British Journal of Neurosurgery, 2014; Early Online: 1–6 © 2014 The Neurosurgical Foundation ISSN: 0268-8697 print / ISSN 1360-046X online DOI: 10.3109/02688697.2014.931349
ORIGINAL ARTICLE
The difference in diffusion-weighted imaging with apparent diffusion coefficient between spontaneous and postoperative intracranial infection Yeong-Jin Kim1, Kyung-Sub Moon1, Seul Kee Kim2, Seong-Ji Kang3, Kyung-Hwa Lee4, Woo-Yool Jang1, Tae-Young Jung1, In-Young Kim1 & Shin Jung1
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
1Department of Neurosurgery, Chonnam National University Research Institute of Medical Sciences, Chonnam National
University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea, 2Department of Radiology, Chonnam National University Research Institute of Medical Sciences, Chonnam National University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea, 3Department of Infectious Medicine, Chonnam National University Research Institute of Medical Sciences, Chonnam National University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea, and 4Department of Pathology, Chonnam National University Research Institute of Medical Sciences, Chonnam National University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea
Introduction
Abstract Background. Although the roles of diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) have been accepted as the initial or confirmatory diagnostic tool for spontaneous intracranial infections, the usefulness of these has rarely been investigated in intracranial infections after a craniotomy procedure. Through an analysis of the clinico-radiological characteristics of spontaneous and postoperative intracranial infections, the authors revealed the specific factors that affect the accuracy of DWI and ADC in diagnosing intracranial infections. Methods. The authors retrospectively analyzed 67 intracranial infections confirmed using preoperative MR imaging, including the DWI, ADC and gadolium-enhanced (Gd) images, and by peroperative pus drainage. Results. In 67 enrolled patients, no or uncertain diffusion restriction on DWI and ADC was found in 9 cases (13%). All the cases showed typical peripheral enhancement on Gd images. Among nine cases without diffusion restriction, postoperative infection was seen in five cases (62.5% [5/8 postoperative infection group] vs. 6.8% [4/59 spontaneous infection group], p ⫽ 0.001). On multivariate analysis, postoperative infection was the predictive factor for false-negative restriction on DWI and ADC (hazard ratio: 41.2, 95% confidential index: 2.39–710.25, p ⫽ 0.01). Conclusion. Despite the excellent availability of DWI and ADC for diagnosing spontaneous intracranial infections, negative restriction results of those images are not sufficient to exclude postoperative intracranial infection.
Intracranial infections can be a fatal condition, leading to marked morbidity and mortality.1 Hence early diagnosis of intracranial infections is important. As advances imaging techniques including brain computed tomography (CT) scan and magnetic resonance (MR) imaging are made, the diagnosis can be made faster and more accurately. On conventional MR imaging, the typical brain abscess has low-signal intensity on T1-weighted images (T1WI), iso-/ high-signal intensity on T2-weighted images (T2WI), and ring enhancement by contrast agents.2–5 Furthermore, diffusionweighted imaging (DWI) and apparent diffusion coefficient (ADC), which are based on the random movement of water molecules known as Brownian motion or translational motion, play an important role in the discrimination of brain abscesses from necrotic or cystic tumors. In DWI, utilization of strong gradient pulses applied to conventional spin echo sequences makes freely moving water molecules lose signal intensity compared to stationary water molecules which have a high-signal intensity. The quantification of this degree of water motion is known as the ADC. The more the water molecules are restricted, the lower the ADC.6–8 Although many studies demonstrated a role for DWI in the diagnosis of subdural empyema, ventriculitis and the differentiation of ring-enhancing lesions,7–10 few studies have validated the sensitivity of DWI in identifying postoperative intracranial infections. Among patients who undergo craniotomy for various kinds of intracranial pathologies, 0.43% develop extradural abscess and 0.11% develop subdural empyema. Only a minor number of patients experience fever
Keywords: apparent diffusion coefficient; brain abscess; diagnosis; diffusion-weighted image; postoperative infection
Correspondence: Kyung-Sub Moon, M.D., Ph.D., Associate Professor, Department of Neurosurgery, Brain Tumor Clinic & Gamma Knife Center, Chonnam National University Hwasun Hospital and Medical School, 322 Seoyang-ro, Hwasun-eup, Hwasun-gun, Jeollanam-do 519–763, Republic of Korea. Tel: ⫹ 82-61-379-7666. Fax: ⫹ 82-61-379-7673. E-mail: [email protected] Received for publication 17 January 2014; accepted 30 May 2014
1
2
Y.-J. Kim et al.
and headache, with or without mild elevated inflammatory markers on the laboratory test.11–13 Postoperative intracranial infections, however, can lead to severe neurological deficits with high mortality rates ranging from 10% to 17%.14,15 Thus, postoperative intracranial infection is a major postoperative complication and detection using imaging modalities is important. Despite a published paper reporting limitations of DWI in the diagnosis of postoperative infections,16 we reaffirm the overall accuracy of DWI and ADC in our cases with postoperative infections and outline the specific factors that affect their usefulness.
Materials and methods Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
Patient population and clinical data The study is in compliance with the Declaration of Helsinki (Sixth Revision, 2008). This study fulfills all the requirements for patient anonymity and was approved by the local institutional review board. Intracranial infection was classified as “spontaneous” and “postoperative” depending on whether the infection occurred after a primary neurosurgical procedure or not. Patients who already had a previous intracranial abscess treated with surgery or drugs were excluded. Based on the operation database of our institution from January 2003 to February 2013, we found 165 patients with spontaneous or postoperative intracranial infection. Among these patients, intracranial infection was confirmed in only 67 patients using preoperative MR imaging, including DWI, ADC, and gadolium-enhanced (Gd) image as a routine MR protocol for infection, and by preoperative finding of pus collection, and these patients were enrolled in this study. Patients whose diagnosis was not confirmed through the preoperative routine MR protocol or peroperative finding of pus drainage were excluded. We gathered the following clinical data: patient age and gender, location and etiology of the infection, microbiology spectrum (gram-positive, gram-negative, polymicrobial, and culure-negative), erythrocyte sedimentation rate (ESR, Nr: ⬍ 20 mm/hr), C-reactive protein level (CRP, Nr: ⬍ 0.3 mg/dL), white blood cell count (WBC, Nr: 4.0– 10.8 ⫻ 103/mm3), neurological or infectious symptoms, and underlying disease.
