Abstract
Background: Approximately 0.9 percent of the white adult population of the United States and 1.1 percent of the adult population in Europe are affected by seropositive rheumatoid arthritis. As many as 10 percent of those patients may need an operation for atlantoaxial subluxation. Severe instability, especially when associated with vertical subluxation of the odontoid process, can result in progressive cervical myelopathy. Typically, occipitocervical fixation has been performed for these patients with use of autograft bone to achieve long-term stability through a solid fusion. Harvesting the bone graft increases the operative risk to the patient and may result in increased morbidity. In our experience, patients who have had no clear radiographic evidence of fusion following use of occipitocervical instrumentation seemed to have done as well as those who have had obvious fusion. One assumption is that the clinical improvement might be attributable simply to stabilization of the joint rather than to osseous fusion.
A longitudinal study was performed on patients with rheumatoid arthritis who required an operation because of craniocervical or upper cervical instability.
Methods: The results of clinical, radiographic, functional, and self-evaluations were studied to determine the efficacy of treatment and to compare the outcomes of bone-grafting with those of procedures done without bone-grafting in a group of 150 patients who underwent posterior occipitocervical stabilization with use of a contoured metal implant (a Ransford loop) that was affixed by sublaminar wires. Internal fixation was performed in 120 patients without bone-grafting and in thirty patients with use of autogenous bone-grafting.
Preoperatively, 23 percent (thirty-five) of the 150 patients had mild neurological involvement (class II, according to the system of Ranawat et al.), 45 percent (sixty-eight) had objective findings of weakness and long-tract signs but were able to walk (class III-A), and 29 percent (forty-three) were quadriparetic and unable to walk (class III-B). The age of the patients at the time of the operation ranged from twelve to eighty-three years (mean, sixty-two years).
Results: There were significant improvements in postoperative Ranawat classes at all time-periods (range, p < 0.00005 to p = 0.0066) and in patient ratings of neck pain (range, p < 0.00005 to p = 0.0044) compared with preoperative scores. With the numbers available, there were no significant differences between the patients managed with a graft and those managed without grafting with respect to survival after the operation, Ranawat class, head or neck-pain rating, presence of subaxial abnormalities, radiographic craniovertebral motion, or vertical subluxation. Overall mortality at one month was 10 percent (fifteen of 150), although this value varied directly with the degree of preoperative disability. A second cervical spine operation was required in 11 percent (sixteen) of the 150 patients.
Conclusions: While patients who have rheumatoid disease with anterior atlantoaxial subluxation should be treated with posterior atlantoaxial arthrodesis with use of bone-grafting and internal fixation, we believe that those who present with vertical instability and multilevel involvement can be treated with posterior occipitocervical stabilization with use of a contoured occipitocervical loop and sublaminar wire fixation without bone-grafting. Furthermore, we believe that the use of preoperative traction, bone cement, or a postoperative halo vest is unnecessary. Avoiding the harvesting of autogenous bone for grafting reduced the morbidity of this operation without compromising the outcome in these already sick patients.
The prevalence of atlantoaxial subluxation in patients with rheumatoid arthritis varies with the population studied, but the published numbers for temperate climates in the Northern Hemisphere indicate that as many as 220,000 persons in the United States and 62,700 persons in the United Kingdom may need an operation13. Approximately 0.9 percent of the white adult population of the United States and 1.1 percent of the adult population in Europe are affected by seropositive rheumatoid arthritis. The prevalence of atlantoaxial subluxation was reported to be five (5 percent) of 100 within two years after the onset of rheumatoid arthritis, and it appears to increase with time48. Progressive atlantoaxial instability increases the risk of neuraxial compression. In a population survey of 1478 patients, Sharp and Purser found seventy-eight patients (5.3 percent) with clinical evidence of rheumatoid arthritis43. Of the seventy-eight, five (6 percent) had atlantoaxial subluxation. In the same survey, fourteen (18 percent) of another seventy-nine rheumatoid patients who were admitted to the hospital in one year had atlantoaxial subluxation. Pellicci et al. evaluated 163 rheumatoid patients with adequate records and found that forty-one (25 percent) had atlantoaxial subluxation at the start of the study and sixty-four (39 percent) had subluxation after a minimum duration of follow-up of five years (mean, 6.1 years)34. Eight percent of the patients had radiographic evidence of deterioration over the course of the study. Neurological deterioration, operative treatments, and deaths were reported in this group and in other series27,44. Sunahara et al. reported that 76 percent of twenty-one patients with rheumatoid arthritis and myelopathy due to atlantoaxial subluxation deteriorated neurologically; none improved, and all were bedridden within three years after the diagnosis of the myelopathy46. In that study, the cumulative probability of survival was 0 percent in the first seven years after the onset of myelopathy.
Typically, occipitocervical stabilization has been performed with the use of autogenous bone graft to achieve long-term stability by means of a solid fusion. Many adjunctive methods have been employed to enhance the fusion rate and to prevent development of a nonunion or a pseudarthrosis. Internal fixation, the use of polymethylmethacrylate bone cement, and postoperative stabilization with a halo vest or skeletal traction have all been utilized. The combination of bone and polymethylmethacrylate, which was pioneered in Lund, Sweden3, has allowed fairly rapid mobilization, and the results over a generation, certainly in the hands of experts, have been gratifying6,7,18,30,31. Preoperative skull traction has been widely used in an attempt to reduce cranial settling, or vertical migration of the odontoid process, which occurs in advanced rheumatoid arthritis affecting the upper cervical spine7,14,28,29,32,36,49. In our experience, traction has not reliably reduced vertical subluxation of the odontoid process. If reduction is achieved, the anterior column support is effectively eliminated (Figs. 1-A and 1-B). Therefore, large forces must be overcome by the posterior fixation and eventually by the arthrodesis, which is often performed in poor-quality bone.
