoroscopic views, the radiofrequency electrode was advanced superomedially from a lateral position in most of the early studies of lumbar and thoracic RF-DRG.9,13–15,18,38An oblique fluoroscopic view that allowed maximum visualization of the IVFs, with coaxial advancement of the electrode, was used in the later studies of lumbar7and thoracic13,21RF-DRG. Almost all of the studies of the cervical RF-DRG used oblique fluoroscopic views, with coaxial advancement of the electrode.4–6,10–12,14,20,38Haspeslagh et al.  6further elaborated the technique of their oblique view in the cervical region; they aligned the vertebral disk plates and rotated the C-arm in the oblique position until the contralateral pedicles were noted to be projecting posterior to the anterior line of the vertebral bodies. Contact was made with a bony landmark before the IVF was entered in almost all the studies. In the lumbar region, this landmark was either the junction of the transverse process and the facet column14,38or the vertebral lamina24; in the thoracic region, it was the junction of the transverse process and the caudal portion of the vertebral body23; and in the cervical region, it was the posterior and caudal border of the IVF.4–6,10–12,20,38In almost all of the studies, after this bony contact was made, the electrode was advanced in the anteroposterior projection, and the target for the electrode tip was the mid-facet column4–6,10–12,20; this target was further described as either a line connecting the mid-facet joints14,38or the inferomedial aspect of the superior pedicle bounding the IVF.13,24On the lateral x-ray projection, the electrode tip was placed in the dorsocranial quadrant of the IVF in the lumbar and thoracic region and in the dorsocaudal quadrant of the IVF in the cervical region. Because of difficult access from the corresponding IVF, the technique of RF-DRG for the first sacral (S1) and upper thoracic (T1–T7) segments was different from the one described above. The S1 DRG was found to be located significantly higher and medial to the S1 IVF,14,38and the upper thoracic DRG (T1–T7) were deemed inaccessible because of the presence of ribs and pleura.13At these sites, the respective DRG was first identified by obtaining a radiculogram through the corresponding IVF, and the electrode was placed through a burr hole drilled in to the osseous structure directly dorsal to the DRG.13,14 

During CRF, the electrical current density is greatest around the electrode tip, and the radiofrequency lesion generated is elliptical, with its long axis formed by the uninsulated electrode tip. Therefore, to effectively coagulate the target nerve during CRF application, the electrode tip is typically placed parallel to the long axis of the target nerve.39During PRF application, however, the greatest electrical current density is distal to the electrode tip,9and placing the electrode parallel to the target nerve is deemed unnecessary. Despite these differences, however, no difference was found between the techniques of electrode placement for pulsed and continuous RF-DRG.

Uematsu et al.  23modeled their technique after surgical rhizotomy40and primarily targeted the sensory spinal nerve roots. However, later publications emphasized direct lesioning of the DRG, arguing that an extraganglionic radiofrequency lesion would be no different than peripheral nerve lesioning.10,14,24,38To avoid deafferentation symptoms, Van Kleef et al.  19proposed lesioning adjacent to and not inside the DRG. The segmental level for radiofrequency application was identified by a single set of diagnostic nerve blocks in most of the studies, and comparative blocks as a diagnostic criterion were not used. In two studies,17,22no diagnostic nerve blocks were used, and the treatment levels were determined entirely by clinical and imaging findings. To avoid motor disturbances and deafferentation symptoms, as observed after the surgical sectioning of multiple sensory spinal nerve roots,40treatment at a single segmental level was advocated by majority of the authors.4,5,7–10,12,14,19Citing overlapping innervation of the adjacent dermatomes,15,41radiofrequency was applied at multiple segments in several of the studies: CRF was applied at two to seven levels by Uematsu et al. ,23one to five levels by Nash,24and one to three levels by Niv and Chayen,15whereas PRF was applied at one to three levels by Teixeira et al. ,21Munglani et al. ,25and Rozen et al.  26 