MR imaging MR imaging was performed using the 3-T Tim Trio MR scanner (Siemens, Erlangen, Germany), which is equipped with a 8-channel phased array head coil, and 1.5-T MR scanners (GE Signa Excite, and GE Signa HDxt; GE Healthcare, Waukesha, WI, USA), which is equipped with a 8-channel high-resolution brain coil and receive head coil. DWI was performed using a single shot, spin-echo planar imaging sequence in the axial plane, with following parameters: TR/ TE, 9000/80 msec; section thickness, 5 mm; matrix, 160 ⫻ 160; field of view, 16 ⫻ 16 cm; b values, 0 and 1000 sec/mm2. The ADC map was obtained from a DWI using a two-point linear regression method at b values of 0 and 1000 s/mm2 and was reconstructed on a pixel-by-pixel basis using the standard software on the console (Siemens and GE). The routine MR
imaging protocol included sagittal and axial T1WI, axial T2WI, axial fluid-attenuated inversion recovery images, and contrast-enhanced axial and coronal fat-suppressed two-dimensional T1WI and contrast-enhanced sagittal T1-weighted three-dimensional images. Two neuroradiologists reviewed the MR images independently and retrospectively. The signal intensities of the lesions on DWI were defined by visual inspection as low- and highsignal intensity compared with contralateral normal brain parenchyma. Measurements of ADC were made in regions of interest (ROI) in the lesion and ROI in contralateral normal brain. The lesion of intracranial infection was classified by the location seen on conventional MR images into intraparenchymal and extraparenchymal (subtype of extradural and subdural), and determined whether restricted diffusion on DWI and ADC map existed or not in the area of the lesion; positive diffusion restriction is high-signal intensity on DWI with relatively lower ADC value of the lesion compared with the ADC value of normal contralateral brain parenchyma, negative diffusion restriction is high-signal intensity on DWI with relatively high ADC value or no high-signal intensity on DWI.
Statistical analysis Statistical analyses were performed with the SSPS version 20.0 (SPSS Inc., Chicago, IL, USA) to determine the significance of the differences in the clinical features and radiological findings between the DWI positive group and DWI falsenegative groups. The relation between presence of diffusion restriction and categorical variables was compared using a Chi-square test, or Fisher‘s exact probability test, when appropriate. Multivariate binary logistic regression analysis was performed to identify the association between the variables. The level of significance was set at p ⬍ 0.05.
Results Clinical findings Of the subjects, 8 patients were classified into the postoperative group and 59 patients into the spontaneous group. The postoperative group consisted of six males (75%) and two females, with a median age of 48 years (range, 28–68 yrs). The spontaneous group showed similar male predominance (40/59, 68%) and median age (46 yrs, 0–80 yrs). In the spontaneous group, diabetes was the most frequent predisposing condition (14/59, 24%) in the medical history, followed by lung disease (13/59, 22%), heart disease (5/59, 9%), and liver disease (2/59, 3%). Only one patient (1/8, 13%) showed diabetes as a predisposing medical condition in the postoperative group. All the patients of the postoperative group complained of neurological symptoms or signs, including seizure, motor/sensory deficits, deteriorated mental status, and cranial nerve deficit. None of the patients demonstrated symptoms or signs of infection, such as fever, chills, or neck stiffness. In contrast, nine patients in the spontaneous group (15%) were admitted with symptoms or signs of infection. Based on the findings of the MR imaging and operation field, the exact location of the infection was determined. The infections in the postoperative group were located in the intraparenchyme (4/8, 50%), extradural space (3/8, 37%)),
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
DWI in postoperative intracranial infection 3 and subdural surface (1/8, 13%). Comparatively, most of the spontaneous infections were located within the parenchyma (55/59, 93%). Causative bacteria were determined using microbiological examination of the gross purulent pus collected in the operation field. Gram-positive bacteria was the prevailing organism in spontaneous infections (28/59, 47%). Culture-negative infection, in which the causative organism was not cultured on pus, was found in 20 cases (32%). Similarly, gram-positive (3/8, 38%) and culture-negative (4/8, 50%) infections were a common bacterial classification in the postoperative group. In 8 patients of the postoperative group, the preceding procedure included craniotomy, and this was done for brain tumor in 5 cases, vascular disease in 2 cases, and trauma in 1 case. Most of the spontaneous infections occurred only in intracranial lesions, and a minority had infections with another primary infection origin (17/59, 29%), such as dental infection, septic embolism, sinusitis and so on.
MR imaging findings In 67 enrolled patients, the usual restrictions were observed in 58 patients (positive restriction group, 87%) on DWI and ADC. Nine patients showed no or uncertain diffusion restriction (negative restriction group) despite having definite preoperative pus drainage (Table I; Fig. 1). However, regardless of the restriction on DWI, the majority of patients in both groups revealed typical peripheral enhancement of the lesion on Gd image. Only six cases in the positive restriction group had lesions located in non-intracerebral spaces (10% vs. 33% in negative restriction group, p ⫽ 0.09). Among nine cases in the negative restriction group, postoperative infection was observed in five cases (55.6% vs. 5.2% in positive restriction group, p ⫽ 0.001). That is, the sensitivity of DWI was 37.5% in postoperative cases (3/8), compared to spontaneous cases with the sensitivity of 93.2% (55/59; p ⫽ 0.001). On multivariate analysis, postoperative infection was the predictive factor for false-negative restriction on DWI and ADC (hazard ratio
[HR] 41.2, 95% confidential index [CI] 2.39–710.25, p ⫽ 0.01). It was impossible to determine the causative organism in five cases of the negative restriction group (56%), which was a higher rate than that of the positive restriction group (31%), but this difference did not reach statistical significance (HR: 5.5, 95% CI: 0.73–40.92, p ⫽ 0.10). Elevated CRP and leukocytosis, as potent predictors for systemic infection, were found less frequently in the negative restriction group than in the positive restriction group. Interestingly, elevated ESR was more frequently observed in the negative restriction group. However, there was no statistical significance for those infection markers between both groups. Table II summarizes the clinical and radiological characteristics of the patients in the postoperative group. The mean interval from the initial surgery to reoperation for infection was 53 days (range, 17–111 days). To investigate the relationship between infection duration and false negatives in DWI, we divided the postoperative group into a short duration group and a long duration group depending on whether the interval from initial surgery to reoperation for infection was less than 2 months or not. In our study, there was no difference in this interval between two groups (60% in short duration vs. 67% in long duration). There was also no difference in the mean interval from the initial surgery to reoperation for infection between the positive and negative restriction cases (24 days in the positive restriction group vs. 60 days in the negative restriction group). The mean operation time for the initial surgery was 339 min (range, 135–645 min), and there was some difference in this time between the positive and negative restriction cases (248 min in positive restriction vs. 406 min in negative restriction). An artificial dural substitute was used in three cases and one of these cases showed negative restriction. Metal fixation material was used in all cases except for one case that had performed craniectomy for traumatic brain injury. First generation cephalosporin antibiotics were routinely used in the perioperative period.