Fixation techniques involving rods and wire permit segmental fixation of the spine while allowing some intersegmental motion. Resina and Alves developed a technique for segmental spinal fixation by fixing a rod to the spinous processes with the use of wires41. Luque, from Mexico City, pioneered the use of segmental sublaminar wire fixation, which provided better control and correction of scoliosis20,21. With this segmental spinal fixation, external support could be minimized or avoided, which permitted its use without bone-grafting in growing children with spinal deformities22,24. Implants consisting of a single rod or a rectangle eliminated the so-called play between the two rods and provided improved rotational control23. Analogous stainless-steel implants (Ransford loops) have been developed for use across the craniocervical junction and have proven efficacious for craniocervical arthrodesis in nonrheumatoid patients (Fig. 2)26,37.
Despite the improved operative and technological measures just described, some patients with rheumatoid arthritis do not obtain a solid fusion. Harvesting the bone graft increases the operative risk to the patient and may result in increased morbidity. The use of a halo vest for external fixation is fraught with dangers, including pressure sores and skin slough. The detrimental effects of recumbency are well known, so forced bed rest in traction should be avoided.
In addition, it is often very difficult to determine the presence of osseous fusion on the basis of plain radiographs of patients with rheumatoid involvement. It is against this background that we present our experience over a twelve-year period in which we began with a conventional approach of fixation and bone-grafting. During the early years of this period, we were vexed by complications at the donor site, and we observed that patients without clear radiographic evidence of fusion seemed to do as well as those with an obvious fusion. Therefore, we questioned whether an osseous fusion was absolutely necessary for an adequate clinical outcome. Our conclusion was that much of the clinical improvement might have been attributable simply to the joint being stabilized rather than to osseous fusion. Since patients with rheumatoid craniocervical and upper cervical instability generally do not make the same demands on the skeleton as do active, nonrheumatoid age-matched individuals, it seemed that bone-grafting to achieve permanent stability might not always be necessary for a satisfactory outcome.
This study involved a group of 150 patients with rheumatoid arthritis who needed occipitocervical stabilization to correct instability. The patients were a subset of 247 patients with rheumatoid arthritis who were evaluated in a uniform fashion according to a set protocol and were operatively treated for instability during an eight-year period. All 150 patients underwent posterior occipitocervical stabilization with use of a contoured metal implant and sublaminar wire fixation with or without autogenous bone-grafting. The outcomes of clinical, radiographic, functional, and self-evaluations were studied to determine the efficacy of treatment and to compare the outcomes of patients who had had bone-grafting and those who had not. Data were not obtained by chart review; rather, they were collected concurrently with patient enrollment and follow-up.
*One or more of the authors has received or will receive benefits for personal or professional use from a commercial party related directly or indirectly to the subject of this article. Funds were received in total or partial support of the research or clinical study presented in this article. The funding source was the Welton Foundation, London, United Kingdom.
†Department of Orthopaedic Surgery, New York University-Hospital for Joint Diseases, 301 East 17th Street, New York, N.Y. 10003. E-mail address for Ronald Moskovich: ronald.moskovich@med.nyu.edu.
‡Department of Surgical Neurology, The National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N, United Kingdom. E-mail address for H. A. Crockard: a.crockard@ion.ucl.ac.uk.
§Biostatistics Unit, Department of Neurosurgery, Rush-Presbyterian-St. Luke's Medical Center, 1725 West Harrison, Suite 755, Chicago, Illinois 60612. E-mail address for Susan Shott: sshott@rush.edu.
#5 Fordington Road, Highgate, London N6 4TD, United Kingdom. E-mail address for A. O. Ransford: ransford@easynet.co.uk.
We performed a clinical and radiographic longitudinal study of all patients with rheumatoid arthritis who underwent posterior occipitocervical arthrodesis from 1984 through 1991 at The National Hospital for Neurology and Neurosurgery. Although the same type of operation has been performed since 1991, more recently treated patients were not included so that the duration of the postoperative follow-up would be adequate. All patients in this study were followed regularly, both clinically and radiographically.
Patients with atlantoaxial or craniocervical instability secondary to rheumatoid arthritis who were referred to our tertiary referral center for treatment were evaluated according to a set protocol. Patients who were found to have more than five millimeters of reducible atlantoaxial subluxation without neurological compromise were treated with posterior atlantoaxial arthrodesis alone and were not included in this study. All of the patients in this report had irreducible atlantoaxial subluxation, vertical translocation of the odontoid process, anterior neuraxial compression with rheumatoid pannus, or any combination of these abnormalities12. Patients who had subaxial instability in addition to atlantoaxial or craniocervical involvement were also included. Those who had rheumatoid subaxial instability alone were excluded.
Data were collected and entered into a customized relational database by means of a user-friendly interface with extensive error-checking protocols.
Clinical Evaluation
All patients had a comprehensive clinical examination including an extensive neurological examination preoperatively. In addition, all recorded tests and evaluations, both preoperatively and postoperatively, were performed by trained physician investigators. A clinical nurse assisted in managing the database, ensured that patients were not lost to follow-up, and obtained and verified the records of patients who died during the study.
