Keywords
sepsis - thrombosis - fusobacterium
Introduction
Pharyngitis leading to sepsis was reported by Schottmuller[1] in 1918, and in 1936 Andre Lemierre reported on a series of 20 similar cases of
postpharyngitis anaerobic sepsis.[2] Only two of the 20 patients survived, and the syndrome, classically comprised of
pharyngeal infection, metastatic septic emboli, internal jugular vein thrombosis,
and the presence of the anaerobic bacterium Fusobacterium necrophorum, was eventually named after Lemierre. Lemierre syndrome is an uncommon disease in
the postantibiotic era, but can carry a high mortality rate if not recognized and
treated aggressively.[3] The most commonly associated organism, F. necrophorum, is a gram-negative, anaerobic rod-shaped bacterium that is part of the normal oral
flora, but multiple other bacteria have been implicated in Lemierre syndrome, such
as streptococcal species, Eikenella corrodens, and Staphylococcus aureus.[4] The venous thrombosis seen in Lemierre syndrome is likely the result of endothelial
dysfunction caused by inflammatory factors from the local infection. [Fig. 1] shows axial magnetic resonance imaging (MRI) demonstrating left internal jugular
vein thrombosis resulting from streptococcal Lemierre syndrome in a young female patient.
Fig. 1 Axial MRI image showing left jugular vein thrombosis in a patient with Lemierre syndrome.
Recent studies have examined the role of anticoagulation in the treatment of Lemierre
syndrome,[5]
[6] but there is no current consensus on the role of anticoagulation and its effect
on vessel recanalization in this syndrome. Recent studies have shown equivocal results,
with 1 recent retrospective study showing that all patients had improvement in their
thrombi by 3 months, with no evident effect on thrombosis outcomes regardless of anticoagulation,[5] and another showing ∼ 91% thrombosis improvement and a low complication rate, but
with no non-anticoagulated group for comparison.[6] To our knowledge, there is no meta-analysis available examining the effect of anticoagulation
in Lemierre syndrome on vessel recanalization/thrombus resolution, or on mortality.
In the present meta-analysis, the aim was to evaluate the effect of anticoagulation
on vessel recanalization in patients with Lemierre syndrome with documented follow-up
imaging, to examine the effect of anticoagulation on mortality, and to collect data
regarding location of thrombosed vessels, antimicrobial treatment type, causative
microorganisms isolated on culture, and frequency of associated complications such
as cranial nerve palsies.
Review of the Literature
A thorough literature search of published studies was performed in the Pubmed database
using the keywords Lemierre syndrome. Studies not published in English were translated using Google translate (Google/Alphabet,
Inc., Mountain View, California, US). Each study was reviewed for relevance and availability
of individual patient data. The present study was exempt from Institutional Review
Board evaluation, given the use of de-identified retrospective patient data, non-involvement
of human subjects, and published literature. The present study was conducted in accordance
with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA)
guidelines statement.[7] [Fig. 2] shows the flow diagram of the study selection. Patients were included in the meta-analysis
if they had evidence of a head-and-neck infection such as pharyngitis/tonsillitis
or mastoiditis with at least one other aspect of Lemierre syndrome: metastatic septic
emboli or thrombosis of the internal jugular vein, or other documented venous or arterial
septic thrombus. Patients with a purely orthopedic or abdominal variant of Lemierre
syndrome (extremity infection with thrombosis only of an extremity vessel such as
the iliac vein, for example) were excluded, but patients with an abdominal, spinal,
or orthopedic thrombosis in conjunction with a head-and-neck infectious source or
thrombosed head-and-neck vessel were included. Patients in whom anticoagulation was
stopped prior to the attainment of therapeutic levels due to thrombocytopenia or allergic
reaction were considered not anticoagulated. Statistical analyses were performed using
the Prism (Graphpad Software, San Diego, California, US) software for the Fisher,
chi-squared, and number needed to treat (NNT) analyses, and the Medcalc (Medcalc Software,
Ostend, Belgium) software for odds ratio (OR) analyses. A two-sided p-value was used and the 95% confidence interval (95%CI) was used for outcome variables.
Fig. 2 Flow diagram of the study selection.