The electrode tip was heated to 67°C in the majority of the studies of continuous RF-DRG4,5,7,11–13,18,19; the electrode temperature selected was 70°C in 2 studies14,15and 75°C in 1 study.10The electrode temperatures were used variably in studies by Uematsu et al.  23(50°–70°C) and Nash (70°–80°C).24The duration of continuous RF-DRG varied, and both 60 s4,10,12,14,19and 90 s5,7,11,13,15,18were selected with almost equal frequency; in 2 studies, continuous RF-DRG was applied for 120 s.23,24The majority of the studies of pulsed RF-DRG used the protocol described by Sluijter et al.  9: Radiofrequency current was applied for 20 ms, at 2 Hz, for 120 s, with maximum tissue temperature increased to 42°C. In 2 studies, the duration of pulsed RF-DRG application was longer than 2 min; Teixeira et al.  22and Cohen et al.  21applied PRF for 3 and 8 min, respectively.

Pain and dysesthesias, which resolved spontaneously in a few weeks to months, were reported in a small number of treated patients in several of the studies of continuous RF-DRG.4,5,10–15,19,24Transient sensory loss in the treated dermatomes was also reported in some studies of continuous RF-DRG.4,10,12,13,19The continuous RF-DRG study by Slappendel et al.  5was the only study that reported motor disturbances, minor weakness in hand strength that was present at 3 months in 3 of 61 patients. Aside from minor immediate postprocedural pain,9,25none of the studies of pulsed RF-DRG reported any significant side effects or complications.

Surgical sectioning of the sensory spinal nerve roots for the relief of pain was first reported as early as late 19th century.42The associated dysesthesias, deafferentation symptoms, and loss of function, however, led to gradual abandoning of these techniques. In the 1930s, neurolytic agents, alcohol,43and phenol44were introduced to chemically destroy the sensory spinal nerve roots, but the unpredictability of the neurolytic effects limited their use to terminally ill patients. In 1974, after the use of thermal radiofrequency for trigeminal neuralgia,45Uematsu et al.  23percutaneously applied radiofrequency to the sensory spinal nerve roots to create a controlled thermal lesion that would safely interrupt the afferent spinal pain pathways. Based on the observations of Brodkey et al. ,46which showed that temperatures above 45°C caused tissue destruction, Uematsu et al.  23increased the electrode temperatures above these levels by applying the radiofrequency currents uninterruptedly— continuous RF-DRG. For selective analgesia, however, these and several subsequent investigators10,14,23,24relied on the findings of Letcher and Goldring,47which showed selective destruction of the small pain fibers but sparing of the larger sensory and motor fibers at the peripheral zones of the thermal radiofrequency lesions. Optimal electrode tip temperatures and lesioning durations were therefore sought in several of the early studies of RF-DRG. In 1991, Vervest and Stolker11first quoted the findings of Smith et al. ,48which reported uniform nerve fibers damage in the radiofrequency lesions created between 45° and 75°C. In 1998, Sluijter et al.  9applied PRF to the DRG. During pulsed RF-DRG, thermal tissue injury was avoided by limiting the peak electrode temperature to 42°C or less, and radiofrequency currents were applied at higher voltages to maximize the delivery of electromagnetic force. These two opposing goals were achieved by applying the radiofrequency currents in a pulsatile manner, and short heat bursts (20 ms) were interspersed by relatively long cooling periods (480 ms), allowing time for the heat to dissipate in between the radiofrequency pulses.

Identified as an enlargement on the dorsal spinal root and located at a variable distance from its takeoff from the central thecal sac, the DRG contain the cell bodies of the afferent spinal nerves. In the lumbar region, the DRG are classified as intraspinal, intraforaminal, and extraforaminal, based on their location in relation to the boundaries of the IVF. The majority of the lumbar DRG were found to be intraforaminal, except the S1 DRG was found to be intraspinal in 80% of individuals.49The size of the lumbar DRG also progressively increases from the first lumbar to the S1 level. In the cervical region, the DRG have been classified as proximal or distal based on their location, proximal or distal to the interpedicular line. Although cervical DRG were found at a progressively greater distance from the thecal sac, from upper to lower levels, no clear trend was found for the proximal or distal types.50 