Specific, well validated tools were used to measure functional activity. First, all patients were examined with the test described by Ritchie et al.42, which evaluates joint inflammatory activity by testing the patient's response to passive movement of the cervical spine, hip joints, and talocalcaneal and mid-tarsal joints and their response to the application of firm pressure over other joints. This instrument does not attempt to quantify functional disability, the patient's assessment of his or her own incapacity, or parameters of joint inflammation such as swelling or warmth. The degree of arthritis or joint irritability can be monitored by serial testing.
Second, total body pain and head and neck pain were assessed with visual-analog pain scales completed by the patient.
Third, the Health Assessment Questionnaire was administered with use of a standard form. The Health Assessment Questionnaire is an independent measurement of the functional capacity of patients with rheumatoid arthritis; it produces a reliably repeatable functional index for patients with rheumatoid arthritis17,19.
Fourth, the functional classification described by Steinbrocker et al. (American Rheumatological Association class) was used to evaluate each patient at each contact (Table I)45. The classification assigns one of four classes to patients with rheumatoid arthritis.
Fifth, the classification of neurological function described by Ranawat et al. was also used (Table II)36. This classification divides patients into groups on the basis of their overall neurological and functional status, and, although it is a crude measure, it has the advantage of common usage, enabling some comparison with other published works.
Many other indices and scales, such as the joint alignment and motion scale33, have also been used in the assessment of patients with rheumatoid arthritis, but, since they have not been as well validated as the tools selected for this study, they were not utilized.
Imaging
All radiographic and computerized measurements were made by physicians who were masked to the procedures that had been performed and to the outcomes.
Cervical radiographs, including lateral flexion and extension views, were made initially and then periodically following the operation. The craniocervical relationships and cervical involvement were documented. All patients underwent preoperative cervical myelography, which included dynamic studies in maximal flexion and extension. Although lateral flexion-extension radiograph pairs were not adequate for formal comparison for every patient, plain radiographs were made at all scheduled postoperative visits. It was possible to evaluate the craniocervical alignment and implant integrity in every patient.
The following measurements were recorded for each flexion-extension radiograph pair: the craniocervical angle (of Wackenheim47), the atlanto-odontoid interval, the space available for the cord, the vertical measurement between the base of the ring of the atlas and the base of the body of the axis, the vertical measurement between McGregor's palato-occipital line25 and the base of the body of the axis (Redlund-Johnell measurement38-40), and the cross-sectional areas of the neuraxis at the level of compression as well as of the subaxial cervical spine.
Follow-up radiographs were carefully inspected for evidence of wire and implant breakage. The presence of metal implants sometimes limited measurements, but the craniocervical angle, Redlund-Johnell measurement, and evaluation of the subaxial spine could usually be determined.
It was exceedingly difficult to determine whether or not osseous union was present. Even with fine-section computerized tomography scanning, the artifact due to metal made this determination virtually impossible. Angular motion was therefore measured between the occiput and the second cervical vertebra to determine whether there was any mobility. Wackenheim's line was drawn along the dorsum of the clivus, and another line was drawn along the posterior cortex of the axis vertebra. The open angle subtended by the lines was the craniocervical angle. The difference in this angle between the flexion and extension radiographs provided an indication of the stability of the occipitocervical complex. As flexion and extension radiographs are always made contemporaneously with use of the same tube-to-film distance, the images could be readily compared.
Operative Treatment
Transoral Decompression
For patients who had marked anterior compression of the neuraxis due to irreducible atlantoaxial subluxation, or who had persistent compression from rheumatoid periodontal pannus, transoral resection of the compressive mass was performed without any attempt at anterior bone-grafting or anterior implant fixation (Figs. 3-A, 3-B, 3-C, 3-D, 4-A, 4-B, 4-C, 4-D, 4-E, 5-A, 5-B, 5-C and 5-D). Details of this operative technique have been published elsewhere8-12. Except for the first few patients, who had staged procedures, posterior occipitocervical stabilization was performed with the patient under the same anesthetic. The patient was placed on his or her side with the skull firmly fixed in a Mayfield clamp so that longitudinal distraction and extension could be applied. Operative access was facilitated by tilting the operating table to the left or right. First anterior decompression and then posterior arthrodesis were performed with the patient in the lateral position.
Posterior Fixation
A contoured stainless-steel occipitocervical loop was implanted and affixed to the occiput through paired, full-thickness burr-holes and by means of the sublaminar passage of malleable 18 or 20-gauge stainless-steel wires at all of the cervical levels indicated for stabilization (Fig. 2). The fixation extended at least as caudad as the second cervical lamina and more caudad if there was demonstrable subaxial instability on the preoperative studies. An attempt was made to include the posterior portion of the ring of the atlas in the construct, but, due to the position and fragility of the ring, this was not always possible.
Initially, fixation was performed with a Luque loop (Zimmer, Warsaw, Indiana) or a modified Hartshill rectangle (Surgicraft; Redditch, Worcester, United Kingdom), contoured to the occipitocervical angle during the operation, was used. Experience with these loops led to the development of the Ransford loop (Surgicraft), a customized loop that has an occipital contour as well as a flare at the level of the second cervical lamina, the purpose of which is to prevent subsequent vertical translocation (cranial settling) (Figs. 2, 3-A, 3-B, 3-C, 3-D, 4-A, 4-B, 4-C, 4-D, 4-E, 5-A, 5-B, 5-C and 5-D). No other forms of fixation were used. Polymethylmethacrylate bone cement was not employed in any patient.