A total of 676 papers were found and reviewed. The abstracts were screened, and review
articles or articles involving only wholly abdominal or orthopedic cases were excluded.
A total of 427 studies remained, and these were reviewed for the availability of individual
patient data that could be analyzed. A remaining 359 studies ([Supplementary Appendix A] [online only]) totaling 394 patients had individual patient data that could be examined.
Studies were examined for data regarding the location of a thrombosed vessel if present,
patient gender, patient age, presence of septic pulmonary or neural tissue emboli,
presence of cranial neuropathies, type of antibiotic treatment, microorganisms isolated
through blood or tissue cultures, mortality, use of anticoagulation, and resolution
of vessel thrombosis on follow-up imaging by ultrasound, computed tomography (CT),
or MRI. A total of 50 patients had specific data on resolution or persistence of vessel
thrombosis on follow-up imaging and use of anticoagulation, and were included in the
vessel recanalization meta-analysis, and 194 patients had data on mortality and use
of anticoagulation, and were included in the mortality meta-analysis.
A total of 300 patients had data specifically mentioning the presence or absence of
thrombosis. Of these, 283/300 (94.3%), showed evidence of thrombosis, while 17/300
(5.7%) showed no evidence of vascular thrombosis. Of the 394 patients included in
the study, 120/394 (30.5%) were treated with antibiotics along with some surgical
procedure, while the remaining 274/394 (69.5%) were treated with antibiotics only.
The average age was 28.8 years, and the median age was 23 years. The majority of the
patients were between ages 7 and 54, with smaller groupings between ages 2 months
to 5 years and 56 to 76 years, with outliers at 80, 81, and 85 years ([Fig. 3]). Of the 344 patients with data reported on gender, 207/344 (60.2%) were male, while
the remaining 137/344 (39.8%) were female. There were 16 deaths for a mortality rate
of 4.1%. There were 8 patients who developed meningitis, resulting in a meningitis
rate of 2.0%. Two patients were pregnant at the time of their diagnosis of Lemierre
syndrome. Septic pulmonary emboli were noted in 148/394 patients (37.6%), while brain
emboli were noted in 12/394 patients (3.0%).
Fig. 3 Age distribution of patients with Lemierre syndrome.
The average length of anticoagulation treatment was 81.1 days (range 16–183 days).
The average length of antibiotic treatment (intavenous [IV] + oral) was 33.7 days
(range 14–60 days). The average time to follow-up imaging was 51.8 days (range 3–365
days). Data collected on the specific vessel or vessels thrombosed, the microbes isolated
from blood or tissue cultures, the type of anticoagulation used, and the type of antibiotic
therapy used are summarized in [Tables 1]
[2]
[3]
[4]. Most patients were treated with multiple antibiotics during their treatment, and
many patients had thrombosis of multiple vessels. Of the 194 patients for whom use
of or lack of use of anticoagulation was specifically noted, 123 (63.4%) were anticoagulated,
while 71 (36.6%) were not anticoagulated. A total of 23 patients (5.8%) were noted
to have cranial nerve palsies during their clinical course. The cranial nerves involved
are summarized in [Table 5] (several patients had multiple cranial neuropathies). One patient was also noted
to have a Horner syndrome/sympathetic trunk palsy as part of their clinical course.
Four patients were noted to have concomitant Epstein-Barr virus (EBV) infection evidenced
by positive EBV immunoglobulin M (IgM), and four patients were noted to have a concomitant
malignancy (one laryngeal squamous cell carcinoma in the setting of recurrent respiratory
papillomatosis, one serous borderline tumor of the ovary, one lung adenocarcinoma,
and one uvular squamous cell carcinoma). Of the 120 patients who underwent a surgical
procedure in addition to antibiotic treatment, 3 (2.5%) underwent endoscopic sinus
surgery, 11 (9.2%) underwent mastoidectomy (2 [1.7%] bilateral), 10 (8.3%) underwent
ligation/excision of the thrombosed internal jugular vein, and 1 (0.8%) underwent
ligation/excision of the thrombosed external carotid artery.
