 |
 |
 |
|
| home | CME | help | feedback | subscriptions | archive |
|
© 1997 The Journal of Bone and Joint Surgery, Inc. Pyogenic Vertebral Osteomyelitis* , STANFORD, CALIFORNIAInvestigation performed at Stanford University Medical Center, Stanford, Palo Alto Veterans Administration Medical Center, Palo Alto, and Santa Clara Valley Medical Center, San Jose
I retrospectively reviewed the records of 111 patients who had pyogenic vertebral osteomyelitis unrelated to an open procedure on the spine. The mean age at the time of the diagnosis was sixty years (range, eighteen to eighty-four years); sixty-one patients (55 per cent) were sixty years old or more. Forty-four patients (40 per cent) had an impaired immune system secondary to diabetes mellitus, the use of corticosteroids, chemotherapy for cancer, rheumatic or immunological disease, renal or hepatic failure, malnutrition, or myelodysplasia. Magnetic resonance imaging, critical for the determination of an early diagnosis, was performed for 103 patients (93 per cent). The infection in sixty-eight patients (61 per cent) was diagnosed within one month after the onset of symptoms. The most frequent infecting organism was Staphylococcus aureus (forty patients; 36 per cent). The infection in forty-one patients (37 per cent) was caused by organisms, such as Staphylococcus epidermidis, Propionibacterium acnes, and diphtheroid species, that are traditionally considered to be of low virulence. The urinary tract was the most frequent source of infection (confirmed in thirteen patients and suspected in twenty-one). The success of non-operative treatment was predicted by four independent variables: an age of less than sixty years, the immune status, infection with Staphylococcus aureus, and a decreasing erythrocyte sedimentation rate. Forty-two patients were managed with débridement and arthrodesis. Fourteen of these patients also had instrumentation of the spine, in the presence of infection, without compromise of the outcome. Eighteen patients died by the time of the latest follow-up evaluation at a mean of four years (range, two years and two months to six years and six months): seven who had been managed non-operatively died in the first month after the diagnosis was made, three died in the acute postoperative period, three died of late complications of paraplegia, and five died of unrelated causes. None of the eighty-nine patients who were seen at a minimum of two years postoperatively had had late recurrence of infection. Chronic, severe back pain was noted in only seven patients.
Pyogenic vertebral osteomyelitis is uncommon31. Before the availability of antibiotics, the course of the disease in adults was characterized by acute, subacute, and chronic syndromes; a mortality rate of 25 per cent (twelve of forty-eight patients)16; and an association with other intractable suppurative conditions16. The course of the disease changed with the introduction of effective antibiotic treatment3,6,15. The development of effective operative approaches to the anterior aspect of the spine for the treatment of tuberculosis17 helped to alter the course further7,8. The demographics of the disease also appear to have changed. The patients in the present study appeared to be older and in poorer health than those previously reported on7,24. This may be related to the greater longevity of the general population as well as to the more intensive treatment of serious illnesses and the use of chemotherapy to treat cancer and immunological disorders. Furthermore, the infecting organisms in my patients were generally not the usual organisms that have been isolated in the past (Staphylococcus aureus, group-B streptococci, and Enterobacteriaceae). Also, the diagnosis is made more quickly now than in the past. I performed a retrospective review of the clinical course and outcome of pyogenic vertebral osteomyelitis in all patients seen with that diagnosis during a five-year period at three teaching hospitals.
I reviewed the records of all patients who were more than sixteen years old and had been managed for pyogenic vertebral osteomyelitis at the Stanford University Medical Center, Veterans Administration Medical Center in Palo Alto, or Santa Clara Valley Medical Center of Santa Clara County between 1988 and 1993. The patients were identified with use of operating-room records and admission and discharge computer diagnostic codes. These were double-checked when possible against bacteriology records containing the results of cultures of deep tissue from the spine. The criteria for inclusion were spinal infection proved by a positive culture or by evidence of osseous destruction on the imaging study and histopathological evidence of osteomyelitis as well as a minimum of two years between the onset of symptoms and the latest follow-up evaluation. Patients were excluded if the infection appeared to be the result of an open procedure on the spine. Initially, 118 patients met the inclusion criteria; however, adequate records and radiographic data were obtained for only 111 (94 per cent). The medical records were reviewed for demographic data such as age and gender, associated medical conditions, immune status, clinical symptoms, findings on physical examination, neurological status, results of laboratory tests on blood and urine and of cultures of various fluids and tissues, findings on radiographic and imaging studies, clinical course, treatment, and findings on follow-up evaluations. Data were collected with use of a standardized record and were entered into a computer database for analysis. Eighteen of the 111 patients died by the time of the latest follow-up evaluation. The cause of death was ascertained from the treating physician or the postmortem examination. Four other patients could not be contacted, and it was not known if they were alive or dead. The remaining eighty-nine patients (96 per cent of the potential survivors) were alive and were contacted for follow-up assessment; 4 per cent were lost to follow-up. The follow-up evaluation was performed at a minimum of two years by one of three independent examiners, not involved in the care of the patient, who conducted a structured interview. Forty-six patients were evaluated by the examiners during a routine visit to their treating physician; twenty-seven patients came specifically to be interviewed; and sixteen were interviewed by telephone. Only the findings from the structured interview segment of the follow-up evaluation are included in this study, as not all of the patients had a physical examination. The follow-up data were entered into the computer database. The paired and unpaired t tests and chi-square analyses were used to compare continuous and categorical parameters, respectively. StatView software (Abacus Concepts, Mountain View, California) was used for statistical calculations.