Table I. Clinical characteristics between negative restriction group and positive restriction group on DWI and ADC. Mutivariate Negative restriction Positive restriction Univariate p-value Variables group (N = 9) group (N = 58) HR 95% CI p-value Age ⬍ 60 years ⱖ 60 years Sex M F Underlying disease# Post-operative Infection Location Extraparenchymal* Intraparenchymal Microbiology spectrum Culture negative! Culture positive Infectious symptom CRP (⬎ 0.3 mg/dL) WBC (⬎ 10.8 103/mm3) ESR (⬎ 20 mm/hr)
7 (77.8%) 2 (22.2%)
41 (70.7%) 17 (29.3%)
1.00
1.00 0.40
6 (66.7%) 3 (33.3%) 6 (66.7%) 5 (55.6%)
40 (69.0%) 18 (31.0%) 45 (77.6%) 3 (5.2%)
1.00
1.00 1.40 2.00 41.20
6 (66.7%) 3 (33.3%)
6 (10.3%) 52 (89.7%)
0.09
5 (55.6%) 4 (44.4%) 1 (11.1%) 4 (44.4%) 1 (11.1%) 7 (77.8%)
18 (31.0%) 40 (69.0%) 9 (15.5%) 38 (65.5%) 26 (44.8%) 20 (34.5%)
0.26
0.44 0.00
1.00 0.23 0.14 0.26
0.04–4.30
0.45
0.21–9.80
0.71
0.17–23.36 2.39–710.25
0.58 0.01
1.00 0.40
0.04–4.01
0.42
5.50 1.00
0.73–40.92
0.10
Not available Not available Not available Not available
ADC: apparent diffusion coefficient, CI: confidential index, CRP: C-reactive protein level, DWI: diffusion-weighted image, ESR: erythrocyte sedimentation rate, HR: hazard ratio, WBC: white blood cell count #Including diabetes mellitus, hypertension, cardiopulmonary disease. *Including epidural abscess or subdural abscess. !No microscopically stained bacteria at gram stain and no macroscopically grown organism at culture media during 5 days.
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
4
Y.-J. Kim et al.
Fig. 1. MR images including Gd image, DWI, and ADC show true positives in the spontaneous infection group (A), true positives in the postoperative infection group (B), partial false negatives in the postoperative infection group (C, D), and false negatives in the postoperative infection group (E). (A) A typical image of a spontaneously developing intracranial abscess in the right parietal lobe. A hypointense central cavity with an enhanced peripheral capsule and hypointense perilesional edema are detected by the Gd image. DWI showed high-signal intensity of restricted diffusion and ADC showed low-signal intensity. Stereotactic biopsy proved the presence of purulent material and the patient recovered with 6 weeks’ treatment of antibiotics. (B) After gross total removal of a parasagittal meningioma, a peripherally ring-enhanced lesion appeared. Along with subacute hemorrhage, a diffusion restricted lesion appeared. (C) After gross total removal of a tentorial meningioma, a post-resection cavity with marginal enhancement appeared. In DWI, only the deepest small-sized portion had high-signal intensity with low-signal intensity on the ADC map. Gross pus is aspirated by free-hand aspiration using a syringe and the patient had a revision operation. (D) The same as C; the most dependent portion had restricted diffusion. (E) After removal of the convexity meningioma, pus discharge from the epidural abscess leaked through the operation suture site. But there is only marginal enhancement, with no diffusion restriction.
Discussion Many articles have established the value of DWI and ADC in the diagnosis of cerebral abscess, and their critical role in discriminating of brain abscesses from necrotic or cystic tumors has also been documented.7,8,17 In 1996, Ebisu et al.8 published a paper reporting that an abscess cavity shows the characteristic findings of high-signal intensity on DWI and low ADC values, which correlate with the ADC of the aspirated purulent material. The biological mechanism underlying the restricted diffusion has not been extensively studied,
but this probably relates to the contents of the abscess cavity which consists of inflammatory cells, necrotic tissues, and protein that make a thick, mucoid, acidic highly concentrated liquid.9 Other studies have reconfirmed the usefulness of DWI in the diagnosis of intracranial abscess.9,18,19 In this study, all spontaneous infections showed high-signal intensity in DWI and low ADC values, ascertaining the role of DWI and ADC in the diagnosis of cerebral abscess without using a preceding procedure. Unlike spontaneous intracranial abscesses, the use of DWI and ADC in postoperative intracranial abscesses is
DWI in postoperative intracranial infection 5 Table II. Clinico-radiological characteristics of postoperative intracranial infection group.
Case no 1 2 3 4 5 6 7 8
Restriction Interval from MRI Age (years)/ on DWI/ scan to re-operation, Sex ADC (days) 68/M 56/F 41/M 62/M 33/M 58/M 36/F 28/M
⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹
Interval from first operation to re-operation, (days)
2 3 2 2 2 1 0 5
32 118 45 57 49 45 17 11
Pathology in initial operation Vascular Tumor Tumor Tumor Trauma Vascular Tumor Tumor
Use of dural substitue
Operation time of initial surgery (minutes)
Previous Operation bed size (Cm)
Fixation material
⫺ ⫹ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹
295 420 265 645 Can’t check 135 275 335
4.5 3.9 4.6 5.4 Can’t check 1.8 4.6 6.0
⫹ ⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫹
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
ADC: apparent diffusion coefficient, DWI: diffusion-weighted image
not well validated. There are only a few studies that investigate the usefulness of DWI in the differential diagnosis of postoperative intracranial infection from other conditions such as remnant brain tumor or postoperative fluid collection.16,20 Hartmann et al.20 reported a case in which the brain abscess after a neurosurgical procedure showed unrestricted diffusion within the independent portion occupying most of the resection cavity, despite the presence of a small gravity-dependent portion showing restricted diffusion. Farrell et al.16 insisted that the absence of high-signal intensity in DWI is not satisfactory for precluding the presence of postoperative intracranial infection and should not be used as the key diagnostic modality in this situation. Nevertheless, MR imaging with DWI is widely used in the diagnosis of postoperative intracranial infection, especially after brain tumor surgery. Similar to this previous report, the radiologic prediction using DWI and ADC for postoperative intracranial infection was so unreliable in our study that the use of the methods in diagnosing postoperative infection is doubtful. Compared to spontaneous brain abscesses which have clear boundaries to adjacent brain tissue, postoperative abscesses develop in a post-neurosurgical resection cavity. This makes the abscess less restricted, and less bounded by brain tissue and also allows purulent material to move around to different areas of the resection cavity depending on the patient’s position. For detection using the DWI and ADC map, the sticky purulent material must stay steady in one place. The development of a sufficient, homogenous infectious collection is disturbed. The high false-negative rate of epidural abscesses can be explained by this phenomenon, because of the wide range of viable spaces for the abscess. Noticeably, we were able to see restricted diffusion and low ADC values in the dependent portion of the abscess. Because purulent material accumulates in gravity-dependent areas, a partial small portion on the bottom of the resection cavity will contain most of the purulent material. Usually, this is overlooked. As well as pus material, the postoperative fluid collection and hemorrhage can converge in the same part. This makes it difficult to interpret MR image. In some cases, we found partial restriction in a dependent portion in the false-negative group. Although the neuroradiologist may use other imaging techniques like MPGR (multiplanar gradientrecalled), the amount of material collected is too small for discrimination with these methods.