Table 1
Thrombosed vessels noted on imaging
|
Thrombosed vessel
|
Number of patients
|
|
Internal jugular vein only
|
158
|
|
Bilateral internal jugular vein
|
11
|
|
Contralateral internal jugular vein
|
1
|
|
Internal jugular vein + another vessel
|
87
|
|
Subclavian
|
10
|
|
Sigmoid
|
36
|
|
External jugular vein
|
19
|
|
Transverse
|
8
|
|
Intracerebral infarct (venous or arterial)
|
4
|
|
Basilic vein
|
1
|
|
Supraclavicular vein
|
1
|
|
Anterior jugular vein
|
2
|
|
Inferior vena cava
|
1
|
|
Iliac vein
|
2
|
|
Superior vena cava
|
2
|
|
Multiple vessels not involving the internal jugular vein
|
5
|
|
Femoral artery
|
1
|
|
Innominate vein
|
1
|
|
Facial vein
|
7
|
|
Thyroid vein
|
1
|
|
Axillary vein
|
4
|
|
Splenic vein
|
1
|
|
Superior mesenteric vein
|
1
|
|
Anterior cerebral artery
|
1
|
|
Middle cerebral artery
|
1
|
|
Maxillary vein
|
1
|
|
Common femoral vein
|
1
|
|
Superior ophthalmic vein or ophthalmic vein
|
9
|
|
Unilateral cavernous sinus
|
20
|
|
Bilateral cavernous sinus
|
7
|
|
Vertebral artery
|
2
|
|
External carotid artery
|
2
|
|
Internal carotid artery
|
10
|
|
Bilateral internal carotid artery
|
1
|
|
Common carotid artery
|
1
|
|
Vertebral vein
|
2
|
|
Peritonsillar vein/pharyngeal plexus/pterygoid plexus
|
3
|
|
Brachiocephalic vein
|
5
|
|
Superior sagittal sinus
|
1
|
|
Straight sinus
|
1
|
|
Retromandibular vein
|
5
|
Table 2
Organisms isolated by culture
|
Culture result
|
Number of patients
|
Percentage
|
|
Fusobacterium necrophorum
|
131
|
33.2%
|
|
Fusobacterium species
|
20
|
5.1%
|
|
Negative culture
|
29
|
7.4%
|
|
Spirocheta/Treponema (Plaut-Vincent angina)
|
1
|
0.3%
|
|
Multiple organisms
|
27
|
6.9%
|
|
Fusobacterium species + another organism
|
13
|
3.3%
|
|
Bacteroides melaninogenicus
|
1
|
0.3%
|
|
Prevotella bivia
|
2
|
0.5%
|
|
Streptococcus intermedius/Streptocossus anginosus/Streptococcus milleri
|
5
|
1.3%
|
|
Streptococcus pneumoniae
|
4
|
1.0%
|
|
Streptococcus viridans
|
3
|
0.8%
|
|
Streptococcus constellatus
|
6
|
1.5%
|
|
Fusobacterium nucleatum
|
9
|
2.3%
|
|
Haemophilus influenza
|
2
|
0.5%
|
|
Proteus species
|
1
|
0.3%
|
|
Klebsiella oxytoca
|
1
|
0.3%
|
|
Corynebacterium species
|
1
|
0.3%
|
|
Bacteroides fragilis
|
4
|
1.0%
|
|
Streptococcal species
|
4
|
1.0%
|
|
Porphyromonas asaccharolytica
|
3
|
0.8%
|
|
Methicillin-sensitive S. aureus
|
6
|
1.5%
|
|
Methicillin-resistant S. aureus
|
16
|
4.1%
|
|
Escherichia coli
|
1
|
0.3%
|
|
Lactobacillus species
|
1
|
0.3%
|
|
Coagulase-negative Staphylococcus
|
4
|
1.0%
|
|
S. aureus unspecified
|
6
|
1.5%
|
|
Arcanobacterium/Arcanobacterium haemolyticum
|
4
|
1.0%
|
|
Proteus mirabilis
|
1
|
0.3%
|
|
Peptostreptococcus species
|
3
|
0.8%
|
|
Streptococcus pyogenes
|
4
|
1.0%
|
|
Slackia exigua
|
1
|
0.3%
|
|
Proprionobacterium species
|
1
|
0.3%
|
|
Mycoplasma pneumoniae
|
1
|
0.3%
|
|
Fusobacterium varium
|
3
|
0.8%
|
|
Klebsiella pneumoniae
|
9
