pontaneous Pyogenic Facet Joint Infectionavier Narváez, MD, PhD,* Joan M. Nolla, MD, PhD,* José A. Narváez, MD,†
aura Martinez-Carnicero, MD,† Eugenia De Lama, MD,†
armen Gómez-Vaquero, MD, PhD,* Oscar Murillo, MD,‡ José Valverde, MD, PhD,*nd Javier Ariza, MD, PhD‡
OBJECTIVE To analyze the clinical features, approaches to management, and out-come of spontaneous pyogenic facet joint infection (PFJI) in adults.PATIENTS AND METHODS Case series of 10 adults with microbiologically proven PFJIdiagnosed during a 10-year period in a teaching hospital, plus a review of 32additional cases previously reported (PubMed 1972 to 2003). Patients with priorspinal instrumentation or surgery and injection drug users were excluded. Onlycases that were sufficiently detailed to be individually analyzed were included.These 32 cases, together with our 10 patients, form the basis of the presentanalysis.RESULTS PFJI represented nearly 20% of all spontaneous pyogenic spinal infectiondiagnosed in our hospital during a 10-year period. This data suggest that PFJI ismore common than was previously thought. Of the 42 patients with PFJI, 26 (62%)were men and 16 (38%) were women, with ages ranging from 20 to 86 years (meanage, 59 � 15 years); 55% of patients were older than 60 years. The most commonlocation of infections was the lumbosacral region (86%). All patients presentedwith severe back pain; fever was noted in 83% of the cases and neurologicalimpairment in nearly 48%. In 38% of patients a systemic predisposing factor forinfection was present; the most common conditions were diabetes mellitus,malignancies, and alcoholism. In almost 36% of cases, one or more concomitantinfectious processes due to the same microorganism was found, mainly arthritis,skin and soft-tissue infections, endocarditis, and urinary tract infections. Staph-ylococcus aureus was the most common etiologic microorganism (86% of cases).Bacteremia was documented in 81% of the cases. The diagnosis of PFJI wasbased mainly on imaging study findings. Paraspinal and/or epidural extension wasfrequent (81% of cases), but its presence did not indicate a worse prognosis.Medical treatment alone was usually successful. The overall prognosis of PFJIwas good, with a mortality rate of only 2%. The great majority of patients werecured without functional sequelae.CONCLUSION Incidence data from our institution reveal that PFJI is not a rare condi-tion, representing approximately 20% of all pyogenic spinal infections. This entityshould be considered in the differential diagnosis of patients with low back pain,especially in the presence of fever, whatever the patient’s immunological status.
KEYWORDS facet or interapophyseal joint, pyogenic arthritis
Rheumatology Department.Radiology Department.Infectious Diseases Department, Hospital Universitari de Bellvitge–IDIBELL, Barcelona, Spain.ddress reprint requests to: Dr. Francisco Javier Narváez García, Department of Rheumatology (Planta 10-2), Hospital Universitari de Bellvitge, Feixa Llarga s/n,
Hospitalet de Llobregat, Barcelona 08907, Spain. E-mail: [email protected].
raditionally, pyogenic arthritis of the facet (interapophy-seal) joints, also called pyogenic facet joint infection
PFJI), has been considered a rare clinical entity. Before the990s, information about this condition was very limited1-7). In the last few years, a considerable number of reportsoncerning PFJI have been published (8-34). However, sinceost reports consist of single cases (2-5,7-9,11-13,16-23,25-
9,31-34) or small series (1,6,10,14,15,24,30), certain clin-cal, microbiological, diagnostic, and therapeutic aspects ofhe disease remain to be clarified.
In this study, we present all cases of spontaneous PFJIiagnosed at our institution during a 10-year period andeview the available literature to summarize the experienceith this infectious entity.
atients and Methodstudy Sitehe study was performed at the Hospital Universitari de Bellvitge, a000-bed tertiary care teaching institution in Barcelona, Spain.
atientsrom the database of our institution, we searched for patientsith PFJI who were admitted from 1994, the first year in which
ccess to magnetic resonance imaging (MRI) in our center wasnrestricted, to 2003. Eleven patients were identified; 1 was anctive IV drug user and was excluded to homogenize the series.o cases of PFJI related to prior spinal instrumentation or sur-
ery were found. The 10 patients included had the following: (1)compatible clinical picture (back pain often accompanied by
ocalized tenderness or limited range of motion); (2) compatiblemaging findings in MRI study (synovitis/joint effusion and sub-hondral bone marrow edema in the facet joint, usually associ-ted with bone erosions and inflammatory periarticular chang-s); and (3) a microbiologically proven diagnosis. Bacteriologicaliagnosis was established on the basis of positive cultures ofpinal biopsy specimens or if blood cultures were positive. Theseases were evenly distributed throughout the study period.