Autogenous bone-grafting was performed with the use of posterior iliac-crest bone in the first few years of the study but was not performed in patients later in the study. Neither bone nor bone substitute was used in any of these later procedures.
Postoperative management consisted of early mobilization out of bed and external support, with either a standard Philadelphia collar or a soft cervical collar if the patient could not tolerate the firmer support, for three months. None of the patients were managed with a halo vest.
Statistical Methods
The postoperative follow-up examinations were performed at a mean of 1.5 months (range, zero to 2.9 months), six months (range, three to 8.9 months), and one year (range, nine months to 1.4 years), and then annually thereafter. The preoperative and postoperative Ranawat grades, pain ratings, and American Rheumatological Association classes were compared with the use of the nonparametric paired sign test. This method compares the observed parameter for each patient before and after treatment, rather than comparing means. The paired t test was performed to compare preoperative and postoperative Health Assessment Questionnaire and Ritchie scores. Comparisons of patients who had had bone-grafting with those who had not were done with the chi-square test of association for nominal variables; with pooled-variance or separate-variance t tests for approximately normally distributed variables; with the nonparametric Mann-Whitney test for non-normal ordinal, ratio, and interval variables; and with the log-rank test for survival. A Levene test was done to determine whether the pooled-variance or separate-variance t test should be used. A significance level of 0.01 was used for all statistical tests to reduce the risk of a type-I (alpha) error. SPSS for Windows (SPSS, Chicago, Illinois) was used for all statistical analyses.
All patients were followed on a regular basis or until death. The longest follow-up period was eleven years. The shortest follow-up period was three years for patients who were still alive at the time of the most recent follow-up. The number of patients who were alive at one, two, three, and four years postoperatively was 114, 107, 100, and ninety-two, respectively.
There were thirty-one male patients and 119 female patients, with a mean age of sixty-two years (range, twelve to eighty-three years) at the time of the index operation. The diagnosis of rheumatoid arthritis had been made from five to sixty years (mean, twenty-three years; median, twenty years) prior to the operation. Twenty-three percent (thirty-five) of the 150 patients had mild neurological involvement (Ranawat class II), 45 percent (sixty-eight) were quadriparetic but able to walk (Ranawat class III-A), and 29 percent (forty-three) were quadriparetic and unable to walk (Ranawat class III-B).
A Ransford loop was used in 118 patients, an earlier version of the Ransford loop that had been made on site was used in eleven patients, a Luque loop was used in three patients, and a Hartshill rectangle was used in eighteen patients. The implant was used without bone-grafting in 120 patients. Eighty-seven patients also had a transoral decompression; 50 percent (fifteen) of the thirty patients managed with a graft and 60 percent (seventy-two) of the 120 patients managed without grafting had a transoral decompression (not a significant difference, according to the chi-square test of association; p = 0.32).
A subsequent cervical spine operation was necessary in sixteen patients (11 percent). Five patients needed a second transoral operation. One patient had inadequate decompression. Increased vertical translocation developed in two patients who had had an early operation with the Hartshill rectangle, and therefore another anterior operation was required. These patients presented with progressive spastic quadriparesis. A subsequent operation reversed the situation, and wiring of the second cervical lamina to the lateral flare of a commercially available Ransford loop prevented further translocation. After this early experience with vertical translocation caused by suboptimal posterior instrumentation, there were no further instances of progressive translocation in any patient, whether or not bone graft had been implanted. Two other patients had received an occipitocervical Luque loop; this device also allowed vertical translocation and recurrent anterior compression. The experiences of these four patients confirmed the necessity of shaping the posterior instrumentation to prevent vertical translocation. Two transoral decompression wounds required débridement due to infection. One patient needed additional subaxial anterior decompression. One revision was required because the wires had cut through the bone. Additional posterior procedures were necessary in seven patients because progressive subaxial disease required later stabilization at levels caudad to those that had already been fixed; one patient needed two procedures.
One patient had broken wires three days after the operation, and broken wires were noted in four other patients six months postoperatively. No other broken wires or loops were found at one, two, or three years postoperatively, at which times eighty, sixty-five, and fifty patients, respectively, were seen. Four years after the insertion of a noncommercial prototype loop, one of the loop limbs was noted to be fractured on routine follow-up radiographs. The patient, who was employed full-time as a secretary, was asymptomatic.
Preoperative studies of forty-one patients revealed a wide range of craniocervical motion, from no motion to 62 degrees of angulation between flexion and extension. Approximately half of the patients (51 percent; twenty-one of forty-one) had 10 degrees or less of motion, and 32 percent (thirteen of forty-one) had 5 degrees or less of motion. At all time-periods postoperatively, motion was markedly reduced or frank angular stability had occurred. Eighty-one percent (forty-six) of fifty-seven patients had 10 degrees or less of craniocervical motion at six months postoperatively; 78 percent (thirty-one) of forty, at one year; 85 percent (thirty-four) of forty, at two years; and 84 percent (twenty-six) of thirty-one, at three years (Fig. 6). Patients with a four-year duration of follow-up demonstrated persistent stability, with thirteen of seventeen patients having 10 degrees or less of angular motion. Seventy-four percent (seventy-six) of 103 patients had 5 degrees or less of motion, consistent with ankylosis or fusion, at some point after the operation when follow-up data up to seven years were considered. An additional indication of the degree of stability achieved was the almost complete lack of anteroposterior gliding motion or instability. Long-term craniocervical stability in all patients who did not have satisfactory flexion and extension radiograph pairs was evaluated by comparing measurements of vertical and anterior subluxation and the craniocervical angles on sequential plain radiographs. No cases of instability were identified.