Data on the Patients A total of 145 segments were involved, including two at the level of the first and second thoracic vertebrae, one at the level of the second and third thoracic vertebrae, three at the level of the third and fourth thoracic vertebrae, three at the level of the fourth and fifth thoracic vertebrae, six at the level of the fifth and sixth thoracic vertebrae, five at the level of the sixth and seventh thoracic vertebrae, seven at the level of the seventh and eighth thoracic vertebrae, ten at the level of the eighth and ninth thoracic vertebrae, nine at the level of the ninth and tenth thoracic vertebrae, ten at the level of the tenth and eleventh thoracic vertebrae, ten at the level of the eleventh and twelfth thoracic vertebrae, twelve at the level of the twelfth thoracic and first lumbar vertebrae, eleven at the level of the first and second lumbar vertebrae, sixteen at the level of the second and third lumbar vertebrae, fifteen at the level of the third and fourth lumbar vertebrae, seventeen at the level of the fourth and fifth lumbar vertebrae, and eight at the level of the fifth lumbar and first sacral vertebrae. The lumbosacral and thoracolumbar segments were predominantly involved. One hundred patients had involvement of only one vertebra or one motion segment (two adjacent vertebrae and the intervening disc), and eleven had involvement of several segments. Concomitant pathological (non-degenerative) conditions were found at or near the site of the infection in eighteen patients: two had a lymphoma, three had an active prostatic carcinoma, eleven had had previous irradiation for a malignant tumor of the spine but did not have evidence of an active neoplasm, one had a compression fracture (osteopenic), and one had a traumatic burst fracture. Forty-four (40 per cent) of the 111 patients had an impaired immune system, implying the presence of an illness that, by its nature or because of the medications used for its treatment, resulted in an impairment of the cellular or humoral immune response of the body; twenty-three of these patients had more than one comorbid condition. The comorbid conditions included diabetes mellitus in twenty-eight patients, therapy with corticosteroids in eighteen, a malignant tumor that was being treated with chemotherapy in eleven, rheumatic or immunological disease in nine, renal failure in three, hepatic failure in three, malnutrition in one, and myelodysplasia in one.