In addition to these 10 cases of PFJI, 40 patients withpondylodiscitis and 2 patients with vertebral osteomyelitisf the posterior elements were diagnosed during the sameeriod, for a total of 52 cases of microbiologically provenpontaneous pyogenic spinal infection (excluding patientsith prior spinal instrumentation or surgery, injection drugsers, and those with tuberculosis or brucellosis). Informa-ion about the clinical course and microbiological features ofome of these patients with spondylodiscitis has been re-orted elsewhere (35).
ata Collectionhe medical records of the patients with PFJI were tran-cribed onto a specific form that included the following in-ormation: (1) year of diagnosis; (2) demographic data; (3)ertebral level involved; (4) clinical presentation; (5) pres-nce of fever before diagnosis, defined as an axillary temper-ture over 37.5°C; (6) predisposing factors; (7) concomitantnfectious processes; (8) erythrocyte sedimentation rate
ESR); (9) white blood cell (WBC) count; (10) bacteriologic a
iagnosis; (11) diagnostic imaging studies; (12) presence ofxtravertebral extension (epidural or paraspinal); and (13)reatment and outcome.
iterature Reviewrevious cases of PFJI reported from January 1972 to Decem-er 2003 were identified using the PubMed (National Libraryf Medicine, Bethesda, MD) search engine. The key wordssed were facet joint, interapophyseal joint, zygapophysial joint,eptic arthritis, pyogenic arthritis, and osteomyelitis. Only En-lish, French, and Spanish reports were considered; the ref-rences of the obtained studies were then examined to iden-ify additional reports.
We excluded the following: (1) patients aged under 18ears (2 cases) (13,26) or injection drug users (1 case) (19);2) 7 cases related to spinal procedures, such as therapeuticoint infiltration (20,23,32,33) or placement of epiduralatheters (10,28); and (3) 2 cases in whom no organism wassolated (32,34). Additionally, we also excluded a case inhich PFJI was the initial manifestation of spondylodiscitis
7).Finally, we also excluded the cases reported by David-
hausse and coworkers (1) and Muffoleto and coworkers (6atients; 2 of them IV drug abusers) (30) because their clin-
cal characteristics were not sufficiently detailed to be indi-idually analyzed.
tatistical Analysisontinuous data were described as mean � standard de-iation (SD) and categorical variables as percentages.omparisons between groups were made using Student t
est for independent continuous variables or the Mann–hitney U test when the assumption of normality was not
ealized. To analyze categorical data, we performed thehi-square test or Fisher’s exact test when the expectedalues were less than 5. Correlations were made by meansf Spearman’s test. Statistical significance was defined as P
0.05.
esultshe main characteristics and outcome of the 10 patients withpontaneous PFJI diagnosed in our institution are summa-ized in Tables 1 and 2; their mean age was similar to the restf patients with pyogenic spinal infection (60 � 14 yearsersus 64 � 14 years, respectively; P � nonsignificant [NS]).FJI represented nearly 20% (10/52) of all cases of sponta-eous pyogenic spinal infection diagnosed in our hospitaluring a 10-year period.In addition to our 10 cases, 32 well-documented cases
f spontaneous PFJI in adult nondrug users were identi-ed in the literature. In all cases, the infection occurred
ndependently of spondylodiscitis. Details of these pa-ients are summarized in Tables 3 and 4. Thus, includinghe patients presented here, a total of 42 cases were avail-
ble for review.
DOwtw1w
VAag
sTo(
CPaw
s(
T
C
T
C
*
274 J. Narváez et al
emographic Dataf the 42 patients, 26 (62%) were men and 16 (38%) wereomen (male:female ratio of 1.6), with ages ranging from 20
o 86 years (mean, 59 � 15 years). The age at onset of PFJIas similar in men and women (58 � 16 years versus 60 �3 years, respectively; P � NS). Fifty-five percent of patientsere older than 60 years (see Table 5).
ertebral Level Involvedll patients had unilateral facet joint involvement, mainlyffecting the right side (55% of cases). The lumbosacral re-ion was affected in 86% (36/42) of the cases, the cervical
able 1 Clinical Characteristics of Our Patients
Patients YearAge/Sex
(yrs)Vertebral Le
Involved
1 1996 33/M Left L4–L5
2 1997 46/M Right L5–S3 1998 60/F Right L4–L4 1998 74/M Right L5–S
5 2000 70/F Left C2–C36 2001 70/F Right L1–L7 2001 82/M Left L1–L28 2002 54/M Right T8–T99 2003 54/F Right L3–L
able 2 Bacteriological Diagnosis, Extravertebral Complicatio
Patients OrganismEpiduralAbscess
ParA
1 Staphylococcus aureus Yes
2 S. aureus Yes3 S. aureus No
4 S. aureus Yes
5 Streptococcus agalactiae No
6 S. aureus Yes7 Enterococcus faecalis Yes
8 S. aureus Yes
9 S. aureus Yes
10 S. agalactiae No
T, computed tomography; IV, intravenous; MRI, magnetic resonanpyrophosphate isotope scan.
pine in 9.5% (4/42), and the thoracic spine in 4.5% (2/42).he most commonly involved site was the L4-L5 level (40%f cases), followed by the L3-L4 level (17%), and L5-S1 level14%).
linical PresentationFJI was community-acquired in 93% (39/42) of the casesnd nosocomial in 7% (3/42). All nosocomial cases of PFJIere related to infection of indwelling vascular catheters.The mean duration of symptoms before hospital admis-
ion was 36 � 45 days (range, 2 days to 6 months). In 69%29/42) of the cases, the average duration of symptoms
ging; ND, not done; 67Ga, gallium-67 citrate scan; 99Tc, technetium
vel
151
2
44
s; ESR
ns, Tr
avertbsce
Yes
YesYes
Yes
No
NoYes
Yes
Yes
No
ce ima
Spinal abscess obtained by CT-guided biopsy or by surgery.
bctfe
pacgtpwt
itnp
PIptmil
T
taneou
T
S
UC
UU
E
C
Spontaneous pyogenic facet joint infection 275
efore diagnosis was less than 1 month. No significantorrelation between the age of the patients and the dura-ion of symptoms was found. No significant difference wasound in the duration of symptoms among the differenttiologic agents.