There were no significant preoperative differences between patients who subsequently had bone-grafting and those who did not with respect to age, gender, clinical symptoms or signs, or radiographic findings or measurements (Table III). It is possible that differences between these two patient groups were not detected because the samples were not large enough to ensure adequate power.
In addition, no significant difference with respect to survival after the operation was found between the thirty patients who had had bone-grafting and the 120 patients who had not (p = 0.59). There were also no significant differences between these two groups at the 0.01 significance level with respect to neurological status, pain perception, and radiographic measurements at any postoperative period for which there was sufficient data for analysis (Table IV). The one patient who had broken wires three days after the operation had had posterior bone-grafting; the other patients with broken wires or loops had not had posterior bone-grafting.
Since no significant differences were found between the patients who had had bone-grafting and the patients who had not, the two groups were combined for comparisons of postoperative data with preoperative data. Significant improvement in the postoperative Ranawat classes compared with the preoperative classes was found at six months (p = 0.0002) in ninety-two patients, at one year (p < 0.00005) in sixty-eight patients, at two years (p = 0.0066) in forty-one patients, at three years (p = 0.0001) in forty-five patients, and at four years (p = 0.0003) in twenty-three patients. Of twenty-two patients who had been confined to bed preoperatively, only ten remained so after operative treatment, and five regained the ability to walk independently. Fifteen (38 percent) of thirty-nine patients with moderate-to-severe myelopathy (Ranawat class III-A) had substantial improvement. A detailed comparison of early and late outcomes was done in a manner similar to that of a Frankel outcome assessment of spinal cord injury16 (Figs. 7-A and 7-B). The degree of neurological improvement was clearly maintained in the long term.
Significant decreases in ratings of neck pain (range, p < 0.00005 to p = 0.0044) were found at all postoperative periods up to five years (Table V). Significant decreases in ratings of head pain (range, p < 0.00005 to p = 0.0001) were found up to three years. The p values were 0.019 and 0.039 at four and five years, respectively. The postoperative improvements in neurological function and head and neck pain ratings cannot be attributed solely to a general lessening of the symptoms of the rheumatoid arthritis since there were no significant changes at the 0.01 level in Steinbrocker classes, Health Assessment Questionnaire scores, or Ritchie scores corrected for neck pain (Table V).
The rate of operative mortality (death within thirty days after the operation) was 10 percent (fifteen of 150). The cause of death of three patients was classified as technical complications related to the operation. A seventy-three-year-old man became quadriplegic postoperatively and died later. A review of the anesthetic record revealed that he had become hypotensive and had had bradycardia for five minutes after the passage of one of the sublaminar wires. Although this was the only serious complication that might have been associated with the use of sublaminar wires, this case illustrates the small but potential risk associated with this instrumentation. Placement as far lateral on the laminae as possible reduces the risk of cord impingement when sublaminar wires are passed. Preliminary resection of the odontoid process may be necessary in patients with irreducible atlantoaxial subluxation to increase the space available for wire passage caudad to the posterior arch of the first cervical lamina. A fifty-eight-year-old woman died from spinal cord and medullary compression and bronchopneumonia. A seventy-one-year-old woman was readmitted to the hospital for a tracheostomy eight days after the operation. She became dyspneic and could not be ventilated thirty hours after the tracheostomy. At autopsy, the tracheostomy tube was noted to have perforated the posterior wall of the trachea. One patient died following bowel infarction. The remainder of the deaths had cardiorespiratory causes. Only one patient with Ranawat class-II disease died; all other deaths were of patients who had had substantially more serious neurological involvement.
To our knowledge, this is the largest study of patients with rheumatoid involvement of the cervical spine who were treated operatively. The study focused exclusively on the long-term outcome of operative fixation in a defined group of patients with advanced rheumatoid myelopathy. Our findings should not be extrapolated to young, physically fit rheumatoid patients with reducible atlantoaxial disease, as these patients were not included in the study. However, the techniques described here have been adopted for the management of patients with myelopathy due to multiple cervical spine metastases, with similar satisfactory results.
There are wide variations in the management of patients with end-stage rheumatoid disease and crippling myelopathy (Ranawat class III-A or III-B) as well as occipitocervical instability. It is always difficult to extract exactly comparable data from different centers with different indications for operative management. However, despite the limitations, some information can be obtained from comparison with similar studies (Table VI). In terms of mortality, Menezes et al.29 reported none, but Peppelman et al.35 and Boden et al.2 reported perioperative mortality rates similar to ours (11 percent [four of thirty-six], 11 percent [two of eighteen], and 10 percent [fifteen of 150], respectively). Although a 10 percent mortality rate seems high, it is important to bear in mind that many of our patients were seriously ill. All but one of the patients who died had severe neurological impairment (Ranawat class III-A or III-B). Reoperation rates of 3 percent35 and 11 percent (the present series) have been reported, caused by progression of the disease, failure of the first procedure, infection, or a combination of these factors. In terms of neurological recovery, it appears that between 38 and 72 percent of patients improve. Thus, despite a wide spectrum of operative approaches, there are similar rates of improvement, reoperation, and mortality. This observation certainly does not suggest that there is one correct procedure and might as easily point to the limits of tolerance of these severely affected patients for this operation.
