Between October 2003 and October 2004, a total of 28 patients underwent RF neurotomies for their cervicobrachialgia in our institution. Patients, who had chronic neck pain with or without referred arm pain more than 3 months, whose pain intensity was greater than 5 in visuoanalogue score (VAS), and whose primary source of pain was confirmed to be cervical zygapophyseal joints by comparative diagnostic blocks, were indicated for RF neurotomies. Thorough clinical and neurological examinations were conducted and simple radiographies were routinely performed. We paid special attention to pain-provocation and its radiation to the corresponding distribution of pain complaints by the light pressure on the cervical zygapophyseal joints, while pain responses were evaluated with the assumed level of cervical zygapophyseal joint arthropathies. All patients were interviewed by a pain-specialized nurse, who recorded pain distribution pattern, nature of pain, pain-related dynamic factors, and factors influencing the life pattern and daily activities. Pain distribution patterns were divided into type A, B, and C (Fig. 1); Type A indicates the pain distribution in the lower neck; Type B demonstrates the pain in the neck and shoulder; Type C with the pain in lower neck, shoulder, sometimes scapula, and upper arm. Excluded are the patients who had objective evidence of cervical radiculopathy with signs of myotomal weakness, sensory disturbances, and loss of reflexes. Patients who had collagen, vascular, or rheumatologic diseases were also excluded.

All patients had diagnostic blocks in the operating room under C-arm intensifier. To decrease the false positive rate, diagnostic blocks were performed more than twice in each patient. Targets of diagnostic blocks were determined mostly by pain distribution patterns. C4, C5, and C6 medial branches were chosen for type A and B pain distribution, and C7 was additionally included for type C. If patients showed pains on inner scapular area, medial branch of T1 was included. If patients showed bilateral pain distribution, targets were determined in both sides. Comparative blocks were performed in all the patients. Those who responded to first block and began to have similar degree of pain again were encouraged to have second block after at least one week. Those whose pain were not relieved by first block went on to other investigations. Those whose pain was markedly improved were not included in this study.

Nerve blocks were performed in a group (e.g., C4, 5, 6 and 7 together) rather than a single trial. Targets for C4, C5, and C6 were the centroid of the articular pillar. Target for C7 was supero-lateral aspect of C7 transverse process (Fig. 2). Target for C8 was supero-lateral aspect of T1 transverse process. After introducing the needle at the targets, and contrast dye was injected to confirm that dye is not drained into the vascular space. If the needle was confirmed to be placed in the target on slightly oblique view, 0.5 mL of 0.5% bupivacaine or 4% lidocaine was used for the first block. As for second block, local anesthetic agent which was not used for their first block were used. After injecting local anesthetics, patients were instructed to mark the changes in their usual pain in ''pain evaluation sheet'' as complete relief, nearly complete relief, 50% reduction of pain, slight reduction of pain, and no change every 1 hr except sleeping time for 3 days. If the patients demonstrated concordant pain relief greater than '50% of pain reduction' on comparative blocks, they were considered the candidates for RF neurotomies.

Only patients who began to experience the return of symptoms after second block returned to RF procedure. If patients' pain was markedly relieved by provocative blocks, they were discouraged to undergo RF procedures.

RF neurotomies were performed in an aseptic manner in the operating room. Patients were placed prone on the table with electrocardiography and blood pressure monitoring. C-arm fluoroscopic unit (OEC, U.S.A or Madison, Korea) was used to visualize the targets. To localize the ''pathological branches'', which are responsible for pain generation, the sensory-stimulation method was adopted. All cervical medial branches included in the diagnostic blocks were subject to sensory stimulation using lesion generator (Leibinger N50, Germany). Twenty one gauge of RF cannula of 100 mm long with 5 mm active tip (Precision Medical Engineering, U.S.A.) was used. RF cannula was advanced parallel to the plane of zygapophyseal joint in C4 - 6 medial branches to locate the needle in the center of waist of articular pillar. Using 50 Hz frequency of stimulation with 1 msec duration, stimulation on every medial branch was performed to locate the pathological branches. Stimulations elicited tingling sensation, electrical pain, or feeling of heaviness. If pain response was same as that of usual pain with stimulation lower than 0.6 V, the cannula was maintained for RF coagulation of the branches. In addition, absence of motor fasciculations in corresponding upper extremities was confirmed with 2 V at 2 Hz stimulation. RF neurotomies were performed upon confirming the pathological branches. After introduction of 1 mL of 2% lidocaine, RF electrode was inserted through RF cannula. Two targets were selected along the course of medial branch for lesioning (Fig. 3). First target was branch on the curving point of waist of the articular pillar on posterior-anterior projection. The second lesion was made after introducing the cannula deeper in the centroid of the second target in the articular pillar on lateral view. To avoid the escape of medial branches from extent of coagulation, additional lesions were made above and below the targets. RF lesion was created on medial branches at 80℃ for 90 sec. After RF Lesioning, 1 mL of mixture of 0.5% bupivacaine and 8 mg/mL of triamcinolone was injected on the lesioned area to decrease postoperative pain and neuritis.