The main symptom of the infection was back pain in 101 patients. Eight of the ten patients who did not complain of back pain had organic dementia secondary to sepsis, and the remaining two patients were comatose. Sixty-seven patients had only back pain at the initial evaluation, thirty-seven had muscle weakness, eighteen were febrile, three had signs of sepsis, and two were paraplegic. The white blood-cell count was elevated in forty-four (41 per cent) of the 108 patients for whom it was measured, and the Westergren erythrocyte sedimentation rate was elevated in ninety-eight (95 per cent) of the 103 patients for whom it was measured. The white blood-cell count was normal in thirty-one (70 per cent) of the forty-four patients who had an impaired immune system, and the sedimentation rate was normal in five (11 per cent). Of the fifty-nine patients who had a normal immune status and for whom the results of both tests were available, thirty-three (56 per cent) had a normal white blood-cell count and none had a normal sedimentation rate. The white blood-cell count was normal in forty-six (75 per cent) of the sixty-one patients who were at least sixty years old, and the sedimentation rate was normal in three (5 per cent). The diagnostic studies performed for the 111 patients included 103 magnetic resonance imaging studies, thirty-one computerized tomography scans, twenty-three myelographic computerized tomography scans, twenty-three technetium radionucleotide scans, four gallium scans, and four white blood-cell scans. For eighteen patients (16 per cent), the clinical findings and magnetic resonance images were considered sufficient, in the opinion of the treating physicians, to establish the diagnosis of a spinal infection. The diagnosis was later corroborated with either a positive culture or histopathological changes consistent with non-granulomatous pyogenic osteomyelitis. For thirty-seven patients (33 per cent), the magnetic resonance images and a positive blood culture were sufficient to establish the diagnosis without a biopsy. Open biopsy (performed posteriorly through a transpedicular approach or anteriorly at the time of the corpectomy) was needed to establish the diagnosis in thirteen patients (12 per cent); fine-needle aspiration and computerized tomography, in twenty-seven (24 per cent); and core biopsy, in sixteen (14 per cent). The blood culture was positive in sixty-six patients (59 per cent) at some time during the course of the disease. Three true-positive blood cultures and three true-positive specimens from needle aspirations were initially interpreted as false-positive or contaminated specimens; these were later verified to be positive for the same organisms at the time of open or repeated biopsy. Thirty-nine of ninety-two urine cultures were positive; however, eleven were apparently contaminated (less than 10,000 colonies per milliliter). The organism isolated from thirteen of the remaining twenty-eight positive urine specimens was the same as that isolated from the blood or a specimen from the spine.
Bacteriological Findings The infecting organism was virulent (capable of causing infection by overcoming the natural defenses of the body) in sixty-six patients and was of low virulence in forty-one. Staphylococcus aureus was isolated from specimens from forty patients (36 per cent); Staphylococcus epidermidis, eighteen (16 per cent); group-B beta-hemolytic streptococcus, eleven (10 per cent); and Escherichia coli and Streptococcus viridans, seven patients (6 per cent) each. Propionibacteria, fusiform bacteria, and diphtheroid species were found in four patients (4 per cent), two patients (2 per cent), and one patient (1 per cent), respectively. The most common pathogen in the fifteen patients who had a history of intravenous drug abuse was Staphylococcus aureus (eleven patients); no organism was found in the specimens from the remaining four patients. Pseudomonas aeruginosa was isolated from the specimens from one of the six other patients who had either a remote or suspected history of intravenous drug abuse. Low-virulence organisms caused twenty-nine (48 per cent) of the infections in the sixty-one elderly patients and twenty-four (55 per cent) of the infections in the forty-four patients who had an impaired immune response; these findings indicate a predilection for unusual infection in these groups.
Source
Treatment Thirty-nine of the 111 patients had less than one month of non-operative treatment. Eight of these patients died within one month after the onset of symptoms and seven, without operative intervention. One died after operative intervention because of persistent sepsis. Twenty-six patients had an anterior or posterior decompression within one month after the onset of symptoms. In eleven patients, corpectomy or drainage was performed to obtain tissue for histopathological evaluation. Forty-two patients had operative intervention. An anterior procedure alone was done in seventeen patients, a posterior procedure alone in one, a combined anterior and posterior procedure on the same day in twenty, and a staged anterior and posterior procedure in four. Posterior instrumentation was used in fourteen patients to allow early mobilization after anterior corpectomy or because the posterior decompression had destabilized the spine. There were no persistent infections in these fourteen patients.
Outcome
Survival There was no evidence of active vertebral infection in the eight patients who died more than six months after the diagnosis was made. Three of these patients had paraplegia: one committed suicide, one had renal disease, and one had a systemic infection due to a neglected sacral decubitus. The deaths of the remaining five patients, all of whom were elderly, could not be attributed to a recurrence of the infection. It is likely that the effects of a prolonged serious illness contributed to their death. There was an association between mortality and the age of the patient, the immune status, and infection with Staphylococcus aureus. Nine (21 per cent) of the forty-two patients who were at least seventy years old and nine (13 per cent) of the sixty-nine patients who were less than seventy years old had died by the time of the latest follow-up examination. The mortality rate appeared to be higher when the organism was Staphylococcus aureus: of the eight patients who died within the first month, six (including all three who died in the acute postoperative period) had had an infection with Staphylococcus aureus (p = 0.05). Six (14 per cent) of the forty-four patients who had an impaired immune response died within the first six months after the diagnosis, whereas only four (6 per cent) of the sixty-seven patients who had a normal immune status died in this period (p = 0.25).
