Differential diagnosis of intracranial ring enhancing cystic mass lesions—Role of diffusion-weighted imaging (DWI) and diffusion-tensor imaging (DTI)

Clinical Neurology and Neurosurgery 112 (2010) 218–225

Contents lists available at ScienceDirect

Clinical Neurology and Neurosurgery journal homepage: www.elsevier.com/locate/clineuro

Differential diagnosis of intracranial ring enhancing cystic mass lesions—Role of diffusion-weighted imaging (DWI) and diffusion-tensor imaging (DTI) Werner Reiche a,∗ , Volker Schuchardt b , Thomas Hagen c , Kamil A. Il’yasov d , Peter Billmann a , Johannes Weber e a

Institut of Radiology, Ortenau Klinikum Lahr-Ettenheim, 77933 Lahr, Germany Clinic for Neurology, Ortenau Klinikum Lahr-Ettenheim, 77933 Lahr, Germany Radiology Centre, 86150 Augsburg, Germany d Medical Physics, Department of Diagnostic Radiology, University Hospital Freiburg, 79106 Freiburg, Germany e Department of Neuroradiology, University Hospital Freiburg, 79106 Freiburg, Germany b c

a r t i c l e

i n f o

Article history: Received 18 April 2009 Received in revised form 16 November 2009 Accepted 26 November 2009 Available online 6 January 2010 Keywords: Brain Ring enhancing lesion Abscess Brain tumour Magnetic resonance imaging (MRI) Diffusion-weighted imaging (DWI) Diffusion-tensor imaging (DTI) Fractional anisotropy (FA)

a b s t r a c t Objective: To evaluate the contribution of DWI and DTI to the differential diagnosis of cerebral ring enhancing lesions by describing DWI and ADC (apparent diffusion coefficient) findings and measuring the two DTI parameters mean diffusivity (MD) and fractional anisotropy (FA). Materials and methods: A total of 17 patients presenting with 26 rim enhancing cysts were investigated with DWI and DTI. Parameter maps of the DTI metrics MD and FA were calculated and quantified using regions of interest (ROIs). Results: Five patients suffered from abscesses with a total of 10 cysts, 9 from glioblastomas with 10 cysts, and 3 from metastases with 6 cysts. All abscess cavities showed hyperintense DWI signal intensity compared to normal appearing white matter (NAWM), low ADC, low MD and high FA. Eight out of 10 glioblastoma cysts and all 6 metastatic cysts revealed hypointensity on DWI, high ADC, high MD and low FA (p < 0.0001 compared to abscess group). DWI findings of 2/10 glioblastoma cysts overlapped with those of abscesses showing hyperintensity on DWI, low ADC and low MD and hence mimicked abscesses. FA of these 2 glioblastoma cysts was significantly lower than in abscess cavities (p = 0.032). Conclusion: The findings of reduced diffusion compared to NAWM and increased FA within a ring enhancing cyst strongly indicate a cerebral abscess. In contrast, the majority of neoplastic cysts revealed high diffusion and low FA. Reduced diffusion is also found in a very small number of tumour cysts, but in these low FA refers to a non-infectious origin and thus helps distinguishing from infectious abscess. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Intracranial ring enhancing lesions of different etiologies have a similar appearance on conventional MRI. Differential diagnosis of ring enhancing cystic mass lesions includes primary and secondary brain tumours, mainly glioblastomas, and metastases, non-neoplastic cysts, and brain abscesses. Even in the era of MR imaging it is still a challenge to differentiate infectious abscesses from other ring enhancing lesions, especially from cystic glioblastomas and metastases. Clinical findings are almost unspecific and do not lead to a final diagnosis. For therapy, it is most important to differentiate cerebral abscesses from tumour cysts in order to initi-