Unlike Menezes et al.29 and Clark et al.6, some authors have reported nonunion in a few patients. Apostolides et al.1 described two patients who, despite bone-grafting and contoured metal loop fixation as well as postoperative immobilization in a halo vest, had no evidence of union. These "stable fibrous unions" did not progress to instability at thirty-four and forty-two months after the procedure. We contend that the majority of our patients may have had at least a stable fibrous union without incurring donor-site morbidity and without the use of a halo vest. Perhaps the restraint of the fixation device allows union to occur. Whatever the mechanism, the data suggest that the functional outcome and the survival rate are the same whether or not bone graft had been inserted.
Techniques of internal fixation have evolved with the use of onlay bone graft, wire, cement, and a contoured implant. The limiting factor is bone quality, and, while plate-and-screw techniques are possible, in our experience the bone quality in many of these patients is so poor that our aim has been to spread the load over as many segments as possible. After the cephalad two joints have been fixed, there is little further loss of mobility when the load is spread over additional segments until the cervicothoracic segment is immobilized26. It was also apparent from this study that very little subaxial instability developed following occipitocervical stabilization without bone-grafting. It is possible that the small amount of motion present within the construct acts to prevent the transmission of large forces to the subaxial spine that might act to accelerate arthritis and instability.
Indications for occipitocervical stabilization include vertical atlantoaxial subluxation or fixed anterior subluxation, or both. Some forms of multidirectional instability may also require such stabilization. Reducible anterior subluxation should be managed with atlantoaxial arthrodesis alone. Odontoidectomy itself (in the absence of vertical atlantoaxial subluxation) is not an absolute indication for occipitocervical stabilization if the anterior part of the ring of the atlas remains intact. If the anterior part of the ring of the atlas is completely resected to facilitate access during anterior craniocervical decompression, posterior stabilization to the occiput is recommended. In this series, any patient who presented with concomitant subaxial instability was treated with inclusion of the affected levels at the index operative procedure, thereby reducing the risk of progression of the instability at the affected levels.
Very few wires or implants broke during the course of this study. There were no late catastrophic failures associated with the lack of osseous fusion, and the few broken wires and the one late loop fracture were clinically inconsequential. Although we observed only one loop fracture over eleven years of routine postoperative radiographic observation, long-term follow-up is recommended for patients who do not obtain an osseous fusion, as very late failure may occur. A possible reason for the absence of late failure in the rheumatoid patients in our study is that the usual pathological development subaxially is ankylosis, which is also the hallmark of juvenile rheumatoid arthritis; there is instability only at the unrestrained atlantoaxial joint, which lacks a disc. Perhaps the restraint of the fixation device permits a similar fibrous or osseous ankylosis to occur.
In many instances, the anterior compression was caused by odontoid translocation, and, unless continuous translocation was prevented, additional problems arose. It soon became apparent that the uncontoured Hartshill and Luque loops were associated with problems15. Maintaining permanent occiput-to-axis distraction by affixing the flare of the Ransford loop to the second cervical lamina was the method chosen to prevent these problems.
Our own experience before we used transoral decompression convinced us of the necessity of decompression prior to fixation. Similarly, the lack of additional morbidity and mortality more than justifies the use of the procedure to reduce the chances of neuraxial damage with posterior fixation. Transoral decompression is indicated in patients with marked anterior compression of the neuraxis due to irreducible atlantoaxial subluxation or in patients who have persistent compression from rheumatoid periodontal pannus. Although rheumatoid pannus may slowly resolve after atlantoaxial stability is achieved, its presence reduces the space available for the cord and increases the risk of sublaminar wire passage at the atlas. The deformity that occurs in anterior subluxation of the odontoid process is one of kyphosis. Odontoidectomy directly removes the compressing structure and permits the thecal sac and the spinal cord to displace anteriorly or to expand into the space created. Resection of the posterior arch of the first cervical lamina in patients with irreducible atlantoaxial subluxation has no effect with regard to reducing anterior compression of the cord. This is especially true when there is associated vertical subluxation of the odontoid process. Resection of the arch of the first cervical lamina also eliminates a potentially valuable fixation level when sublaminar wires are used.
The value of an operation can be questioned for some patients with rheumatoid disease. For bed-bound patients (Ranawat class IIIB), the mortality rate is high: in one study seven (13 percent) of fifty-five patients died within one month after the operation, and in another thirty-two (58 percent) of fifty-five died within three years4,5. Very few of the patients in these studies left institutional care in the postoperative period. The operation does have important long-term benefits for patients who are less severely affected. Nevertheless, even the patients with severe disease in our study experienced notable reductions in head and neck pain and felt that the procedure was worthwhile. We believe that our results indicate that posterior occipitocervical fixation without bone-grafting in patients with advanced rheumatoid arthritis provides long-term benefits.