|
2.3%
|
|
Eikenella corrodens
|
1
|
0.3%
|
|
Peptostreptococcus anaerobius
|
2
|
0.5%
|
|
Leptotrichia buccalis
|
1
|
0.3%
|
|
Streptococcus salivarius
|
1
|
0.3%
|
|
Prevotella intermedia
|
2
|
0.5%
|
|
Prevotella species
|
1
|
0.3%
|
|
Streptococcus sanguinis
|
1
|
0.3%
|
|
Group A strep
|
1
|
0.3%
|
|
Group C strep
|
6
|
1.5%
|
|
Group G strep
|
2
|
0.5%
|
|
Group B strep
|
1
|
0.3%
|
|
Pseudomonas aeruginosa
|
1
|
0.3%
|
|
Staphylococcus capitus
|
1
|
0.3%
|
|
Staphylococcus epidermidis
|
2
|
0.5%
|
|
Actinomyces myeri
|
1
|
0.3%
|
|
Parvimonas micra
|
1
|
0.3%
|
|
Chromobacterium violaceum
|
1
|
0.3%
|
Table 3
Type of anticoagulant used
|
Anticoagulant used
|
Number of patients
|
|
Total anticoagulation
|
123
|
|
Total no anticoagulation
|
71
|
|
Heparin/LMWH/unfractionated heparin
|
53
|
|
Xarelto/Rivaroxaban
|
2
|
|
Edoxaban/Savaysa
|
2
|
|
Warfarin/Coumadin
|
29
|
|
Enoxaparin/Lovenox
|
19
|
|
Tinzaparin/Innohep
|
2
|
|
Bemiparin
|
1
|
|
Dabigatran/Pradaxa
|
1
|
|
Fondaparinux/Arixtra
|
4
|
|
Danaparoid/Orgaran
|
1
|
|
Dalteparin/LMWH
|
1
|
|
Nadroparin/Fraxiparine/LMWH
|
1
|
Abbreviation: LMWH, low molecular weight heparin.
Table 4
Type of antibiotic used
|
Antibiotic used
|
Number of patients
|
|
Imipenem or imipenim cilastatin
|
10
|
|
Ceftazidime
|
3
|
|
Cefpodoxime
|
1
|
|
Cefuroxime
|
5
|
|
Cefepime
|
6
|
|
Cefotetan
|
1
|
|
Cefoperazone
|
1
|
|
Flucloxacillin
|
1
|
|
Clindamycin
|
63
|
|
Ampicillin/sulbactam or ticarcillin/clavulanate
|
23
|
|
Metronidazole
|
83
|
|
Chloramphenicol
|
2
|
|
Piperacillin/tazobactam
|
28
|
|
Nafcillin
|
1
|
|
Rifampin
|
1
|
|
Penicillin/amoxicillin/ampicillin
|
34
|
|
Methicillin/flucloxacillin/floxacillin
|
4
|
|
Dicloxacillin
|
2
|
|
Gentamicin
|
12
|
|
Aztreonam
|
1
|
|
Moxifloxacin
|
2
|
|
Ofloxacin
|
1
|
|
Pazufloxacin
|
1
|
|
Daptomycin
|
1
|
|
Trimethoprim/sulfamethoxazole
|
1
|
|
Tigecycline
|
1
|
|
Linezolid
|
4
|
|
Ertapenem
|
3
|
|
Vancomycin
|
44
|
|
Doxycycline
|
1
|
|
Azithromycin
|
2
|
|
Cefotaxime
|
10
|
|
Levofloxacin
|
8
|
|
Ciprofloxacin
|
8
|
|
Roxithromycin
|
1
|
|
Meropenem
|
30
|
|
Amikacin
|
3
|
|
Teicoplanin
|
2
|
|
Biapenem
|
1
|
|
Clarithromycin
|
6
|
Table 5
Cranial nerve palsies observed
|
Cranial nerve palsy
|
Number of Patients
|
|
Total of patients with cranial nerve palsy
|
23
|
|
I
|
3
|
|
II
|
0
|
|
III
|
3
|
|
IV
|
2
|
|
V
|
1
|
|
VI
|
9
|
|
VII
|
3
|
|
VIII
|
3
|
|
IX
|
0
|
|
X (recurrent laryngeal nerve)
|
2
|
|
XI
|
0
|
|
XII
|
5
|
After review of the studies, 50 patients were identified with individual data on anticoagulation
and recanalization on follow-up imaging. Of the patients treated with anticoagulation,
26 (60.4%) showed vessel recanalization and 17 (39.5%) showed persistent vessel thrombosis
on follow-up imaging. Of the non-anticoagulated patients, five (71.4%) showed vessel
recanalization and two (28.5%) showed persistent vessel thrombosis on follow-up imaging.