Infection and Pain
Paralysis Twenty (45 per cent) of the forty-four patients who had an impaired immune system had some neurological deficit, whereas thirteen (19 per cent) of the remaining sixty-seven patients had a neurological deficit. This difference was significant (p = 0.01). Similarly, paralysis of the bladder was found in nine (20 per cent) of the forty-four patients who had an impaired immune system and in only six (9 per cent) of the sixty-seven patients who had a normal immune system. This difference was also significant (p = 0.05). The infecting organism was Staphylococcus aureus in 36 per cent (forty) of the 111 patients and in 42 per cent (fourteen) of the thirty-three patients who had a neurological deficit. This difference was not significant, with the numbers available.
Early reports of pyogenic vertebral osteomyelitis described the difficulty involved in the diagnosis, an association with diabetes mellitus, and the serious neurological sequelae1,3,6,13,15,20,21,27,31. Despite the advent of antibiotics, the initial results of laminectomy were not encouraging15,19. Subsequently, anterior decompression and reconstruction of the spine proved to be much more effective, and laminectomy was contraindicated except in special circumstances4,5,8,33. At the same time, computerized tomography and radionucleotide imaging improved the diagnostic capability7,9,28. However, despite these changes, few reports of the results of treatment of pyogenic vertebral osteomyelitis were published. The patients who had pyogenic vertebral osteomyelitis in this series appeared to be older than those in other reported series, and they were more likely to have a comorbid condition than previously reported13-15,18. This is possibly because diagnostic and therapeutic advances in the treatment of cancer and diseases of the immune system have led to greater longevity. Advances in therapeutics have altered the importance of organisms that had been discounted as contaminants, and newer imaging techniques have made it possible to diagnose vertebral osteomyelitis early without the need for invasive diagnostic tests. In the past, staphylococci and streptococci had been the most commonly isolated organisms, to the virtual exclusion of other non-granulomatous agents1,16,27. Staphylococcus aureus is still the most common organism; however, gram-negative and low-virulence, atypical organisms are now frequently isolated11,14,20-22,24,31. The primary cause of pyogenic vertebral osteomyelitis in intravenous drug users has been reported to be Pseudomonas24,32. In the present series, only one of twenty-one patients who had a confirmed or suspected history of intravenous drug abuse was infected with Pseudomonas. Wiesseman et al.32 reported an association between pyogenic vertebral osteomyelitis due to infection with Pseudomonas and heroin addiction, and Patzakis et al.24 found that fifteen of twenty-three drug addicts had Pseudomonas pyogenic vertebral osteomyelitis. In the present study, the low prevalence of infection may be related to the intensive needle-exchange program in my area, the use of chlorine bleach to sterilize needles, a failure to detect spinal infection when an infection with Pseudomonas at a distant site was being treated, or the fact that some of the drug abusers medicated themselves with antibiotics at the onset of a fever, making it difficult to obtain a positive culture. One of the striking features of this study is the advanced age of the patients. Griffiths and Jones15, in a study of twenty-eight patients, noted a high prevalence of pyogenic infection of the spine in young patients and observed no such infections in patients who were more than seventy years old. Eismont et al.7 reported that only two of sixty-one patients who had pyogenic vertebral osteomyelitis were more than seventy years old. Emery et al.8 and Frederickson et al.11, in studies of twenty-one and seventeen patients, each had three patients who were more than seventy years old. Jones et al.18 noted, in a study conducted in a general hospital setting, that pyogenic vertebral osteomyelitis was becoming more widespread, particularly in older patients. Sixty-one (55 per cent) of the patients in the present series were at least sixty years old. Along with the increased age, there was an increase in the prevalence of concomitant illness, particularly in association with diseases that suppressed the immune system (forty-four of 111 patients). The usual delay in diagnosis has been reported to be two to four months3,6,18,27,33. Kemp et al.21 noted a mean time to diagnosis of more than six months. In the present study, twenty-seven (24 per cent) of the 111 infections were diagnosed within one week after the onset of symptoms and eighty-nine (80 per cent), within two months. It is likely that the advent of magnetic resonance imaging has made it possible to diagnosis this disease in the early stages9,23,25,29. The most common source of infection in the present study, as in other studies1,3,6,7,11,31, was the genito-urinary system. This association is troubling because of the potential increase in vertebral osteomyelitis secondary to seeding of the spine with organisms as a result of urethroscopy performed on a regular basis for the early detection of prostatic cancer. The high prevalence of suspected or diagnosed prostatic carcinoma (twenty [30 per cent] of the sixty-seven men) in this study has not been reported previously, to my knowledge. I also found that eleven (10 per cent) of the 111 patients had had radiation therapy for cancer with spinal