Radicular pain has traditionally been attributed to compression of the spinal nerve root by a herniated disk, causing antidromic spread of impulses along the peripheral nerve.51Compression neuropathies, however, are often painless,52and in experimental conditions, sustained response was not generated by direct compression of the sensory nerve roots.53These observations cast doubts on the assumption that the sensory nerve root compression was the main cause of radicular pain. Spontaneous54and enhanced55DRG activity, however, was observed in response to the injury in experimental conditions. In addition, inflammatory mediators released at the site of herniated disk material have been shown to modulate the Na⁺, K⁺, and Ca²⁺ion channels on the DRG surface, causing its ectopic and sustained firing.3Such sustained DRG discharges have also been linked to sensitization of the spinal dorsal horn cells and the resulting state of hyperalgesia.56Based on these observations, the DRG are considered the most likely focus of ectopic impulse origin in patients with radicular pain, and the prime target for neurodestructive and neuromodulatory pain treatments.

The proposed mechanism of pain relief from RF-DRG is neuronal dysfunction and/or nerve fiber damage, resulting in interruption of the afferent nociceptive impulses. However, none of the clinical or laboratory studies we reviewed provided evidence supporting this assumption. The basic science study by De Louw et al.  27reported small (< 2.0-mm) intraganglionic thermal lesions, or larger extraganglionic lesions and a possible DRG cell injury response; these findings do not equate to interruption of the nociceptive impulses and the consequent pain relief. Van Kleef et al.  12recorded sensory evoked potentials and electromyography 1 week before and at 3–4 weeks after the continuous RF-DRG and reported no difference in the results, supporting only the assumption that continuous RF-DRG spared the large sensory and motor nerve fibers. Because no thermal lesion is created during PRF, its mode of action is even less obvious. The proposed mechanisms for PRF effect have ranged from cellular dysfunction from high electromagnetic fields9,57and heat bursts30,57to neuromodulation.58The experimental studies by Higuchi et al. ,28Van Zundert et al. ,29and Hamann et al.  30showed that pulsed RF-DRG caused neuronal activation. The study by Erdine et al.  32also suggested that PRF can cause damage to the cellular substructure of the DRG. However, it is unclear whether either the neuronal activation or the structural cell damage was caused by the transient heat bursts or the high electromagnetic fields generated during the PRF application.30,57In addition, the significance of these findings in relation to the interruption of nociceptive impulses remains unclear.

Although Uematsu et al.  23targeted the sensory spinal nerve roots, majority of the later authors recommended direct (intraganglionic) lesioning of the DRG for the following reasons: (1) to make denervation more permanent, by targeting the neuronal cell bodies; (2) to make denervation more complete, by targeting the sensory fibers in the anterior spinal roots; and (3) to reduce the incidence of postprocedural neuropathic pain, by preventing neuronal hyperactivity by direct DRG lesioning.10,14,24,38These assumptions were supported only by the experimental findings of DRG cell hyperactivity observed after peripheral nerve lesions59and the presence of sensory fibers in the motor nerve roots.60Although the two techniques of RF-DRG are deemed distinct, both continuous and pulsed RF-DRG were used to treat similar pain syndromes. Only 1 clinical study compared continuous with pulsed RF-DRG,9and even this study used a mode of continuous RF-DRG—continuous RF-DRG with maximum temperature of 42°C—not used in routine clinical practice. Despite their high false-positive rates,61a single set of diagnostic spinal blocks was used almost exclusively, and none of the studies we reviewed used comparative diagnostic blocks.

Although the majority of the uncontrolled clinical studies of RF-DRG reported the efficacy of both continuous and pulsed RF-DRG, results of the controlled clinical trials varied. Of the 2 RCTs of continuous RF-DRG in patients with cervicobrachial pain, the trial by Van Kleef et al.  4scored 4 out of 5 points (table 8) on the 5-point Jadad scale—it lacked the description of random patient allocation—and was graded as a high-quality trial; it reported short-term efficacy of the continuous RF-DRG in the treatment of cervicobrachial pain. The second trial, by Slappendel et al. ,5scored 5 out of 5 points on the Jadad scale (table 8) and was graded as a high-quality trial. This trial concluded that the two treatment groups (RF-DRG at 67°C and at 40°C) had equal clinical efficacy; the efficacy of either of the groups, however, is unclear. In addition to the lack of a placebo control group, the overall pain relief for the two groups was not significant; at 3 months in either group, less than 52% of the patients experienced clinically significant pain relief (> 2 points on the VAS), and less than 12% of the patients had complete pain relief. Based on the studies by Brodkey et al. ,46the effects of 40°C lesion on nervous tissue should be minimal and reversible. Therefore, if one of the study groups (40°C) were considered as a placebo group, the other would be shown to have no clinical effect. The results of this trial therefore do not provide conclusive evidence of the efficacy of continuous RF-DRG. Using the best evidence synthesis method (table 1), with positive results in one high-quality RCT, there is thus level C or limited evidence of short-term efficacy of continuous RF-DRG in the treatment of cervicobrachial pain (table 9).