∗ Corresponding author at: Institut of Radiology, Ortenau Klinikum LahrEttenheim, Post Box 1727, 77931 Lahr, Germany. Tel.: +49 7821 932601; fax: +49 7821 932063. E-mail address: [email protected] (W. Reiche). 0303-8467/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.clineuro.2009.11.016

ate adequate medical and surgical treatments early. In recent years, diffusion-weighted MRI techniques were added to the diagnostic. Diffusion-weighted imaging (DWI) gains increasing importance in the diagnosis of cerebral tumours and cystic cerebral lesions. DWI is a kind of molecular imaging providing information about the mobility of water molecules for example in brain parenchyma, tumour tissue, pus, or cysts. High signal intensity of abscess pus on DWI scans associated with low apparent diffusion coefficient (ADC) was first reported by Ebisu et al. [1]. In contrast, cystic tumours showed low DWI signal intensity and high ADC [2]. Further reports [3–5] supported these findings and confirmed, that a discrimination of brain abscesses from cystic brain tumours is possible, as most abscesses showed restricted diffusion compared to non-affected brain. High DWI signal intensity is however not a specific finding for abscesses. Holtås et al. [6] and Hartmann et al. [7], each reported a ring enhancing metastasis of adenocarcinoma with high signal on DWI and low ADC mimicking an abscess. Tung et al. [8] presented two

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

metastases, both from squamous cell carcinomas, and one radiation necrosis with hyperintensity on DWI relative to normal white matter and low ADC. Further Reddy et al. [9] observed in 2 out of 50 glioblastomas high DWI signal. Diffusion-tensor imaging (DTI) represents a further development of DWI. In contrast to DWI, DTI measures diffusion ellipsoids by using 6 or more gradient directions, which allows to describe diffusion in 3D space and to calculate mean diffusivity (MD) and fractional anisotropy (FA) [10,11]. FA is a nondimensional parameter of anisotropic diffusion and yields values between 0 and 1 [12,13]. Low FA values indicate isotropic diffusion, whereas high FA values reflect high orientation of diffusion, e.g. along fiber tracts of white matter. Following Gupta et al. [14] and Nath et al. [15] pus in brain abscess cavities shows increased FA values with restricted MD compared to other cystic lesions. Those mainly reveal high MD and low FA values. Hence FA derived from DTI may help to better characterise brain abscesses and to better distinguish them from cystic tumours with less overlap, compared to DWI findings. The purpose of this study was to examine intracranial ring enhancing mass lesions with DWI and DTI and to evaluate the contribution of these techniques to differential diagnosis. For this purpose DWI and ADC findings were described and the DTI parameters MD and FA were measured quantitatively. The results were compared between different types of ring enhancing lesions. 2. Materials and methods 2.1. Patients From June 2006 to July 2008, 17 consecutive patients (9 males and 8 females, aged 58.8 ± 14.8 years (mean ± standard deviation)) presenting with one or more cystic cerebral lesions with variable perifocal edema and with rim enhancement on MRI postcontrast studies were included in this study. The MRI investigations including DWI and DTI sequences were acquired immediately after patients were admitted. The study concept was reviewed by the local ethical committee. All patients themselves, their deputies, or their relatives had given informed consent. The abscess and tumour diagnoses were confirmed by histopathology in 16 cases (14 open surgeries with total tumour resection and 2 stereotactic biopsies). Additionally, in all cases with brain abscesses, the causing microorganisms were isolated from abscess pus. In one case with multiple cerebral metastases after malignant melanoma the histopathological verification by open surgery was refused. In this patient the final diagnoses were based on the patient’s history, on MRI and CT studies of head, chest and abdomen, showing multiple metastases in the brain, the right axilla, the abdominal wall, and in the right gluteal area. 2.2. MRI protocol The DTI sequence was integrated in a routine MR investigation protocol and was acquired on a 1.5 T MR unit (Siemens, Erlangen, Germany) using a standard head coil. The conventional MR imaging protocol consisted of an axial diffusion-weighted EPI sequence with calculation of an ADC map, an axial T2 weighted TSE sequence, an axial T1 weighted GE sequence and a coronal FLAIR sequence. A T1 weighted 3D-MPRage sequence was run after intravenous administration of gadolinium contrast (0.5 molar) in a dose of 0.2 mmol/kg body weight followed by multiplanar reconstruction (MPR) performing axial, coronal, and sagittal slices. The DTI sequence was inserted prior to contrast media injection. We used a diffusionweighted 2D EPI sequence (TR 9000 ms, TE 109 ms, averages 4, slices 45, slice thickness 3 mm, FOV 230 mm, matrix 128 × 128, spatial resolution 1.8 mm × 1.8 mm). The diffusion b-factors were