metastases; this association has rarely been reported2,12,26,30. The high prevalence of neurological involvement (30 per cent; thirty-three patients) in this series is similar to previously reported rates7,11,15,20,21,26 of 29 per cent (eight of twenty-eight15) to 51 per cent (thirty-one of sixty-one7). The recovery of neurological function after treatment has also been well documented7,11,15,20,21,26. In the present study, infection with Staphylococcus aureus impaired immune status, and thoracic lesions were associated with greater neurological loss. Remarkably few patients had serious residual back pain after more than two years of follow-up. Several factors have been mentioned as a cause of the failure of treatment of pyogenic vertebral osteomyelitis. Eismont et al.7 noted that rheumatic conditions and diabetes mellitus increased the risk of complications secondary to pyogenic vertebral osteomyelitis. Cahill et al.4 reported a favorable response to operative treatment in elderly patients who did not have other risk factors, but they did not discuss non-operative treatment. Gepstein et al.14 commented on the association between paralysis and pyogenic vertebral osteomyelitis in elderly patients. The frequent failure of non-operative treatment despite prolonged use of antibiotics in patients who have pyogenic vertebral osteomyelitis and an impaired immune system has been reported previously5. In the present series, age, an impaired immune system, and a response to therapy reflected by a decrease in the erythrocyte sedimentation rate appeared to be associated with the outcome of non-operative treatment. The rate of mortality (16 per cent; eighteen of 111) in this series is higher than the 5 per cent rate (two of forty) reported by Garcia and Grantham13 (both of their patients who died were elderly). Cahill et al.4 noted that the ten elderly patients who were managed operatively in their study were still alive six months to three years postoperatively. Eismont et al.7 reported that five of twenty-two patients who had paralysis and one of thirty patients who did not have a neurological deficit died; two of the sixty-one patients in the study were thought to have died secondary to infection. The cause of death of six patients in the present study who died soon after the diagnosis was made was apparently not related to the osteomyelitis. Without magnetic resonance imaging, pyogenic vertebral osteomyelitis would not have been diagnosed in these patients. A retrospective report of this nature has inherent limitations. The patients were identified with use of hospital computer records employed for admission, discharge, and billing. Although these were double-checked against operating-room schedules and bacteriological records, omissions may have occurred. There may have been a patient-selection bias in the record-coding, and some diagnoses may not have been correct. It is possible that the number of low-virulence organisms was underestimated and such organisms may have been discounted as contaminants. My interest in this disease probably led to referral to these centers, and it would not be valid to project the prevalence of pyogenic vertebral osteomyelitis from these data to the community as a whole. The method of treatment varied among physicians. Operative indications were not consistent, and the administration of antibiotics varied, with regard to both the type and the duration, according to the practitioner. Finally, neurological status was often poorly documented until the neurosurgeon or the orthopaedic consultant evaluated the patient. Despite these limitations, the findings of an older population, an increased prevalence of immunodeficiency, diversification of infecting agents, and rapidity of diagnosis probably represent a bona fide evolution of pyogenic vertebral osteomyelitis in modern clinical practice. A review of these results and of published data suggests that the clinical features of pyogenic vertebral osteomyelitis are continually evolving. Pyogenic vertebral osteomyelitis is being diagnosed more frequently in older patients who have serious medical conditions. Magnetic resonance imaging has proved to be an effective early diagnostic tool. Staphylococcus aureus remains the most commonly isolated organism, but uncommon pathogens or low-virulence organisms were seen in about 40 per cent of the patients. Non-operative treatment was more successful in patients who were less than sixty years old, had an intact immune system, and had a decreasing erythrocyte sedimentation rate during the first month. The outcome at two years was good: there was general neurological improvement, most of the patients had clinical resolution of the infection, and few patients had residual back pain.
NOTE: The author thanks Dr. Stuart Goodman for his support of this research with funding, time, and space and Dr. Donald Nagel, Dr. Lawrence Rinsky, Dr. Glen O'Sullivan, Dr. James Billys, and Dr. Charles Sonu for their contributions during the preparation of this paper. The author also acknowledges Elizabeth Helms, Theresa van der Vugt, and Daniel Vittum for their assistance in reviewing the medical records and radiographs and in performing the follow-up of the patients.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
  CiteULike  Connotea  Del.icio.us  Facebook  Technorati  Twitter What's this?
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
| home | CME | help | feedback | subscriptions | archive | site index |
|
Stanford University Libraries' HighWire Press (TM) assists in the publication of JBJS Online. |