Clinically, severe pain was the presenting complaint in allatients. Presentations ranged from subacute to acute, and inll cases pain became constant without relief with rest. Inases of PFJI involving the thoracic or lumbar spine, pain wasenerally unilateral, usually radiating into the flank or but-ocks, and was often accompanied by muscle spasm andainful palpation of the paravertebral musculature. Patientsith cervical involvement were clinically characterized by
phlebitis, and staphylococcalbacteremia (2 months earlier)
Rig
Yes Metastatic breast cancer AcuYes None LefNo COPD UTYes None NoYes None NoYes Hepatic cirrhosis Cu
able 2 Continued
Other Site ofIsolation
DiagnosticImaging Studies
ynovial fluid 67Ga, 99Tc, MRI IV cloxacgentam
rine 67Ga, 99Tc, CT, MRI IV cloxacerebrospinal fluid MRI IV cloxac
gentam67Ga, 99Tc, MRI IV cloxac
and cip67Ga, 99Tc, MRI IV penicil
2 weekrine 67Ga, 99Tc, MRI IV cloxacrine 67Ga, 99Tc, MRI IV ampici
weeks;weeks
pidural/paravertebralabscess
MRI IV cloxac
67Ga, 99Tc, MRI IV cloxacoral rifaweeks
utaneous ulcer MRI IV penicil2 week
Fever was noted in 83% (35/42) of the cases. Neurologicmpairment was present in nearly 48% (20/42) of the pa-ients: 16 (38%) had radiculopathy, and 4 (9.5%) had severeeurologic deficits (hemiparesis in 1 patient, paraparesis in 1atient, and cauda equina syndrome in 2 patients).
redisposing Factorsn 38% (16/42) of the patients, 1 or more systemic predis-osing factors for infection were found. In 33% (14/42) ofhe cases, the predisposing factor was an underlying im-unosuppressive disease; the most common associated
llnesses were diabetes mellitus, malignancies, chroniciver diseases, and alcoholism (see Table 6). Patients with
nderlying immunosuppressive diseases were on averagelder than patients without these conditions, but the dif-erence was not statistically significant (61 � 14 yearsersus 57 � 16 years, respectively; P � NS). Finally, 5% ofhe patients (2/42) had an extraspinal infection due to theame microorganism several months before the diagnosis
able 3 Clinical Characteristics of 32 Reported Adult Cases
Patient(Reference) Year
Age/Sex(yrs)
VertebralLevel
Involved Feve
1 (2) 1987 66/M Right L3–L4 Yes2 (3) 1987 45/F Left L4–L5 No3 (4) 1988 66/M Left L4–L5 No
4 (5) 1989 65/M Right L2–L3 Yes
5 (6) 1989 63/F Left C4–C5 No
6 (6) 1989 63/M Left C4–C5 No7 (6) 1989 82/M Left L4–L5 Yes8 (8) 1991 76/M Right C1–C2,
odontoidprocess
Yes
9 (9) 1992 68/M Left L3–L4 Yes10 (10) 1992 59/M Left L5–S1 Yes11 (11) 1993 71/F Right L4–L5 Yes12 (12) 1995 61/M Right L4–L5 Yes
13 (14) 1996 23/M Left L5–S1 Yes14 (14) 1996 62/M Right L2–L3 Yes15 (14) 1996 20/M Right L4–L5 Yes16 (15) 1997 60/F Left L5–S1 Yes17 (15) 1997 63/M Right L4–L5 Yes
18 (15) 1997 74/F Right L3–L4 Yes19 (15) 1997 49/M Left L2–L3 Yes20 (15) 1997 66/F Right L4–L5 Yes21 (15) 1997 63/M Right L4–L5 Yes22 (16) 1997 55/F Left L5–S1 Yes23 (17) 1997 41/F Left L4–L5 Yes24 (18) 1998 56/M Left L4–L5 Yes25 (21) 2000 38/M Right L3–L4 Yes
26 (22) 2000 54/F Right L4–L5 Yes
27 (24) 2000 42/M Right T3–T4 Yes28 (24) 2000 65/M Left L4–L5 Yes29 (25) 2000 35/F Right L2–L3 Yes 30 (27) 2001 86/F Left L4–L5 Yes31 (29) 2001 66/M Right L3–L4 No32 (31) 2002 73/M Left L4–L5 No
M, diabetes mellitus; NA, data not available; PMR, polymyalgia rh
f PFJI. v
oncomitant Infectious Processesn 36% (15/42) of patients, 1 or more concomitant infec-ious processes due to the same microorganism wereound (Table 7). Patients with concomitant infectious pro-esses were on average older than those without, but theifference was not statistically significant (60 � 15 years
ne Buttock abscess 76 9000ne None NA NAne None 108 7600
UTI 66 11,700ne None 95 7200ne UTI NA 22,500ne None NA 17,000ng-termhemodialysis
Soft tissueabscess
NA 11,700
rcinoma insitu of thecervix
None 135 20,000
ne None NA NAne None NA NAne None 51 NAne Endocarditis 72 NAne None 88 24,500
None 140 NA
ica; UTI, urinary tract infection; WBC, white blood cell count.