219

set to 0 and 800 s/mm2 and diffusion encoding was done along 12 diffusion directions. The total acquisition time was 7.6 min. 2.3. DTI evaluation and region of interest (ROI) analysis We used the software program DTI & Fiber Tool (Medical Physics, Dept. of Diagnostic Radiology, University Hospital Freiburg, Germany) [16] for calculating mean diffusivity (MD) and fractional anisotropy (FA). After calculating a diffusion tensor the program computed and created voxel by voxel parameter maps for MD and FA. For this purpose the tensor was transformed in a diagonal form, i.e. the eigenvalues (1 , 2 and 3 ) and eigenvectors (e1 , e2 and e3 ) of the tensor were calculated analytically. The eigenvalues describe the diffusion along the axis of the diffusion ellipsoid and the eigenvectors reflect the orientation of the ellipsoid [10]. These tensor data were then used to compute the DTI metrics MD (Eq. (1)) and FA (Eq. (2)): ¯ = MD = 

 FA =

1 + 2 + 3 3

(1)

¯ 2 + (2 − ) ¯ 2 + (3 − ) ¯ 2 3 (1 − ) 2 2 2 2  + + 1

2

(2)

3

The DTI derived parameter maps were quantified by an experienced neuroradiologist using regions of interest (ROIs). These were placed manually in abscess cavities and tumour cysts respectively, in contrast enhancing rims, in perifocal edema zones and for the purpose of intern control in normal appearing white matter (NAWM) of the contralateral hemisphere and in the lateral ventricles. ROI sizes varied due to the different size of abscesses, tumour cysts and the anatomy of the control structures and were manually adapted pixel by pixel to their different size and configuration. The evaluation software enabled to place ROIs three dimensional in several successive slices by using axial, coronal and sagittal planes and to add measured values together. The ROI sizes ranged for the lesion’s cysts from 12 to 1407 pixel (median 85), for contrast enhancing rims from 20 to 802 (median 155), for perifocal edema from 40 to 1223 (median 356), for NAWM from 211 to 935 (median 427) and for lateral ventricles from 43 to 739 (median 193). ADC maps derived from DWI trace sequence were not quantified, because ADC and MD values, both describe diffusion and show rather similar values. Therefore DWI scans and ADC maps were evaluated only descriptively, because in clinical practice DWI sequence generates immediately images for diagnostic analysis while DTI needs additional post-processing work. We used the software SPSS for statistical analysis. Medians and ranges (minimum to maximum) were used to compare DTI metrics MD and FA of cysts, ring enhancing tissue, perifocal edema, NAWM and CSF in the lateral ventricles. Due to the low number of patients non-parametric Mann–Whitney U-test was chosen to evaluate significances. The significance level was set to 0.05. 3. Results Five patients suffered from cerebral abscesses with a total of 10 cysts (abscess group). The infectious agents were (1) Nocardia, (2) Micromonas micros, (3) a combination of Actinomyces meyeri and Streptococcus milleri, (4) a combination of Peptostreptrococcus species and Streptococcus intermedius and (5) Fusobacterium nucleatum. The other 12 patients suffered from brain tumours with a total of 16 cystic lesions (tumour group), 9 glioblastomas (WHO grade IV) with a total of 10 cysts and 3 metastases with a total of 6 cysts. The primary tumours were (1) melanoma, (2) small cell lung cancer and (3) anaplastic non-small cell lung cancer. Median values and ranges (minimum to maximum) of MD and FA measured from cysts, ring enhancing tissue, adjacent perifocal edema, NAWM and CSF of

0.78 (0.75–0.81) 0.11 (0.10–0.13) 1.07 (0.87–1.27) 0.31 (0.24–0.39) 1.34 (1.18-1.50) 0.24 (0.21–0.26) 0.78 (0.77–0.79) 0.45 (0.44–0.47) 3.21 (3.18–3.24) 0.10 (0.09–0.10) MD [×10−3 mm2 /s] FA [nondimensional] MD [×10−3 mm2 /s] FA [nondimensional] MD [×10−3 mm2 /s] FA [nondimensional] MD [×10−3 mm2 /s] FA [nondimensional] MD [×10−3 mm2 /s] FA [nondimensional]