The OR analysis for anticoagulation versus recanalization demonstrated an OR of 1.6
for anticoagulation versus no anticoagulation and vessel recanalization, with a 95%CI
of 0.3 to 9.4 The two-tailed p-value was 0.6, demonstrating no statistically significant relationship between anticoagulation
and vessel recanalization on follow-up imaging. There were 194 patients identified
with individual data on anticoagulation and mortality. Of the patients treated with
anticoagulation, 120 (97.6%) survived and 3 (2.4%) died, while of the non-anticoagulated
patients, 68 (95.8%) survived and 3 (4.2%) died. The OR for anticoagulation and death
was 0.6, with a 95%CI of 0.1 to 2.9. The two-tailed p-value for anticoagulation and death was 0.5, demonstrating no statistically significant
decrease in mortality with anticoagulation. The number needed to treat (NNT) for anticoagulation
and vessel recanalization was 9, while the number needed to treat for anticoagulation
and mortality was 55. [Table 6] summarizes the OR ratio analysis for anticoagulation and vessel recanalization and
anticoagulation and mortality.
Table 6
Odds ratio analysis for the effect of anticoagulation treatment on vessel recanalization
and anticoagulation treatment on mortality in patients with Lemierre syndrome
|
Vessel recanalization
|
|
|
Odds ratio
|
1.6
|
|
95% confidence interval
|
0.3 to 9.4
|
|
Significance level
|
p = 0.6
|
|
Number needed to treat
|
9
|
|
Mortality
|
|
|
Odds ratio
|
0.6
|
|
95% confidence interval
|
0.1 to 2.9
|
|
Significance level
|
p = 0.5
|
|
Number needed to treat
|
55
|
Discussion
The present study analyzed the diagnosis and management in 394 patients with a diagnosis
of Lemierre syndrome reported between 1980 and 2017. A meta-analysis was conducted
on 194 of these patients to examine the effect of anticoagulation on mortality, and
on 50 patients to examine the effect of anticoagulation on vessel recanalization on
follow-up imaging. The OR for anticoagulation and mortality was 0.6, while the OR
for anticoagulation and vessel recanalization was 1.6. Neither relationship was statistically
significant, showing no effect on mortality or vessel recanalization with anticoagulation
in these subsets of Lemierre syndrome patients. Cupit-Link et al[5] retrospectively examined the charts of 18 pediatric and adult patients with Lemierre
syndrome. In total, 6 of the patients were anticoagulated for at least 4 weeks (range
6.9–32.9 weeks). They noted that all patients had improvement in their thrombi by
three months, with complete resolution in nine of the nonanticoagulated patients and
partial resolution in three of the nonanticoagulated patients, and complete thrombus
resolution in two of the anticoagulated patients, and partial resolution in four of
the anticoagulated patients. Their data suggested that anticoagulation in Lemierre
syndrome patients did not appear to affect the rate of thrombus resolution. Rebelo
et al[6] retrospectively examined 11 pediatric patients with Lemierre syndrome and thrombosis
(sigmoid sinus: 6; internal jugular vein: 5). All 11 patients were anticoagulated
with low molecular weight heparin (LMWH) for an average duration of 105.8 days. In
total, 10 of the 11 patients had partial or complete resolution of the thrombus within
a median of 3.4 months. No control non-anticoagulated group was analyzed, but they
postulated that the benefit of anticoagulation might outweigh the low risks given
their observation of no adverse effects from anticoagulation in their retrospective
patient series. Their review of the literature revealed anticoagulant use in 63.7%
of pediatric cases reported since 2002, similar to our observed rate of 63.4% of patients
anticoagulated overall since 1980. Interestingly, they noted one middle cerebral artery
(MCA) infarct, four patients with septic pulmonary emboli, three cranial nerve palsies,
and two patients with Horner syndrome. Phan et al[8] reported a single pediatric Lemierre case treated with enoxaparin with persistent
internal jugular (IJ) vein thrombosis on three-week follow-up imaging, but proposed