of Spo
rP
NoNoPh
St
DM
DMNoNo
NoNoNoNo
SpAlNoNoDM
NoNoNoDMNoNoNoLo
Ca
NoNoNoNoNoDM
ersus 57 � 15 years, respectively; P � NS). The fre-
qpc[
LD((a5(
TR
A*†
TP
>80 3 (6)
Spontaneous pyogenic facet joint infection 277
uency of underlying conditions causing immunosup-ression was higher in the subgroup of patients with con-omitant infectious processes (50% [8/16] versus 31%8/26], respectively; P � 0.05).
aboratory Testsata about ESR were available in 32 patients; it was elevated
above 30 mm/h) in all cases. Mean ESR was 88 � 26 mm/hrange 46 to 140 mm/h). Data about WBC count were avail-ble in 34 patients. The mean WBC count was 12,081 �111 � 109/L (range, 5300 to 24,500 � 109/L). In 53%
ent, and Outcome of 32 Adult Cases of Spontaneous PFJI
ParaspinalExtension BC
SpinalSpecimen
Culture* Treatment
No � ND ATBNo � � ATBNo � � ATB � SurgeryYes � � ATB � SurgeryNo NA � ATB � Surgery
Yes NA � ATBYes � NA ATB � SurgeryYes NA � ATB � SurgeryYes � ND ATBNo � ND ATBYes � � ATBYes � ND ATBYes � ND ATBNo � � ATB � Surgery
a not available; ND, not done.
able 4 Bacteriological Diagnosis, Extravertebral Extension, Treatmeported Previously (1972–2003)
Patient(Reference) Organism
EpiduralExtension
1 (2) Staphylococcus aureus No2 (3) S. aureus Yes3 (4) S. aureus Yes4 (5) S. aureus Yes5 (6) Staphylococcus epidermidis
and Propionibacteriumsp.
No
6 (6) S. aureus No7 (6) S. aureus No8 (8) S. aureus9 (9) Klebsiella pneumoniae No
10 (10) S. aureus No11 (11) S. aureus No12 (12) S. aureus No13 (14) S. aureus No14 (14) S. aureus Yes15 (14) S. aureus Yes16 (15) S. aureus Yes17 (15) S. aureus Yes18 (15) S. aureus No19 (15) S. aureus Yes20 (15) S. aureus Yes21 (15) S. aureus Yes22 (16) S. aureus Yes23 (17) S. aureus, Streptococcus
viridans, and Neisseriasubflava
Yes
24 (18) S. aureus Yes25 (21)† S. aureus Yes26 (22) S. aureus Yes27 (24) S. aureus Yes28 (24) S. aureus No29 (25) S. aureus No30 (27) S. aureus No31 (29) S. aureus Yes32 (31) Streptococcus agalactiae
(GBS)Yes
TB, antibiotic; BC, blood cultures; GBS, group B streptococcus; NA, datSpinal specimens obtained by CT guided biopsy or by surgery.All patients survived except for patient 25
able 5 Age Distribution in 42 Patients with SpontaneousFJI
acteriological Diagnosishe infection was monomicrobial in 95% (40/42) of theases; in 5% (2/42) of patients more than 1 microorganismas responsible for the process.The causative microorganisms in the 42 cases of sponta-
eous PFJI are listed in Table 8. Staphylococcus aureus was theost common etiologic organism (86% of the cases), fol-
owed by Streptococcus species (9% of the cases); Gram-neg-tive organisms were responsible for 2% of the cases.
Blood cultures gave positive results in 81% of the cases30/37) in which this information was reported. In patients inhom an associated infectious process was observed, the
requency of positive blood cultures was higher than in thoseho did not present with this condition (100% [16/16] ver-
us 68% [15/21], respectively; P � 0.05).Etiologic diagnosis was established on the basis of blood
ultures alone in 62% (26/42) of the cases; in 29% (12/42) ofatients the etiological diagnosis was established on the basisf positive cultures of spinal specimens, obtained by ultra-ound or computed tomography (CT) -guided percutaneouseedle biopsy, or by surgery. In 4 patients (9%), both blood
able 6 Predisposing Factors in 42 Patients with Spontane-us PFJI
nd spinal cultures were positive, identifying the same mi-roorganism.
maging Testsll patients underwent plain radiography of the spine, 29
69%) had CT, 29 (69%) had MRI, 26 (62%) had technetiumone scans, 11 (26%) had gallium scans, and 1 (2%) hadingle-photon emission computed tomography (SPECT). Inhe majority of cases, a combination of studies was employedo reach the diagnosis.
Plain radiographs were usually of no value in the firsttages of the disease. Technetium-99 scintigraphy showedncreased uptake of tracer at the involved vertebral level in allases tested, as early as 3 days after symptom onset. In con-rast to spondylodiscitis, the uptake was usually more verti-ally oriented than horizontally, although in the majority ofases it was difficult to distinguish between the 2 conditionsased only on scintigraphy findings. SPECT was used in 1ase to show more clearly that the infection was in the pos-erior elements. Gallium-67 scintigraphy was positive foracet joint infection in 82% (9/11) of the cases tested; the 2egative scans were obtained at 3 days and 5 weeks, respec-ively, after the beginning of symptoms.