Apostolides, P. J.; Dickman, C. A.; Golfinos, J. G.; Papadopoulos, S. M.; and Sonntag, V. K.: Threaded Steinmann pin fusion of the craniovertebral junction. Spine,21: 1630-1637, 1996.211630
1996
[PubMed]
Boden, S. D.; Dodge, L. D.; Bohlman, H. H.; and Rechtine, G. R.: Rheumatoid arthritis of the cervical spine. A long-term analysis with predictors of paralysis and recovery. J. Bone and Joint Surg.,75-A: 1282-1297, Sept. 1993.75-A1282
1993
Brattström, H.: Surgery in atlanto-axial dislocation in rheumatoid arthritis. Reconstr. Surg. and Traumatol.,18: 16-29, 1981.1816
1981
Casey, A. T.; Crockard, H. A.; Bland, J. M.; Stevens, J.; Moskovich, R.; and Ransford, A.: Predictors of outcome in the quadriparetic nonambulatory myelopathic patient with rheumatoid arthritis: a prospective study of 55 surgically treated Ranawat class IIIb patients. J. Neurosurg.,85: 574-581, 1996.85574
1996
[PubMed]
Casey, A. T.; Crockard, H. A.; Bland, J. M.; Stevens, J.; Moskovich, R.; and Ransford, A. O.: Surgery on the rheumatoid cervical spine for the non-ambulant myelopathic patient—too much, too late?. Lancet,347: 1004-1007, 1996.3471004
1996
[PubMed]
Clark, C. R.; Keggi, K. J.; and Panjabi, M. M.: Methylmethacrylate stabilization of the cervical spine. J. Bone and Joint Surg.,66-A: 40-46, Jan. 1984.66-A40
1984
Clark, C. R.; Goetz, D. D.; and Menezes, A. H.: Arthrodesis of the cervical spine in rheumatoid arthritis. J. Bone and Joint Surg.,71-A: 381-392, March 1989.71-A381
1989
Crockard, H. A.: The transoral approach to the base of the brain and upper cervical cord. Ann. Roy. Coll. Surgeons,67: 321-325, 1985.67321
1985
Crockard, H. A.; Essigman, W. K.; Stevens, J. M.; Pozo, J. L.; Ransford, A. O.; and Kendall, B. E.: Surgical treatment of cervical cord compression in rheumatoid arthritis. Ann. Rheumat. Dis.,44: 809-816, 1985.44809
1985
[PubMed]
Crockard, H. A.; Pozo, J. L.; Ransford, A. O.; Stevens, J. M.; Kendall, B. E.; and Essigman, W. K.: Transoral decompression and posterior fusion for rheumatoid atlantoaxial subluxation. J. Bone and Joint Surg.,68-B(3): 350-356, 1986.68-B(3)350
1986
Crockard, H. A.: Anterior approaches to lesions of the upper cervical spine. Clin. Neurosurg.,34: 389-416, 1988.34389
1988
[PubMed]
Crockard, H. A.; Calder, I.; and Ransford, A. O.: One-stage transoral decompression and posterior fixation in rheumatoid atlanto-axial subluxation. J. Bone and Joint Surg.,72-B(4): 682-685, 1990.72-B(4)682
1990
Crockard, H. A.: Surgical management of cervical rheumatoid problems. Spine,20: 2584-2590, 1995.202584
1995
[PubMed]
Ferlic, D. C.; Clayton, M. L.; Leidholt, J. D.; and Gamble, W. E.: Surgical treatment of the symptomatic unstable cervical spine in rheumatoid arthritis. J. Bone and Joint Surg.,57-A: 349-354, April 1975.57-A349
1975
Flint, G. A.; Hockley, A. D.; McMillan, J. J.; and Thompson, A. G.: A new method of occipitocervical fusion using internal fixation. Neurosurgery,21: 947-950, 1987.21947
1987
[PubMed]
Frankel, H. L.; Hancock, D. O.; Hyslop, G.; Melzak, J.; Michaelis, L. S.; Ungar, G. H.; Vernon, J. D.; and Walsh, J. J.: The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia. I. Paraplegia,7: 179-192, 1969.7179
1969
[PubMed]
Helewa, A.; Goldsmith, C. H.; and Smythe, H. A.: Independent measurement of functional capacity in rheumatoid arthritis. J. Rheumatol.,9: 794-797, 1982.9794
1982
[PubMed]
Krieg, J. C.; Clark, C. R.; and Goetz, D. D.: Cervical spine arthrodesis in rheumatoid arthritis: a long-term follow-up. Yale J. Biol. and Med.,66: 257-262, 1993.66257
1993
Lee, P.; Jasani, M. K.; Dick, W. C.; and Buchanan, W. W.: Evaluation of a functional index in rheumatoid arthritis. Scandinavian J. Rheumatol.,2: 71-77, 1973.271
1973
Luque, E. R.: The anatomic basis and development of segmental spinal instrumentation. Spine,7: 256-259, 1982.7256
1982
[PubMed]
Luque, E. R.: Segmental spinal instrumentation for correction of scoliosis. Clin. Orthop.,163: 192-198, 1982.163192
1982
[PubMed]
Luque, E. R.: Paralytic scoliosis in growing children. Clin. Orthop.,163: 202-209, 1982.163202
1982
[PubMed]
Luque, E. R.: Segmental spinal instrumentation of the lumbar spine. Clin. Orthop.,203: 126-134, 1986.203126
1986
[PubMed]
McAfee, P. C.; Lubicky, J. P.; and Werner, F. W.: The use of segmental spinal instrumentation to preserve longitudinal spinal growth. J. Bone and Joint Surg.,65-A: 935-942, Sept. 1983.65-A935
1983
McGregor, M.: The significance of certain measurements of the skull in the diagnosis of basilar impression. British J. Radiol.,21: 171-181, 1948.21171
1948
Malcolm, G. P.; Ransford, A. O.; and Crockard, H. A.: Treatment of non-rheumatoid occipitocervical instability. Internal fixation with the Hartshill-Ransford loop. J. Bone and Joint Surg.,76-B(3): 357-366, 1994.76-B(3)357
1994
Mathews, J. A.: Atlanto-axial subluxation in rheumatoid arthritis. A 5-year follow-up study. Ann. Rheumat. Dis.,33: 526-531, 1974.33526
1974
[PubMed]
Meijers, K. A. E.; van Beusekom, G. T.; Luyendijk, W.; and Duijfjes, F.: Dislocation of the cervical spine with cord compression in rheumatoid arthritis. J. Bone and Joint Surg.,56-B(4): 668-680, 1974.56-B(4)668
1974
Menezes, A. H.; VanGilder, J. C.; Clark, C. R.; and el-Khoury, G.: Odontoid upward migration in rheumatoid arthritis. An analysis of 45 patients with "cranial settling". J. Neurosurg.,63: 500-509, 1985.63500
1985
[PubMed]
Menezes, A. H., and Ryken, T. C.: Instrumentation of the craniocervical region. In Spinal Instrumentation, pp. 47-62. Edited by E. C. Benzel. Park Ridge, Illinois, American Association of Neurological Surgeons, 1994.