that anticoagulation should be considered in high-risk patients in the absence of
any contraindication to anticoagulation. In a case treated by the author of the present
study noted in [Fig. 1], the Infectious Disease Service recommended against anticoagulation due to the risk
of intracranial hemorrhage given the patient's thrombocytopenia and sigmoid sinus
involvement, while Vascular Surgery recommended anticoagulation. Ultimately no anticoagulation
was given, and the patient had an uneventful clinical course with aggressive IV antibiotic
treatment, while an MRI at one year (shown in [Fig. 1]) showed persistence of the left internal jugular vein thrombosis. The patient had
no long-term sequelae of the disease.
A search of the literature did not reveal any studies that systematically examined
the relationship between mortality and anticoagulation, which may be due to the low
overall mortality rate (4.1% in the present study) in the postantibiotic era. Interestingly,
the use of ligation/excision of the thrombosed internal jugular vein was low, with
10 patients in this series treated with IJ vein ligation/excision reported between
1980 and 2017. No studies could be identified systematically examining the effect
of vessel ligation/excision on mortality, but excision of the thrombosed vessel appeared
to be more likely in patients who continued to decline clinically despite aggressive
medical therapy.[9]
There are several limitations to the present study. The retrospective nature of the
meta-analysis makes recall bias and selection bias more likely. The varied geographical
locations and causative organisms, as well as the variation in the type and length
of therapy of both the antibiotic and, when used, anticoagulation regimens make it
more difficult to draw conclusions from the meta-analysis data. Additionally, there
was variability in the timing of follow-up imaging, ranging from several days to one
year, introducing the possibility that over time there might be a greater rate of
vessel recanalization with a longer follow-up time regardless of anticoagulation.
The majority of patients in both the anticoagulated and non-anticoagulated groups
showed vessel recanalization. The modality of follow-up imaging also varied depending
on the study, with some patients followed up using ultrasound, while CT or MRI was
used in other patients. Nevertheless, the present study represents the largest retrospective
series of Lemierre syndrome patients analyzed, and is the first study to our knowledge
to formally compare the rates of vessel recanalization and mortality in anticoagulated
and non-anticoagulated patients. Neither group showed a statistically significant
advantage for anticoagulation. Given the relatively low incidence of Lemierre syndrome,
and the difficulty of assembling prospective, randomized patient groups, and the findings
of this meta-analysis, it may be reasonable to treat patients primarily with aggressive
empiric and then culture directed antibiotic therapy to cover the most common organisms
(Fusobacterium, Staphylococcus, and streptococcal species), with surgical treatment as indicated for coalescent
mastoiditis, empyema, severe sinusitis, abscesses etc., reserving anticoagulation
for patients with low risk of anticoagulation side effects or who have refractory
sequelae of vessel thrombosis despite aggressive antibiotic +/− surgical treatment.
A more routine follow-up imaging protocol may be useful in analyzing vessel recanalization,
but this may be difficult as many Lemierre patients may be lost to follow-up.
Final Comments
The present meta-analysis examined 394 patients presented in the literature between
1980 and 2017 with a diagnosis of Lemierre syndrome. A meta-analysis of the effect
of anticoagulation on vessel recanalization and mortality failed to show a statistically
significant benefit for either outcome. Aggressive IV antibiotic therapy combined
with surgical intervention when indicated appeared to be the mainstay of treatment
in Lemierre syndrome, with a low mortality rate overall with or without anticoagulation.