In all patients CT and/or MRI was performed. MRI wasoth sensitive and specific in diagnosing PFJI as early as 2ays after the onset of symptoms, especially when enhancedith gadolinium. By contrast, CT can appear normal early in
he disease. In the present series CT only detected abnormal-ties after 2 weeks of symptoms. Both imaging techniqueshowed additional information about the presence of extra-ertebral extension (epidural, paraspinal, or psoas abscess).T scan was also of great value in guiding percutaneouseedle biopsies of the paravertebral abscesses or of the con-erned facet joint.
Paraspinal and/or epidural extension was seen in 81% (34/2) of the patients: paraspinal alone (paravertebral musclesnd/or psoas muscle) in 21% (9/42) of the patients, epidural
able 8 Causative Microorganisms in 42 Patients with Spon-aneous PFJI*
lone in 12% (5/42), and both in 48% (20/42) of the patients.
Some patients had more than 1 process.
Twsfiohns
TAml8Cawpsma
iroc
DIttopau
2TDwpimcMmp
nstWPnpa
tamfcrtta
dsdrnttsmtddsdt
mcipsn
tTwaTvi
wiut
strismc
sB
Spontaneous pyogenic facet joint infection 279
hus, in nearly 60% (25/42) of the cases, epidural extensionas found. All patients with neurologic impairment pre-
ented with extravertebral extension, but the presence of thisnding was not necessarily associated with the developmentf neurologic symptoms. There was no association betweenigh leukocyte count and the presence of epidural or paraspi-al extension. Diagnostic delay and etiologic agents wereimilar in patients with and without extravertebral extension.
reatment and Outcomell patients initially received parenteral antibiotics, with aean duration of 4 weeks (range, 1 to 6 weeks), often fol-
owed by oral therapy (mean 1 � 3 months; range, 2 weeks tomonths). In addition to antibiotic treatment, percutaneousT-guided drainage of the infected joint and/or paraspinalbscess was performed in 19% (8/42) of the patients. Surgeryas indicated in 26% (11/42) of the patients, mainly in theresence of severe neurologic dysfunction. Bed rest was pre-cribed in all cases for at least 3 weeks; usually, when move-ent was authorized, external immobilization was indicated
t least during the first month.One of the 42 patients died due to an asystolic arrest dur-
ng surgery. Of the remaining 41 patients, 1 became parapa-etic; the rest were cured without functional sequelae or withnly residual pain, which did not impede an active life. Noases relapsed.
iscussionn this study we present the largest published series of pa-ients with PFJI seen in a tertiary care institution and reviewhe available literature in an effort to increase the knowledgef this disease. To standardize the information, patients withrior spinal instrumentation or surgery, injection drug users,nd pediatric patients were excluded, since all of these pop-lation groups present specific differential characteristics.Data from our institution reveal that PFJI represents nearly
0% of the cases of spontaneous pyogenic spinal infections.his prevalence is higher than that previously reported byavid-Chausse and coworkers (1) and Muffoletto and co-orkers (30) in the 1980s. These discrepancies may be, inart, attributed to methodologic differences but, in our opin-
on, it is mainly a consequence of the routine employment ofore sensitive imaging techniques during recent years. PFJI
ould easily be misdiagnosed as spondylodiscitis if CT orRI is not used. Similarly, some epidural infections or psoasuscle abscesses believed to be primary may in fact be com-lications of unrecognized PFJI.The demographic characteristics of patients with sponta-
eous PFJI were similar to those reported with spontaneouspondylodiscitis (35-42). Data from this review demonstratehat PFJI mainly appears in patients over the age of 60 years.
e did not find differences between the age of patients withFJI and the rest of patients with spontaneous pyogenic spi-al infections diagnosed at our hospital during the studyeriod. As occurs in the majority of osteoarticular infections,
male predominance was seen. b
As in spondylodiscitis (35-42), PFJI is most common inhe lumbosacral area. Of interest, three-quarters of the casesre located between L3 and S1, the L4-L5 level being theost common site involved. Patients with PFJI are usually
ebrile and present with severe pain and stiffness that be-omes constant without relief upon resting. The presence ofadicular symptoms is relatively frequent (in nearly 40% ofhe cases), probably as a consequence of the localization ofhe infectious process; however, severe neurologic deficitsre rare, occurring in less than 10% of the cases.
It is difficult to differentiate between PFJI and spondylo-iscitis based only on clinical grounds. There are, however,ome clues that may be useful. When compared with spon-ylodiscitis, PFJI symptoms tend to be unilateral, usuallyadiating into the flank or buttocks, and are often accompa-ied by muscle spasm and painful palpation of the paraver-ebral musculature. In addition, the mean duration of symp-oms before diagnosis of PFJI is shorter than those reported inpondylodiscitis (35-42). In PFJI, symptoms tend to presentore acutely and with greater severity in the early stages. In
he absence of neurologic deficit or fever, a patient may beiagnosed with acute lumbago if further investigation is notone. Thus, in uncomplicated cases a high degree of clinicaluspicion is necessary. In Table 9 we summarized the mainifferences and similarities between PFJI and spondylodisci-is according to our experience (35).