Menezes, A. H.: Posterior occipitocervical fixation. Tech. Neurosurg.,1: 72-81, 1995.172
1995
Newman, P., and Sweetnam, R.: Occipito-cervical fusion. An operative technique and its indications. J. Bone and Joint Surg.,51-B(3): 423-431, 1969.51-B(3)423
1969
Parker, J. W.; Harrell, P. B.; and Alarcón, G. S.: The value of the joint alignment and motion scale in rheumatoid arthritis. J. Rheumatol.,15: 1212-1215, 1988.151212
1988
[PubMed]
Pellicci, P. M.; Ranawat, C. S.; Tsairis, P.; and Bryan, W. J.: A prospective study of the progression of rheumatoid arthritis of the cervical spine. J. Bone and Joint Surg.,63-A: 342-350, March 1981.63-A342
1981
Peppelman, W. C.; Kraus, D. R.; Donaldson, W. F., III; and Agarwal, A.: Cervical spine surgery in rheumatoid arthritis: improvement of neurologic deficit after cervical spine fusion. Spine,18: 2375-2379, 1993.182375
1993
[PubMed]
Ranawat, C. S.; O'Leary, P.; Pellicci, P.; Tsairis, P.; Marchisello, P.; and Dorr, L.: Cervical spine fusion in rheumatoid arthritis. J. Bone and Joint Surg.,61-A: 1003-1010, Oct. 1979.61-A1003
1979
Ransford, A. O.; Crockard, H. A.; Pozo, J. L.; Thomas, N. P.; and Nelson, I. W.: Craniocervical instability treated by contoured loop fixation. J. Bone and Joint Surg.,68-B(2): 173-177, 1986.68-B(2)173
1986
Redlund-Johnell, I., and Pettersson, H.: Radiographic measurements of the cranio-vertebral region. Designed for evaluation of abnormalities in rheumatoid arthritis. Acta Radiol.,25: 23-28, 1984.2523
1984
Redlund-Johnell, I., and Pettersson, H.: Vertical dislocation of the C1 and C2 vertebrae in rheumatoid arthritis. Acta Radiol.,25: 133-141, 1984.25133
1984
Redlund-Johnell, I.: Atlanto-occipital dislocation in rheumatoid arthritis. Acta Radiol.,25: 165-167, 1984.25165
1984
Resina, J., and Alves, A. F.: A technique of correction and internal fixation for scoliosis. J. Bone and Joint Surg.,59-B(2): 159-165, 1977.59-B(2)159
1977
Ritchie, D. M.; Boyle, J. A.; McInnes, J. M.; Jasani, M. K.; Dalakos, T. G.; Grieveson, P.; and Buchanan, W. W.: Clinical studies with an articular index for the assessment of joint tenderness in patients with rheumatoid arthritis. Quart. J. Med.,37: 393-406, 1968.37393
1968
[PubMed]
Sharp, J., and Purser, D. W.: Spontaneous atlanto-axial dislocation in ankylosing spondylitis and rheumatoid arthritis. Ann. Rheumat. Dis.,20: 47-77, 1961.2047
1961
[PubMed]
Smith, P. H.; Benn, R. T.; and Sharp, J.: Natural history of rheumatoid cervical luxations. Ann. Rheumat. Dis.,31: 431-439, 1972.31431
1972
[PubMed]
Steinbrocker, O.; Traeger, C. H.; and Batterman, R. C.: Therapeutic criteria in rheumatoid arthritis. J. Am. Med. Assn.,140: 659-662, 1949.140659
1949
Sunahara, N.; Matsunaga, S.; Mori, T.; Ijiri, K.; and Sakou, T.: Clinical course of conservatively managed rheumatoid arthritis patients with myelopathy. Spine,22: 2603-2607, 1997.222603
1997
[PubMed]
Wackenheim, A.: Roentgen Diagnosis of the Craniovertebral Region. Berlin, Springer, 1974.
Winfield, J.; Young, A.; Williams, P.; and Corbett, M.: Prospective study of the radiographic changes in hands, feet, and cervical spine in adult rheumatoid disease. Ann. Rheumat. Dis.,42: 613-618, 1983.42613
1983
[PubMed]
Zoma, A.; Sturrock, R. D.; Fisher, W. D.; Freeman, P. A.; and Hamblen, D. L.: Surgical stabilisation of the rheumatoid cervical spine. A review of indications and results. J. Bone and Joint Surg.,69-B(1): 8-12, 1987.69-B(1)8
1987