Aspirated pus in a culture-negative abscess had less restricted DWI than that of culture-positive brain abscesses. Like the previous study,21 this shows that proliferous bacterial content correlated with restricted diffusion. In our study, culture negative abscess was also frequently found in the negative restriction group, but without reaching statistical significance. Although the mechanism responsible for this relationship has not been firmly established, it may be the result of production of pro-inflammatory cytokines and chemokines that increase inflammatory cells in the pus. A previous study showed that the time interval from the initial operation to revision for the intracranial abscess may affect the imaging findings.16 That is, the increased heterogeneity that develops during infection in the postoperative abscess cavity may affect MR imaging. The finding that intracranial infections developing within 2 weeks after the initial operation demonstrated a high rate of diffusion restriction compared to infections that had delayed onset (over 2 months after initial operation), which implies that a more aggressive and rapidly setting infection shows a typical diffusion restriction. In our study, however, there was no correlation between infection duration and false negativity in diffusion restriction, possibly due to the low number of samples in the postoperative group. It is known that restricted diffusion is the most common abnormality observed in the early postoperative DWI of brain parenchyma at the operation site after neurosurgery, which suggests that tissue damage including cytotoxic edema is caused by the surgical procedure. In addition, hemorrhage in the operation bed can constitute another cause of highsignal intensity in DWI.22 Because of this, in the early postoperative imaging, it is important to be aware of the possible misinterpretation of restricted diffusion as an intracranial abscess when it is actually due to tissue damage associated with a fluid collection near the operation site. The brain abscesses showed with no remarkable change of inflammatory indices such as WBC count, CRP, ESR, and body temperature. Absence of change in these markers was more definite in patients without meningitis or ventriculitis.13 This clinical finding makes it difficult to analyze imaging findings accurately, even though the presence of scalp erythematic swelling in postoperative infection. In our study, there was no difference in the abnormality of these markers regardless of whether there was diffusion restriction or not.
6
Y.-J. Kim et al.
Our study has several shortcomings. First, it consists of a small number of patients. Second, selection bias may exist. Not all patients with an intracranial abscess were examined with DWI. After tumor surgery, DWI was more frequently performed. Third, the study design was retrospective in nature. Further studies need to be performed prospectively with large numbers of subjects for validating the usefulness of DWI in the postoperative setting.
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
Conclusion Although DWI with an ADC map is an excellent tool for diagnosis spontaneous brain abscesses, it is not sufficient for excluding intracranial infection after a neurosurgical operation, when it is based only on the negative restriction of those images.
Declaration of interest: The authors report no declarations of interest. The authors alone are responsible for the content and writing of the paper.
References 1. Yang SY, Zhao CS. Review of 140 patients with brain abscess. Surg Neurol 1993;39:290–6. 2. Haimes AB, Zimmerman RD, Morgello S, et al. MR imaging of brain abscesses. AJR Am J Roentgenol 1989;152:1073–85. 3. Kjos BO, Brant-Zawadzki M, Kucharczyk W, et al. Cystic intracranial lesions: magnetic resonance imaging. Radiology 1985;155:363–9. 4. Holtas S, Tornquist C, Cronqvist S. Diagnostic difficulties in computed tomography of brain abscesses. J Comput Assist Tomogr 1982;6:683–8. 5. Braun IF, Chambers E, Leeds NE, Zimmerman RD. The value of unenhanced scans in differentiating lesions producing ring enhancement. AJNR Am J Neuroradiol 1982;3:643–7. 6. Castillo M, Mukherji SK . Diffusion-weighted imaging in the evaluation of intracranial lesions. Semin Ultrasound CT MR 2000;21:405–16.
7. Leuthardt EC, Wippold FJ II, Oswood MC, Rich KM. Diffusionweighted MR imaging in the preoperative assessment of brain abscesses. Surg Neurol 2002;58:395–402. 8. Ebisu T, Tanaka C, Umeda M, et al. Discrimination of brain abscess from necrotic or cystic tumors by diffusion-weighted echo planar imaging. Magn Reson Imaging 1996;14:1113–6. 9. Bukte Y, Paksoy Y, Genc E, Uca AU. Role of diffusion-weighted MR in differential diagnosis of intracranial cystic lesions. Clin Radiol 2005;60:375–83. 10. Nadal Desbarats L, Herlidou S, de Marco G, et al. Differential MRI diagnosis between brain abscesses and necrotic or cystic brain tumors using the apparent diffusion coefficient and normalized diffusion-weighted images. Magn Reson Imaging 2003;21:645–50. 11. Hlavin ML, Kaminski HJ, Fenstermaker RA , White RJ. Intracranial suppuration: a modern decade of postoperative subdural empyema and epidural abscess. Neurosurgery 1994;34:974–80. 12. Mathisen GE, Johnson JP. Brain abscess. Clin Infect Dis 1997;25: 763–79. 13. Oyama H, Kito A , Maki H,et al. Inflammatory index and treatment of brain abscess. Nagoya J Med Sci 2012;74:313–24. 14. Helweg-Larsen J, Astradsson A , Richhall H, et al. Pyogenic brain abscess, a 15 year survey. BMC Infect Dis 2012;12:332. 15. Landriel F, Ajler P, Hem S, et al. Supratentorial and infratentorial brain abscesses: surgical treatment, complications and outcomes—a 10-year single-center study. Acta Neurochir 2012;154:903–11. 16. Farrell CJ, Hoh BL, Pisculli ML, et al. Limitations of diffusionweighted imaging in the diagnosis of postoperative infections. Neurosurgery 2008;62:577–83. 17. Chang SC, Lai PH, Chen WL, et al. Diffusion-weighted MRI features of brain abscess and cystic or necrotic brain tumors: comparison with conventional MRI. Clin Imaging 2002;26:227–36. 18. Desprechins B, Stadnik T, Koerts G, et al. Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. AJNR Am J Neuroradiol 1999;20:1252–7. 19. Kim YJ, Chang KH, Song IC, et al. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusionweighted MR imaging. AJR Am J Roentgenol 1998;171:1487–90. 20. Hartmann M, Jansen O, Heiland S, et al. Restricted diffusion within ring enhancement is not pathognomonic for brain abscess. AJNR Am J Neuroradiol 2001;22:1738–42. 21. Mishra AM, Gupta RK , Saksena S, et al. Biological correlates of diffusivity in brain abscess. Magn Reson Med 2005;54:878–85. 22. Ozturk A , Oguz KK, Akalan N, Geyik PO, Cila A . Evaluation of parenchymal changes at the operation site with early postoperative brain diffusion-weighted magnetic resonance imaging. Diagn Interv Radiol 2006;12:115–20.