Of interest, no systemic risk factors for PFJI are detected inore than 60% of patients, indicating that spontaneous PFJI
an also affect immunocompetent patients without debilitat-ng conditions. When underlying medical illnesses areresent, they are similar to those observed in spontaneouspondylodiscitis (35-42), including diabetes mellitus, malig-ancies, alcoholism, and chronic liver diseases.In nearly 36% of patients, 1 or more concomitant infec-
ious processes due to the same microorganism were found.his subgroup of patients was on average older than thoseithout them. Moreover, in most of these cases, bacteremia
nd underlying immunosuppressive illness were present.he most common diagnoses were peripheral and sternocla-icular arthritis, skin and soft-tissue infections, urinary tractnfections, and endocarditis.
Laboratory data also are similar to those found in patientsith spondylodiscitis. Leukocyte count is elevated in approx-
mately 50% of cases. The ESR is uniformly elevated, being aseful parameter for monitoring the course of the disease andhe response to treatment.
Regarding etiologic diagnosis, the causal microorganism inpontaneous PFJI is typically S. aureus, accounting for 86% ofhe reported cases, a prevalence clearly higher than the oneeported in the series of spondylodiscitis (35,36,38-42). Ofnterest, Gram-negative bacilli, an important pathogen inpondylodiscitis of the elderly (35,36,41-43), are an uncom-on cause of spontaneous PFJI, producing only 2% of the
ases.Hematogenous spread from a distant focus is the pre-
umed route of infection in the cases of spontaneous PFJI.acteremia was demonstrated in 81% of the cases in which
lood culture results were available, stressing the usefulness
oaeabcmn
o(ipdt
tp
Pcbna(parid
T
D
L
C
E
I
280 J. Narváez et al
f this microbiological technique to establish the causativegent. In fact, in 62% of patients the etiologic diagnosis wasstablished on the basis of blood cultures, since PFJI is almostlways a monomicrobial disease and isolates from positivelood cultures can be assumed to be responsible for the pro-ess. When blood cultures are negative, spinal specimensust be obtained, preferably by CT-guided percutaneouseedle biopsy, or by open surgical biopsy when necessary.Several cases of PFJI have been described as a complication
f spinal procedures, such as therapeutic joint infiltration20,23,32,33) or placement of epidural catheters (10,28). Its strange that the number of these cases is not higher (7atients), taking into account the number of invasive proce-ures that are performed in the lumbar region. Thus, al-
hough rare, PFJI should be suspected as a complication of e
hese interventions in patients whose pain worsens after therocedure, especially if fever appears.Data from this review demonstrate that the diagnosis of
FJI was based mainly on imaging study findings and thatlinical data failed in the majority of cases to discriminateetween PFJI and spondylodiscitis. Plain films are, in general,ot useful for diagnosis, since the radiologic abnormalitiesppear 6 to 12 weeks after the onset of symptoms2,4,12,15,18,30). The signs that appear earlier in obliquelain radiographs are soft-tissue swelling, with narrowingnd irregularity of joint space. Technetium-99 bone scintig-aphy seems to be 100% sensitive in detecting facet jointnvolvement, but is not specific (12,14). In contrast to spon-ylodiscitis, the uptake in PFJI is usually more vertically ori-
Spondylodiscitis
(55%)
in
Mainly in patients over 60 years (60%)Male predominance (56%)
Predisposing factor for infection in50% of the cases
nted than horizontally oriented (3,5,10,12,15). Gallium-67
Spontaneous pyogenic facet joint infection 281
Figure 1 Pyogenic arthritis of the right L5-S1 facet joint. A. Enhanced CT scan shows joint destruction, with wideningof the joint space and erosive changes (white arrows). There is also periarticular inflammatory changes, without clearfluid collections, in the corresponding foramina and posterior paraspinal muscles (black arrowheads), as well as in theepidural space (black arrows). B. Sagittal T1-weighted MR image demonstrates bone marrow edema and irregularity ofthe joint surfaces (short arrows), and extensive periarticular inflammatory changes in the posterior paraspinal muscles(long arrows) and in the neural foramina. C. sagittal short tau inversion recovery (STIR) MR image better depicts the
joint and periarticular inflammatory changes.
stSptMoeumftsisMtt4satesea
wot
fdwepq
oiawor
diaotpue
demo(m
oqal
R
1
FTbcc
282 J. Narváez et al
cintigraphy may be falsely negative in cases of PFJI if ob-ained earlier than 1 week after symptom onset (8,12).PECT can be used to confirm that the infection is in theosterior elements (5,32), but currently this is still a researchool. The most useful imaging techniques in PFJI are CT and
RI. CT is the investigation that best delineates the extent ofsseous lesions of the facet joints, but may appear normalarly in the disease (25,44,45). For this reason, it is moreseful in patients with subacute or chronic illness. CT abnor-alities include loss of subchondral bone associated with the
acet joint, loss of ligamentum flavum density, and oblitera-ion of fat planes. The presence of paraspinal soft-tissue le-ions and/or collections can also be detected by CT usingntravenous injection of contrast; however, CT depiction ofoft-tissue structures is less sensitive than MRI (44-46).oreover, CT provides radiologic guidance for interven-
ional procedures (ie, biopsy, drainage). MRI is consideredhe modality of choice for diagnosis of PFJI (18,30,34,45-8). High-quality MRI with gadolinium enhancement is bothensitive and specific in diagnosing PFJI as early as 2 daysfter the onset of symptoms (18), ruling out involvement ofhe disc space and vertebral body. Moreover, it is especiallyffective for evaluating the neural structures of the spine (ie,pinal cord, nerve roots) and the spread of infection to thepidural space and/or paraspinal soft tissues (see Figs. 1nd 2).