The only RCT of RF-DRG use in the treatment of cervicogenic headaches was by Haspeslagh et al.  6This trial concluded that the sequential continuous radiofrequency treatments of the facet joints and the DRG were as effective as the local nerve blocks in the treatment of cervicogenic headaches. Because of the lack of appropriate randomization and blinding, this trial scored only 2 out of 5 points on the 5-point Jadad scale and was graded as a low-quality trial (table 8). In addition to the lack of a placebo control group, the results reported were of the combined efficacy of facet joint radiofrequency and continuous RF-DRG. With only 3 of 15 patients in the treatment group receiving the continuous RF-DRG, this trial did not directly evaluate the efficacy of continuous RF-DRG, and its results were therefore regarded as inconclusive. With inconclusive results in one low-quality RCT trial, there is therefore level D or inconclusive evidence (table 1) of the efficacy of continuous RF-DRG in the treatment of cervicogenic headaches (table 9).

The trial by Geurts et al.  7was the only RCT of RF-DRG use in the treatment of lumbar radicular pain. It reported that continuous RF-DRG was not effective in the treatment of chronic lumbar radicular pain and recommended against its routine use. This trial scored 5 out of 5 points on the Jadad scale and was ranked as a high-quality trial (table 8). With negative results from one high-quality RCT, there is thus level C or limited evidence (table 1) against the use of continuous RF-DRG in the treatment of lumbar radicular pain (table 9).

The trial by Van Zundert et al.  8in the treatment of cervicobrachial pain was the only RCT of pulsed RF-DRG encountered. With different pretrial characteristics of the two study groups, this trial reported only possible short-term clinical efficacy of pulsed RF-DRG in the treatment of cervicobrachial pain. This trial scored 5 out of 5 on the 5-point Jadad scale and was graded as a high-quality trial (table 8). With possible positive results in one high-quality RCT, it provided level C or limited evidence (table 1) of possible short-term efficacy of pulsed RF-DRG in the treatment of cervicobrachial pain (table 9).

The two primary modalities of radiofrequency application to the DRG used in the clinical practice include continuous RF-DRG, with electrode temperatures in thermodestructive range, and pulsed RF-DRG. Although the uncontrolled studies reported the clinical efficacy of both continuous and pulsed RF-DRG, the controlled clinical data provided results that were variable depending on the pain syndrome being treated and the mode of RF-DRG used. For continuous RF-DRG, limited evidence of short-term efficacy existed in the treatment of cervi-cobrachial pain, the evidence was inconclusive in the treatment of cervicogenic headaches, and limited evidence against its use existed in the treatment of lumbar radicular pain. For pulsed RF-DRG, limited evidence of possible short-term efficacy existed in the treatment of cervicobrachial pain. The complications reported from continuous RF-DRG were limited mainly to sensory disturbances that were infrequent and self-limiting, and no notable complications of pulsed RF-DRG were reported. Although proximity to the DRG was sought in all of the studies of RF-DRG, its exact target, the optimal number of treated segments, and the preferred mode, whether continuous or pulsed radiofrequency, are not clear. The mechanisms of action of both pulsed and continuous RF-DRG remain poorly understood. More basic science research and larger, long-term outcome, controlled clinical trials are needed to clearly understand the efficacy and the mechanism(s) of action of RF-DRG.

The authors thank Naveen Nathan, M.D. (Instructor in Anesthesiology, Northwestern University, Evanston, Illinois), for preparing figures 2 and 3, and Douglas Taylor, M.D. (Resident in Anesthesiology, Northwestern University), for help with data collection.