ORIGINAL ARTICLE
The difference in diffusion-weighted imaging with apparent diffusion coefficient between spontaneous and postoperative intracranial infection Yeong-Jin Kim1, Kyung-Sub Moon1, Seul Kee Kim2, Seong-Ji Kang3, Kyung-Hwa Lee4, Woo-Yool Jang1, Tae-Young Jung1, In-Young Kim1 & Shin Jung1
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
1Department of Neurosurgery, Chonnam National University Research Institute of Medical Sciences, Chonnam National
University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea, 2Department of Radiology, Chonnam National University Research Institute of Medical Sciences, Chonnam National University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea, 3Department of Infectious Medicine, Chonnam National University Research Institute of Medical Sciences, Chonnam National University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea, and 4Department of Pathology, Chonnam National University Research Institute of Medical Sciences, Chonnam National University Hwasun Hospital & Medical School, Hwasun-gun, Jeollanamdo, Republic of Korea
Introduction
Abstract Background. Although the roles of diffusion-weighted imaging (DWI) and apparent diffusion coefficient (ADC) have been accepted as the initial or confirmatory diagnostic tool for spontaneous intracranial infections, the usefulness of these has rarely been investigated in intracranial infections after a craniotomy procedure. Through an analysis of the clinico-radiological characteristics of spontaneous and postoperative intracranial infections, the authors revealed the specific factors that affect the accuracy of DWI and ADC in diagnosing intracranial infections. Methods. The authors retrospectively analyzed 67 intracranial infections confirmed using preoperative MR imaging, including the DWI, ADC and gadolium-enhanced (Gd) images, and by peroperative pus drainage. Results. In 67 enrolled patients, no or uncertain diffusion restriction on DWI and ADC was found in 9 cases (13%). All the cases showed typical peripheral enhancement on Gd images. Among nine cases without diffusion restriction, postoperative infection was seen in five cases (62.5% [5/8 postoperative infection group] vs. 6.8% [4/59 spontaneous infection group], p ⫽ 0.001). On multivariate analysis, postoperative infection was the predictive factor for false-negative restriction on DWI and ADC (hazard ratio: 41.2, 95% confidential index: 2.39–710.25, p ⫽ 0.01). Conclusion. Despite the excellent availability of DWI and ADC for diagnosing spontaneous intracranial infections, negative restriction results of those images are not sufficient to exclude postoperative intracranial infection.
Intracranial infections can be a fatal condition, leading to marked morbidity and mortality.1 Hence early diagnosis of intracranial infections is important. As advances imaging techniques including brain computed tomography (CT) scan and magnetic resonance (MR) imaging are made, the diagnosis can be made faster and more accurately. On conventional MR imaging, the typical brain abscess has low-signal intensity on T1-weighted images (T1WI), iso-/ high-signal intensity on T2-weighted images (T2WI), and ring enhancement by contrast agents.2–5 Furthermore, diffusionweighted imaging (DWI) and apparent diffusion coefficient (ADC), which are based on the random movement of water molecules known as Brownian motion or translational motion, play an important role in the discrimination of brain abscesses from necrotic or cystic tumors. In DWI, utilization of strong gradient pulses applied to conventional spin echo sequences makes freely moving water molecules lose signal intensity compared to stationary water molecules which have a high-signal intensity. The quantification of this degree of water motion is known as the ADC. The more the water molecules are restricted, the lower the ADC.6–8 Although many studies demonstrated a role for DWI in the diagnosis of subdural empyema, ventriculitis and the differentiation of ring-enhancing lesions,7–10 few studies have validated the sensitivity of DWI in identifying postoperative intracranial infections. Among patients who undergo craniotomy for various kinds of intracranial pathologies, 0.43% develop extradural abscess and 0.11% develop subdural empyema. Only a minor number of patients experience fever
Keywords: apparent diffusion coefficient; brain abscess; diagnosis; diffusion-weighted image; postoperative infection
Correspondence: Kyung-Sub Moon, M.D., Ph.D., Associate Professor, Department of Neurosurgery, Brain Tumor Clinic & Gamma Knife Center, Chonnam National University Hwasun Hospital and Medical School, 322 Seoyang-ro, Hwasun-eup, Hwasun-gun, Jeollanam-do 519–763, Republic of Korea. Tel: ⫹ 82-61-379-7666. Fax: ⫹ 82-61-379-7673. E-mail: [email protected] Received for publication 17 January 2014; accepted 30 May 2014
1
2
Y.-J. Kim et al.
and headache, with or without mild elevated inflammatory markers on the laboratory test.11–13 Postoperative intracranial infections, however, can lead to severe neurological deficits with high mortality rates ranging from 10% to 17%.14,15 Thus, postoperative intracranial infection is a major postoperative complication and detection using imaging modalities is important. Despite a published paper reporting limitations of DWI in the diagnosis of postoperative infections,16 we reaffirm the overall accuracy of DWI and ADC in our cases with postoperative infections and outline the specific factors that affect their usefulness.
Materials and methods Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
Patient population and clinical data The study is in compliance with the Declaration of Helsinki (Sixth Revision, 2008). This study fulfills all the requirements for patient anonymity and was approved by the local institutional review board. Intracranial infection was classified as “spontaneous” and “postoperative” depending on whether the infection occurred after a primary neurosurgical procedure or not. Patients who already had a previous intracranial abscess treated with surgery or drugs were excluded. Based on the operation database of our institution from January 2003 to February 2013, we found 165 patients with spontaneous or postoperative intracranial infection. Among these patients, intracranial infection was confirmed in only 67 patients using preoperative MR imaging, including DWI, ADC, and gadolium-enhanced (Gd) image as a routine MR protocol for infection, and by preoperative finding of pus collection, and these patients were enrolled in this study. Patients whose diagnosis was not confirmed through the preoperative routine MR protocol or peroperative finding of pus drainage were excluded. We gathered the following clinical data: patient age and gender, location and etiology of the infection, microbiology spectrum (gram-positive, gram-negative, polymicrobial, and culure-negative), erythrocyte sedimentation rate (ESR, Nr: ⬍ 20 mm/hr), C-reactive protein level (CRP, Nr: ⬍ 0.3 mg/dL), white blood cell count (WBC, Nr: 4.0– 10.8 ⫻ 103/mm3), neurological or infectious symptoms, and underlying disease.