Paraspinal and/or epidural extension are a frequent findinghen MRI and/or CT are used for the diagnosis of PFJI (81%f patients in this review). As other authors (49), we used the
igure 2 Pyogenic arthritis of the left L5-S1 facet joint. An axial2-weighted MR image shows widening of the joint space, occupiedy fluid, and a well-defined fluid collection in the paraspinal mus-les (long arrows) corresponding to an abscess. Note also small fluidollection in the posterior epidural space (short arrows).
erm extension, since we did not attempt to differentiate dif-
use inflammation or granulation tissue from a true abscess,ue to the retrospective nature of our study. No relationshipas found between extension and clinical findings. Its pres-
nce did not indicate a worse prognosis; in fact, in mostatients, this radiologic finding had no clinical conse-uences.The therapeutic approach to PFJI should be based mainly
n the clinical manifestations and not on the imaging find-ngs. In our experience, medical treatment alone, includingntibiotic therapy, bed rest for 3 weeks, and immobilizationith orthosis during the first month, is suitable for most casesf PFJI. Surgical intervention is indicated when severe neu-ologic dysfunction is present.
The optimal duration of therapy for PFJI has not been wellefined, since all reports are retrospective. Usually, antibiot-
cs are given initially intravenously for at least 4 weeks, andfter this parenteral treatment, the majority of clinicians usesral antibiotics for several weeks. Probably a total period of 6o 8 weeks is suitable, depending on clinical outcome and theresence of extravertebral extension (2,10,12,14,20). These of oral fluoroquinolones and rifampicin may providearly efficacious oral therapy (35).
MRI seems to be useful for monitoring the course of theisease and response to treatment (18,34,36,45-52). How-ver, it is important to bear in mind that signal and enhance-ent changes in MRI can persist after the clinical resolution
f the infection and gradually decrease over weeks to months42,45,46,52). This illustrates the fact that contrast enhance-ent does not always indicate active infection.The overall prognosis of PFJI is good, with a mortality rate
f 2%. Usually, patients are cured without functional se-uelae or with only residual pain, which did not impede anctive life. Of interest, after appropriate medical therapy, re-apse seems to be an exceptional event.
eferences1. David-Chausse J, Dehais J, Boyer M, Darde M, Imbert Y. Les infections
articulaires che l’adulte atteintes peripheriques et vertebrales a germesbanals et a bacilles tuberculeux. Rev Rhum 1981;48:69-76.
2. Halpin DS, Gibson RD. Septic arthritis of a lumbar facet joint. J BoneJoint Surg Br 1987;69:457-9.
3. Bellen P, Maldague B, Zech F. Arthrite septique d’une articulation in-terapophysaire vertébrale posterieure: à propos d’une cas. Rev ChirOrthop Reparatrice Appar Mot 1987;73:289-91.
4. Roberts W. Pyogenic vertebral osteomyelitis of a lumbar facet joint withassociated epidural abscess: a case report with review of the literature.Spine 1988;13:948-52.
5. Swayne LC, Dorsky S, Caruana V, Kaplan IL. Septic arthritis of a lumbarfacet joint: detection with bone SPECT imaging. J Nucl Med 1989;30:1408-11.
6. Ehara S, Khurana JS, Kattapuram SV. Pyogenic vertebral osteomyelitisof the posterior elements. Skeletal Radiol 1989;18:175-8.
7. Rousselin B, Gires F, Vallée C, Chevrot A. Case report 627: septicarthritis of lumbar facet joint as initial manifestation of spondylodisci-tis. Skeletal Radiol 1990;19:453-5.
8. Halla JT, Bliznak J, Hardin J, Finn S. Septic arthritis of the C1-C2 lateralfacet joint and torticollis: pseudo-Grisel’s syndrome. Arthritis Rheum1991;34:84-8.
9. Chevalier X, Marty M, Larget-Piet B. Klebsiella pneumoniae septic ar-thritis of a lumbar facet joint. J Rheumatol 1992;19:1817-9.
0. Peris P, Brancos MA, Gratacós J, Moreno A, Miró JM, Muñoz J. Septic
arthritis of spinal apophyseal joint. Spine 1992;17:1514-6.
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
44
4
4
4
55
5
Spontaneous pyogenic facet joint infection 283
1. Daumen-Legre V, Chaine M, Chagnaud C, Lafforgue P, Schiano A,Acquaviva PC, et al. Arthrite septique interapophysaire vertebrale pos-terieure d’origine hematogene: une nouvelle observation et revue de lalittérature (abstract). Rev Rhum 1993;60:743.
2. Dauwe DM, Van Oyen JJ, Samson IR, Hoogmartens MJ. Septic arthritisof a lumbar facet joint and a sternoclavicular joint. Spine 1995;20:1304-6.
3. Heenan SD, Britton J. Septic arthritis in a lumbar facet joint: a rare causeof epidural abscess. Neuroradiology 1995;37:462-4.