0.47 (0.39–0.67) 0.20 (0.14–0.45) 1.28 (1.14–1.65) 0.18 (0.12–0.26) 1.82 (1.61–1.99) 0.15 (0.10–0.16) 0.75 (0.73–0.79) 0.48 (0.45–0.50) 3.38 (3.06–3.69) 0.12 (0.09–0.13)

2 6

median (min–max) 2.80 (2.14–2.88) 2.32 (1.84–2.78) 0.065 (0.056–0.077) 0.08 (0.06–0.14) 1.11 (0.91–1.43) 0.89 (0.77–0.99) 0.17 (0.12–0.29) 0.25 (0.23–0.28) 1.75 (1.59–1.94) 1.78 (1.62–1.94) 0.15 (0.14–0.16) 0.12 (0.12-0.18) 0.78 (0.71–0.82) 0.78 (0.77–0.81) 0.50 (0.41–0.54) 0.43 (0.41–0.50) 3.13 (3.04–3.27) 3.14 (3.12–3.15) 0.08 (0.06–0.12) 0.08 (0.07–0.12)

8 10

Glioblastomas (b)

Parameter

Brain abscess group (a)

Tumour group

Metastases (c)

Glioblastomas with hyperintense DWI signal (d)

p-Values

pa/b+c+d < 0.0001; pd/b+c = 0.026; pb/c = 0.033 pa/b+c+d < 0.0001; pa/d = 0.032; pb/c = 0.17 pa/c = 0.006 pa/c = 0.028 n.s. n.s. total median (min–max): 0.77 (0.71–0.82) 0.47 (0.41–0.54) 3.17 (3.04–3.70) 0.09 (0.06–0.13)

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

the lateral ventricles are summarized in Table 1 and displayed as boxplot diagrams in Fig. 1. In the abscess group all 10 cysts showed hyperintense signal on DWI sequence in relation to NAWM, diminished diffusion on correspondent ADC map and low DTI metric MD (0.47 × 10−3 mm2 /s, range 0.39–0.67) (Fig. 2). The abscess cysts presented high FA values (median 0.20, range 0.14–0.45) compared to low FA of CSF in lateral ventricles (median 0.12, range 0.09–0.13) (p = 0.002). FA values of NAWM were highest (0.48, range 0.45–0.50) (p = 0.002). Median MD for contrast enhancing abscess rings was 1.28 × 10−3 mm2 /s (range 1.14–1.65) and median FA 0.18 (range 0.12–0.26). In the tumour group 8 out of 10 glioblastoma cysts and all 6 metastatic cysts were hypointense on DWI scan and revealed high diffusion on ADC maps. Median MD values of cysts were 2.80 × 10−3 mm2 /s (range 2.14–2.88) for glioblastomas and 2.32 × 10−3 mm2 /s (range 1.84–2.78) for metastases respectively (Fig. 3) and differed with a p-value of 0.033, but their ranges overlapped considerably. The difference of MD values between the abscess and tumour group was significant, p < 0.0001. FA values of metastatic cysts (median 0.08, range 0.06–0.14) and 8 out of 10 glioblastoma cysts (median 0.065, range 0.056–0.077) differed not significantly (p = 0.17) and ranged within FA values of CSF. These FA values were significantly lower compared to abscess cavities (p < 0.0001). One metastatic cyst revealed a FA of 0.14, which overlapped with FA value of one abscess cavity. But MD in this neoplastic cyst was with 2.15 × 10−3 mm2 /s significantly higher than in the abscess case (MD = 0.55 × 10−3 mm2 /s). Two out of 10 glioblastoma cysts mimicked abscesses presenting hyperintense cyst contents on DWI sequence and low ADC on corresponding parameter map. Median MD was 0.78 × 10−3 mm2 /s (range 0.75–0.81) and lower compared to the other tumour cysts (p = 0.026). The median FA of these 2 glioblastoma cysts was 0.11 (range 0.10–0.13), which was within the range measured for CSF in lateral ventricles (0.06–0.13) and significantly lower compared to FA of abscess pus (p = 0.032). There was no significant difference in relation to the other tumour cysts (p = 0.056). In the tumour group ring enhancing tissue revealed median MD of 1.11 × 10−3 mm2 /s (range 0.91–1.43) for glioblastomas and 0.89 × 10−3 mm2 /s (range 0.77–0.99) for metastases, median FA were 0.17 (range 0.12–0.29) and 0.25 (range 0.23–0.28) respectively. The differences between the rims of abscesses and glioblastomas were for both DTI metrics not significant, between abscesses and metastases p(MD) was 0.006 and p(FA) 0.028. Median MD and FA for perifocal edema of abscesses and tumours did not show any statistically significant differences. For calculating sensitivity, specificity and predictive values for differentiation of cerebral abscesses from cystic brain tumours cut-off values for MD and FA were evaluated. For the parameter MD, best differentiation was achieved with a cut-off value ≤0.9 × 10−3 mm2 /s for abscesses yielding in a sensitivity of 100%, a specificity of 87.5% and a predictive value of 83.3%. For FA best discrimination was attained with a cut-off value of ≥0.14 for abscesses resulting in a sensitivity of 100% and improvement of specificity to 93.8% and of predictive value to 90.9% respectively. Specificity and predictive value could be improved up to 100% by combining MD and FA.