MR imaging MR imaging was performed using the 3-T Tim Trio MR scanner (Siemens, Erlangen, Germany), which is equipped with a 8-channel phased array head coil, and 1.5-T MR scanners (GE Signa Excite, and GE Signa HDxt; GE Healthcare, Waukesha, WI, USA), which is equipped with a 8-channel high-resolution brain coil and receive head coil. DWI was performed using a single shot, spin-echo planar imaging sequence in the axial plane, with following parameters: TR/ TE, 9000/80 msec; section thickness, 5 mm; matrix, 160 ⫻ 160; field of view, 16 ⫻ 16 cm; b values, 0 and 1000 sec/mm2. The ADC map was obtained from a DWI using a two-point linear regression method at b values of 0 and 1000 s/mm2 and was reconstructed on a pixel-by-pixel basis using the standard software on the console (Siemens and GE). The routine MR
imaging protocol included sagittal and axial T1WI, axial T2WI, axial fluid-attenuated inversion recovery images, and contrast-enhanced axial and coronal fat-suppressed two-dimensional T1WI and contrast-enhanced sagittal T1-weighted three-dimensional images. Two neuroradiologists reviewed the MR images independently and retrospectively. The signal intensities of the lesions on DWI were defined by visual inspection as low- and highsignal intensity compared with contralateral normal brain parenchyma. Measurements of ADC were made in regions of interest (ROI) in the lesion and ROI in contralateral normal brain. The lesion of intracranial infection was classified by the location seen on conventional MR images into intraparenchymal and extraparenchymal (subtype of extradural and subdural), and determined whether restricted diffusion on DWI and ADC map existed or not in the area of the lesion; positive diffusion restriction is high-signal intensity on DWI with relatively lower ADC value of the lesion compared with the ADC value of normal contralateral brain parenchyma, negative diffusion restriction is high-signal intensity on DWI with relatively high ADC value or no high-signal intensity on DWI.
Statistical analysis Statistical analyses were performed with the SSPS version 20.0 (SPSS Inc., Chicago, IL, USA) to determine the significance of the differences in the clinical features and radiological findings between the DWI positive group and DWI falsenegative groups. The relation between presence of diffusion restriction and categorical variables was compared using a Chi-square test, or Fisher‘s exact probability test, when appropriate. Multivariate binary logistic regression analysis was performed to identify the association between the variables. The level of significance was set at p ⬍ 0.05.
Results Clinical findings Of the subjects, 8 patients were classified into the postoperative group and 59 patients into the spontaneous group. The postoperative group consisted of six males (75%) and two females, with a median age of 48 years (range, 28–68 yrs). The spontaneous group showed similar male predominance (40/59, 68%) and median age (46 yrs, 0–80 yrs). In the spontaneous group, diabetes was the most frequent predisposing condition (14/59, 24%) in the medical history, followed by lung disease (13/59, 22%), heart disease (5/59, 9%), and liver disease (2/59, 3%). Only one patient (1/8, 13%) showed diabetes as a predisposing medical condition in the postoperative group. All the patients of the postoperative group complained of neurological symptoms or signs, including seizure, motor/sensory deficits, deteriorated mental status, and cranial nerve deficit. None of the patients demonstrated symptoms or signs of infection, such as fever, chills, or neck stiffness. In contrast, nine patients in the spontaneous group (15%) were admitted with symptoms or signs of infection. Based on the findings of the MR imaging and operation field, the exact location of the infection was determined. The infections in the postoperative group were located in the intraparenchyme (4/8, 50%), extradural space (3/8, 37%)),
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
DWI in postoperative intracranial infection 3 and subdural surface (1/8, 13%). Comparatively, most of the spontaneous infections were located within the parenchyma (55/59, 93%). Causative bacteria were determined using microbiological examination of the gross purulent pus collected in the operation field. Gram-positive bacteria was the prevailing organism in spontaneous infections (28/59, 47%). Culture-negative infection, in which the causative organism was not cultured on pus, was found in 20 cases (32%). Similarly, gram-positive (3/8, 38%) and culture-negative (4/8, 50%) infections were a common bacterial classification in the postoperative group. In 8 patients of the postoperative group, the preceding procedure included craniotomy, and this was done for brain tumor in 5 cases, vascular disease in 2 cases, and trauma in 1 case. Most of the spontaneous infections occurred only in intracranial lesions, and a minority had infections with another primary infection origin (17/59, 29%), such as dental infection, septic embolism, sinusitis and so on.
MR imaging findings In 67 enrolled patients, the usual restrictions were observed in 58 patients (positive restriction group, 87%) on DWI and ADC. Nine patients showed no or uncertain diffusion restriction (negative restriction group) despite having definite preoperative pus drainage (Table I; Fig. 1). However, regardless of the restriction on DWI, the majority of patients in both groups revealed typical peripheral enhancement of the lesion on Gd image. Only six cases in the positive restriction group had lesions located in non-intracerebral spaces (10% vs. 33% in negative restriction group, p ⫽ 0.09). Among nine cases in the negative restriction group, postoperative infection was observed in five cases (55.6% vs. 5.2% in positive restriction group, p ⫽ 0.001). That is, the sensitivity of DWI was 37.5% in postoperative cases (3/8), compared to spontaneous cases with the sensitivity of 93.2% (55/59; p ⫽ 0.001). On multivariate analysis, postoperative infection was the predictive factor for false-negative restriction on DWI and ADC (hazard ratio
[HR] 41.2, 95% confidential index [CI] 2.39–710.25, p ⫽ 0.01). It was impossible to determine the causative organism in five cases of the negative restriction group (56%), which was a higher rate than that of the positive restriction group (31%), but this difference did not reach statistical significance (HR: 5.5, 95% CI: 0.73–40.92, p ⫽ 0.10). Elevated CRP and leukocytosis, as potent predictors for systemic infection, were found less frequently in the negative restriction group than in the positive restriction group. Interestingly, elevated ESR was more frequently observed in the negative restriction group. However, there was no statistical significance for those infection markers between both groups. Table II summarizes the clinical and radiological characteristics of the patients in the postoperative group. The mean interval from the initial surgery to reoperation for infection was 53 days (range, 17–111 days). To investigate the relationship between infection duration and false negatives in DWI, we divided the postoperative group into a short duration group and a long duration group depending on whether the interval from initial surgery to reoperation for infection was less than 2 months or not. In our study, there was no difference in this interval between two groups (60% in short duration vs. 67% in long duration). There was also no difference in the mean interval from the initial surgery to reoperation for infection between the positive and negative restriction cases (24 days in the positive restriction group vs. 60 days in the negative restriction group). The mean operation time for the initial surgery was 339 min (range, 135–645 min), and there was some difference in this time between the positive and negative restriction cases (248 min in positive restriction vs. 406 min in negative restriction). An artificial dural substitute was used in three cases and one of these cases showed negative restriction. Metal fixation material was used in all cases except for one case that had performed craniectomy for traumatic brain injury. First generation cephalosporin antibiotics were routinely used in the perioperative period.