4. Douvrin F, Callonnec F, Proust F, Janvresse A, Simonet J, Thiebot J.Arthrite septique interapophysaire lombaire: a propos de 3 cas. J Neu-roradiol 1996;23:234-40.
5. Ergan M, Macro M, Benhamou CL, Vandermarcq P, Colin T,L=Hirondel JL, et al. Septic arthritis of lumbar facet joints. A review ofsix cases. Rev Rhum Engl Ed 1997;64:386-95.
6. Ben Hamouda M, Rajhi H, Golli M, Bergaoui N, Chaouch A, Hassine H,et al. Arthrite septique interapophysaire postérieure lombaire. J Radiol1997;78:373-6.
7. Baltz MS, Tate DE, Glaser JA. Lumbar facet joint infection associatedwith epidural and paraspinal abscess. Clin Orthop 1997;339:109-12.
8. Fujiwara A, Tamai K, Yamato M, Yoshida H, Saotome K. Septic arthritisof a lumbar facet joint: report of a case with early MRI findings. J SpinalDisord 1998;11:452-3.
9. Louie T, Morgan B, Standiford HC. Facet and facet-joint infections: casereport and review. Clin Infect Dis 1998;26:510-2.
0. Cook NJ, Hanrahan P, Song S. Paraspinal abscess following facet jointinjection. Clin Rheumatol 1999;18:52-3.
1. Dayer MJ, Gransden W, Goldsmith DJ. Facet joint osteomyelitis in apatient on long-term hemodialysis. Am J Kidney Dis 2000;36:1041-4.
2. Hickey NA, White PG. Septic arthritis of a lumbar facet joint causingmultiple abscesses. Clin Radiol 2000;55:481-3.
3. Magee M, Kannangara S, Dennien B, Lonergan R, Emmett L, Van derWall H. Paraspinal abscess complicating facet joint injection. Clin NuclMed 2000;25:71-3.
4. Pilleul F, Garcia J. Septic arthritis of the spine facet joint: early positivediagnosis on magnetic resonance imaging. A review of two cases. RevRhum Fr Ed 2000;67:237-40.
5. Rombauts PA, Linden PM, Buyse AJ, Snoecx MP, Lysens RJ, GryspeerdtSS. Septic arthritis of a lumbar facet joint caused by Staphylococcusaureus. Spine 2000;25:1736-8.
6. Ishibe M, Inoue M, Saitou K. Septic arthritis of a lumbar facet joint dueto pyonex. Arch Orthop Trauma Surg 2001;121:90-2.
7. Derouet N, Haettich B, Temmar Z, Dugard D, Puechal X. Septic arthri-tis of a lumbar facet joint. A case report. Ann Med Interne (Paris)2001;152:279-82.
8. Glaser JA, El-Khoury GY. Unknown case; diagnosis: facet joint septicarthritis T12-L1 on the left with extension of the infection into thespinal canal producing a large epidural abscess. Spine 2001;26:991-2.
9. Marson F, Cognard C, Guillem P, Sévely A, Manelfe C. Septic arthritisof a lumbar facet joint associated with epidural and paravertebral softtissue abscess. J Radiol 2001;82:63-6.
1. Coscia MF, Trammell TR. Pyogenic lumbar facet joint arthritis with
intradural extension: a case report. J Spinal Disord Tech2002;15:526-8.
2. Okazaki K, Sasaki K, Matsuda S, Yuge I, Omiya K, Kido H, et al.Pyogenic arthritis of a lumbar facet joint. Am J Orthop 2002;29:222-4.
3. Orpen NM, Birch NC. Delayed presentation of septic arthritis of alumbar facet joint after diagnostic facet joint injection. J Spinal DisordTech 2003;16:285-7.
4. Doita M, Nishida K, Miyamoto H, Yoshiya S, Kurosaka M, NabeshimaY. Septic arthritis of bilateral lumbar facet joints. Report of a case withMRI findings in the early stage. Spine 2003;28:E198-E202.
7. Guri JP. Pyogenic osteomyelitis of the spine: differential diagnosisthrough clinical and roentgenographic observations. J Bone Joint Surg1946;28:29–9.
8. Liebergall M, Chaimsky G, Lowe J, Robin GC, Floman Y. Pyogenicvertebral osteomyelitis with paralysis: prognosis and treatment. ClinOrthop 1991;269:142-50.
9. Perronne C, Saba J, Behloul Z, Salmon-Ceron D, Leport C, Vilde JL, etal. Pyogenic and tuberculous spondylodiskytis (vertebral osteomyeli-tis) in 80 adult patients. Clin Infect Dis 1994;19:746-50.
1. Colmenero JD, Jimenez Mejías ME, Sánchez Lora FJ, Reguera JM, Pal-omino Nicás J, Martos F, et al. Pyogenic, tuberculous, and brucellarosteomyelitis: a descriptive and comparative study of 219 cases. AnnRheum Dis 1997;56:709-15.
2. Belzunegui J, Del Val N, Intxausti JJ, De Rios JR, Queiro R, González C,et al. Vertebral osteomyelitis in northern Spain. Report of 62 cases. ClinExp Rheumatol 1999;17:447-52.
3. Sapico FL. Microbiology and antimicrobial therapy of spinal infections.Orthop Clin North Am 1996;27:9-13.
4. Tay BK, Deckey J, Hu SS. Spinal infections. J Am Acad Orthop Surg2002;10:188-97.