Patients were followed up for a minimum of 6 months up to 12 months after RF neurotomies. Patient outcome was measured with VAS score at 1, 2, 3, 6, 9, and 12 months after the operations. Outcome was arbitrarily graded as a successful outcome (more than 75% improvement), moderate improvement (between 50 and 75% improvement), and failure (less than 50% improvement). Categorical variables were assessed and compared by using chi-square, Pearson correlation, and ANOVA by SPSS statistical software program.

Table 1 summarizes the clinical characteristics. Patients consisted of 15 male (54%) and 13 female patients (46%). Age at the time of evaluation and surgery ranged from 36 to 83 yr (mean 54.4 yr). Duration of pain prior to surgery ranged from 3 months to 40 yr (mean 2.9 yr). Preoperative severity of pain ranged from 5 to 10 (mean 6.6) in VAS. Analysis of pain distribution pattern showed type A in 6 (21%), type B in 6 (21%), and type C in 16 (57%) (Fig. 1). Five patients had history of whiplash injuries, and one patient had penetrating injury in the neck. Twenty patients had unilateral pain (9 in right side and 11 in left side), while 8 patients had bilateral pain.

Controlled block was performed at least twice in all patients. The third test was required in two patients, because the result of the second block was questionable. Interval between first and second blocks was 16.5 days. Pain distribution pattern was not significantly correlated with the results of diagnostic blocks in the study. All patients obtained significant (≥50% at least) pain relief on each controlled block to be potential candidates for RF neurotomies. Meanwhile, C4-C6 levels were most commonly selected for the targets of diagnostic blocks. Four patients showed incongruent level of pain relief on two consecutive diagnostic blocks.

All patients underwent electrical stimulation on each medial branch. Average stimulation intensity was 0.45 V, ranging from 0.3 to 0.69, to elicit pain response in cervical medial branches. Average number of lesioning was 8 per patient. Number of RF-lesioned nerve branches was 6 per patient. Most common involvement of nerve branches was C4 (89%), followed by C5 (82%), C6 (75%), and C7 (43%).

Postoperative pain relief was relatively constant until 6 months once patients obtained improvement by RF procedures. However, one patient among 28 patients showed the recurrence with same pain distribution between 6 and 9 months, and three patients recurred between 9 and 12 months. However, their degree of recurrent pain were not same as preoperative level, 64% of postoperative pain relief, to justify repeating the procedure. Interestingly, 6 patients out of 28 patients showed only modest decrease (32% improvement) of VAS. Increased physical activity after decrease of pain might contribute to the additional decrease of VAS in these patients.

Nineteen patients (68%) reported successful outcome according to outcome criteria after 6 months of follow-up (p=0.001), and among them, eight patients (29%) reported complete (100%) relief of pain. Two patients reported failure. Five patients who showed incongruent level of pain relief on two consecutive diagnostic blocks had unsuccessful outcome. These outcomes were relatively maintained until 9 months. However, patients with failure category were increased due to recurrences of pain 12 months after the procedure. Outcome was not significantly influenced either by the side or level of operation.

Among five patients with whiplash injury, three obtained complete relief of pain. One patient had 60% of relief of pain of preoperative level, yet pain relief was enough not to repeat the procedure. One patient who had history of whiplash injury showed earlier recurrence of pain 3 months after RF procedures, despite successful outcome in early period. He was advised to repeat diagnostic procedure but refused to do so.

Postoperatively, patients showed side effects including muscle cramping pain and numbness which lasted one or two weeks in most cases, but no longer than four weeks. Taking analgesic drugs for several days was enough to control these postoperative side effects. No patients were admitted for intensive control of post-operative pain. Complications such as motor deficit, infection, neuropathic pains, were not encountered after the procedures in our series.