Table I. Clinical characteristics between negative restriction group and positive restriction group on DWI and ADC. Mutivariate Negative restriction Positive restriction Univariate p-value Variables group (N = 9) group (N = 58) HR 95% CI p-value Age ⬍ 60 years ⱖ 60 years Sex M F Underlying disease# Post-operative Infection Location Extraparenchymal* Intraparenchymal Microbiology spectrum Culture negative! Culture positive Infectious symptom CRP (⬎ 0.3 mg/dL) WBC (⬎ 10.8 103/mm3) ESR (⬎ 20 mm/hr)
7 (77.8%) 2 (22.2%)
41 (70.7%) 17 (29.3%)
1.00
1.00 0.40
6 (66.7%) 3 (33.3%) 6 (66.7%) 5 (55.6%)
40 (69.0%) 18 (31.0%) 45 (77.6%) 3 (5.2%)
1.00
1.00 1.40 2.00 41.20
6 (66.7%) 3 (33.3%)
6 (10.3%) 52 (89.7%)
0.09
5 (55.6%) 4 (44.4%) 1 (11.1%) 4 (44.4%) 1 (11.1%) 7 (77.8%)
18 (31.0%) 40 (69.0%) 9 (15.5%) 38 (65.5%) 26 (44.8%) 20 (34.5%)
0.26
0.44 0.00
1.00 0.23 0.14 0.26
0.04–4.30
0.45
0.21–9.80
0.71
0.17–23.36 2.39–710.25
0.58 0.01
1.00 0.40
0.04–4.01
0.42
5.50 1.00
0.73–40.92
0.10
Not available Not available Not available Not available
ADC: apparent diffusion coefficient, CI: confidential index, CRP: C-reactive protein level, DWI: diffusion-weighted image, ESR: erythrocyte sedimentation rate, HR: hazard ratio, WBC: white blood cell count #Including diabetes mellitus, hypertension, cardiopulmonary disease. *Including epidural abscess or subdural abscess. !No microscopically stained bacteria at gram stain and no macroscopically grown organism at culture media during 5 days.
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
4
Y.-J. Kim et al.
Fig. 1. MR images including Gd image, DWI, and ADC show true positives in the spontaneous infection group (A), true positives in the postoperative infection group (B), partial false negatives in the postoperative infection group (C, D), and false negatives in the postoperative infection group (E). (A) A typical image of a spontaneously developing intracranial abscess in the right parietal lobe. A hypointense central cavity with an enhanced peripheral capsule and hypointense perilesional edema are detected by the Gd image. DWI showed high-signal intensity of restricted diffusion and ADC showed low-signal intensity. Stereotactic biopsy proved the presence of purulent material and the patient recovered with 6 weeks’ treatment of antibiotics. (B) After gross total removal of a parasagittal meningioma, a peripherally ring-enhanced lesion appeared. Along with subacute hemorrhage, a diffusion restricted lesion appeared. (C) After gross total removal of a tentorial meningioma, a post-resection cavity with marginal enhancement appeared. In DWI, only the deepest small-sized portion had high-signal intensity with low-signal intensity on the ADC map. Gross pus is aspirated by free-hand aspiration using a syringe and the patient had a revision operation. (D) The same as C; the most dependent portion had restricted diffusion. (E) After removal of the convexity meningioma, pus discharge from the epidural abscess leaked through the operation suture site. But there is only marginal enhancement, with no diffusion restriction.
Discussion Many articles have established the value of DWI and ADC in the diagnosis of cerebral abscess, and their critical role in discriminating of brain abscesses from necrotic or cystic tumors has also been documented.7,8,17 In 1996, Ebisu et al.8 published a paper reporting that an abscess cavity shows the characteristic findings of high-signal intensity on DWI and low ADC values, which correlate with the ADC of the aspirated purulent material. The biological mechanism underlying the restricted diffusion has not been extensively studied,
but this probably relates to the contents of the abscess cavity which consists of inflammatory cells, necrotic tissues, and protein that make a thick, mucoid, acidic highly concentrated liquid.9 Other studies have reconfirmed the usefulness of DWI in the diagnosis of intracranial abscess.9,18,19 In this study, all spontaneous infections showed high-signal intensity in DWI and low ADC values, ascertaining the role of DWI and ADC in the diagnosis of cerebral abscess without using a preceding procedure. Unlike spontaneous intracranial abscesses, the use of DWI and ADC in postoperative intracranial abscesses is
DWI in postoperative intracranial infection 5 Table II. Clinico-radiological characteristics of postoperative intracranial infection group.
Case no 1 2 3 4 5 6 7 8
Restriction Interval from MRI Age (years)/ on DWI/ scan to re-operation, Sex ADC (days) 68/M 56/F 41/M 62/M 33/M 58/M 36/F 28/M
⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹
Interval from first operation to re-operation, (days)
2 3 2 2 2 1 0 5
32 118 45 57 49 45 17 11
Pathology in initial operation Vascular Tumor Tumor Tumor Trauma Vascular Tumor Tumor
Use of dural substitue
Operation time of initial surgery (minutes)
Previous Operation bed size (Cm)
Fixation material
⫺ ⫹ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹
295 420 265 645 Can’t check 135 275 335
4.5 3.9 4.6 5.4 Can’t check 1.8 4.6 6.0
⫹ ⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫹
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
ADC: apparent diffusion coefficient, DWI: diffusion-weighted image
not well validated. There are only a few studies that investigate the usefulness of DWI in the differential diagnosis of postoperative intracranial infection from other conditions such as remnant brain tumor or postoperative fluid collection.16,20 Hartmann et al.20 reported a case in which the brain abscess after a neurosurgical procedure showed unrestricted diffusion within the independent portion occupying most of the resection cavity, despite the presence of a small gravity-dependent portion showing restricted diffusion. Farrell et al.16 insisted that the absence of high-signal intensity in DWI is not satisfactory for precluding the presence of postoperative intracranial infection and should not be used as the key diagnostic modality in this situation. Nevertheless, MR imaging with DWI is widely used in the diagnosis of postoperative intracranial infection, especially after brain tumor surgery. Similar to this previous report, the radiologic prediction using DWI and ADC for postoperative intracranial infection was so unreliable in our study that the use of the methods in diagnosing postoperative infection is doubtful. Compared to spontaneous brain abscesses which have clear boundaries to adjacent brain tissue, postoperative abscesses develop in a post-neurosurgical resection cavity. This makes the abscess less restricted, and less bounded by brain tissue and also allows purulent material to move around to different areas of the resection cavity depending on the patient’s position. For detection using the DWI and ADC map, the sticky purulent material must stay steady in one place. The development of a sufficient, homogenous infectious collection is disturbed. The high false-negative rate of epidural abscesses can be explained by this phenomenon, because of the wide range of viable spaces for the abscess. Noticeably, we were able to see restricted diffusion and low ADC values in the dependent portion of the abscess. Because purulent material accumulates in gravity-dependent areas, a partial small portion on the bottom of the resection cavity will contain most of the purulent material. Usually, this is overlooked. As well as pus material, the postoperative fluid collection and hemorrhage can converge in the same part. This makes it difficult to interpret MR image. In some cases, we found partial restriction in a dependent portion in the false-negative group. Although the neuroradiologist may use other imaging techniques like MPGR (multiplanar gradientrecalled), the amount of material collected is too small for discrimination with these methods.
Aspirated pus in a culture-negative abscess had less restricted DWI than that of culture-positive brain abscesses. Like the previous study,21 this shows that proliferous bacterial content correlated with restricted diffusion. In our study, culture negative abscess was also frequently found in the negative restriction group, but without reaching statistical significance. Although the mechanism responsible for this relationship has not been firmly established, it may be the result of production of pro-inflammatory cytokines and chemokines that increase inflammatory cells in the pus. A previous study showed that the time interval from the initial operation to revision for the intracranial abscess may affect the imaging findings.16 That is, the increased heterogeneity that develops during infection in the postoperative abscess cavity may affect MR imaging. The finding that intracranial infections developing within 2 weeks after the initial operation demonstrated a high rate of diffusion restriction compared to infections that had delayed onset (over 2 months after initial operation), which implies that a more aggressive and rapidly setting infection shows a typical diffusion restriction. In our study, however, there was no correlation between infection duration and false negativity in diffusion restriction, possibly due to the low number of samples in the postoperative group. It is known that restricted diffusion is the most common abnormality observed in the early postoperative DWI of brain parenchyma at the operation site after neurosurgery, which suggests that tissue damage including cytotoxic edema is caused by the surgical procedure. In addition, hemorrhage in the operation bed can constitute another cause of highsignal intensity in DWI.22 Because of this, in the early postoperative imaging, it is important to be aware of the possible misinterpretation of restricted diffusion as an intracranial abscess when it is actually due to tissue damage associated with a fluid collection near the operation site. The brain abscesses showed with no remarkable change of inflammatory indices such as WBC count, CRP, ESR, and body temperature. Absence of change in these markers was more definite in patients without meningitis or ventriculitis.13 This clinical finding makes it difficult to analyze imaging findings accurately, even though the presence of scalp erythematic swelling in postoperative infection. In our study, there was no difference in the abnormality of these markers regardless of whether there was diffusion restriction or not.
6
Y.-J. Kim et al.
Our study has several shortcomings. First, it consists of a small number of patients. Second, selection bias may exist. Not all patients with an intracranial abscess were examined with DWI. After tumor surgery, DWI was more frequently performed. Third, the study design was retrospective in nature. Further studies need to be performed prospectively with large numbers of subjects for validating the usefulness of DWI in the postoperative setting.
Br J Neurosurg Downloaded from informahealthcare.com by Universitaet Zuerich on 07/09/14 For personal use only.
Conclusion Although DWI with an ADC map is an excellent tool for diagnosis spontaneous brain abscesses, it is not sufficient for excluding intracranial infection after a neurosurgical operation, when it is based only on the negative restriction of those images.
Declaration of interest: The authors report no declarations of interest. The authors alone are responsible for the content and writing of the paper.
References 1. Yang SY, Zhao CS. Review of 140 patients with brain abscess. Surg Neurol 1993;39:290–6. 2. Haimes AB, Zimmerman RD, Morgello S, et al. MR imaging of brain abscesses. AJR Am J Roentgenol 1989;152:1073–85. 3. Kjos BO, Brant-Zawadzki M, Kucharczyk W, et al. Cystic intracranial lesions: magnetic resonance imaging. Radiology 1985;155:363–9. 4. Holtas S, Tornquist C, Cronqvist S. Diagnostic difficulties in computed tomography of brain abscesses. J Comput Assist Tomogr 1982;6:683–8. 5. Braun IF, Chambers E, Leeds NE, Zimmerman RD. The value of unenhanced scans in differentiating lesions producing ring enhancement. AJNR Am J Neuroradiol 1982;3:643–7. 6. Castillo M, Mukherji SK . Diffusion-weighted imaging in the evaluation of intracranial lesions. Semin Ultrasound CT MR 2000;21:405–16.
7. Leuthardt EC, Wippold FJ II, Oswood MC, Rich KM. Diffusionweighted MR imaging in the preoperative assessment of brain abscesses. Surg Neurol 2002;58:395–402. 8. Ebisu T, Tanaka C, Umeda M, et al. Discrimination of brain abscess from necrotic or cystic tumors by diffusion-weighted echo planar imaging. Magn Reson Imaging 1996;14:1113–6. 9. Bukte Y, Paksoy Y, Genc E, Uca AU. Role of diffusion-weighted MR in differential diagnosis of intracranial cystic lesions. Clin Radiol 2005;60:375–83. 10. Nadal Desbarats L, Herlidou S, de Marco G, et al. Differential MRI diagnosis between brain abscesses and necrotic or cystic brain tumors using the apparent diffusion coefficient and normalized diffusion-weighted images. Magn Reson Imaging 2003;21:645–50. 11. Hlavin ML, Kaminski HJ, Fenstermaker RA , White RJ. Intracranial suppuration: a modern decade of postoperative subdural empyema and epidural abscess. Neurosurgery 1994;34:974–80. 12. Mathisen GE, Johnson JP. Brain abscess. Clin Infect Dis 1997;25: 763–79. 13. Oyama H, Kito A , Maki H,et al. Inflammatory index and treatment of brain abscess. Nagoya J Med Sci 2012;74:313–24. 14. Helweg-Larsen J, Astradsson A , Richhall H, et al. Pyogenic brain abscess, a 15 year survey. BMC Infect Dis 2012;12:332. 15. Landriel F, Ajler P, Hem S, et al. Supratentorial and infratentorial brain abscesses: surgical treatment, complications and outcomes—a 10-year single-center study. Acta Neurochir 2012;154:903–11. 16. Farrell CJ, Hoh BL, Pisculli ML, et al. Limitations of diffusionweighted imaging in the diagnosis of postoperative infections. Neurosurgery 2008;62:577–83. 17. Chang SC, Lai PH, Chen WL, et al. Diffusion-weighted MRI features of brain abscess and cystic or necrotic brain tumors: comparison with conventional MRI. Clin Imaging 2002;26:227–36. 18. Desprechins B, Stadnik T, Koerts G, et al. Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. AJNR Am J Neuroradiol 1999;20:1252–7. 19. Kim YJ, Chang KH, Song IC, et al. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusionweighted MR imaging. AJR Am J Roentgenol 1998;171:1487–90. 20. Hartmann M, Jansen O, Heiland S, et al. Restricted diffusion within ring enhancement is not pathognomonic for brain abscess. AJNR Am J Neuroradiol 2001;22:1738–42. 21. Mishra AM, Gupta RK , Saksena S, et al. Biological correlates of diffusivity in brain abscess. Magn Reson Med 2005;54:878–85. 22. Ozturk A , Oguz KK, Akalan N, Geyik PO, Cila A . Evaluation of parenchymal changes at the operation site with early postoperative brain diffusion-weighted magnetic resonance imaging. Diagn Interv Radiol 2006;12:115–20.