Normal appearing white matter CSF in lateral ventricles

Perifocal edema

Ring enhancing tissue

Cyst

Number of cysts

4. Discussion Lesion’s compartment

Table 1 Quantitative results (median and min–max) of mean diffusivity (MD) and fractional anisotropy (FA) values derived from DTI scan of ring enhancing lesions (cysts, ring enhancing tissue, adjacent perifocal edema) and for comparison of normal appearing white matter (NAWM) and CSF of the lateral ventricles.

220

Cerebral ring enhancing lesions may represent cerebral abscesses as well as other cystic mass lesions, especially cystic or necrotic brain tumours. Differential diagnosis of such ring enhancing lesions is difficult on conventional MRI sequences and they can generally not be differentiated precisely. It has been shown, that DWI can help to distinguish between abscesses and cystic tumours [1–5]. In our series, all 10 abscess cysts revealed

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

221

Fig. 1. Boxplot diagrams of quantitative results of MD (a) and FA (b) of the cysts of ring enhancing lesions and for comparison of normal appearing white matter (NAWM) and CSF in lateral ventricles. (Centre bars correspond to medians, boxes indicate lower and upper quartiles, range lines represent the Whiskers, their maximum lengths are the 1.5 times of interquartile ranges, small circles indicate mild and stars extreme outliers.)

hyperintense DWI signal intensity with low ADC and low DTI metric MD respectively. In the non-abscess group 14 out of 16 tumour cysts were hypointense on DWI scan and showed high ADC and high MD. Our results are in agreement with previous reports showing increased DWI signal intensity in abscess cavities compared to normal brain parenchyma reflecting restricted diffusion (sensitivity 72–96%), whereas it is usually hypointense in tumour cysts and necroses respectively (specificity 96–100%) reflecting more isotropic diffusion like in CSF [9,17,18]. Different diffusion values (ADC derived from DWI and MD derived from DTI) in abscess cavities compared to tumour cysts are due to the different compositions of their cyst content. Abscess pus is a creamy and viscous fluid containing inflammatory cells, bacteria, mucoid proteins and cell debris. The high cellularity of pus represents the main biological parameter leading to a diminution of the extracellular space and to decreased diffusion [19]. Furthermore the binding of water to carboxyl-, hydroxyl- and amino-groups on the surface of large macromolecules like mucoid proteins and fibrinogen may contribute additionally to restricted

water motion and diffusion in abscess pus [19–21]. Thus MD reflects the density of pus and inflammatory cells in abscess cavities. However diffusion restriction in cystic brain lesions is not specific for brain abscesses [6–9,17,18]. In our series 2 out of 16 neoplastic cysts showed hyperintense DWI finding with low ADC and low MD, and thus mimicked cerebral abscesses. Reddy et al. [9] reported about 2 out of 50 glioblastomas with hyperintensity on DWI and low ADC value, Lai et al. [18] found in one case (out of 23) with metastasis from colon adenocarcinoma also high DWI signal and low ADC. Therefore we used in our study additional DTI sequence besides DWI to better distinguish infectious from neoplastic cysts. DTI allows to calculate two parameters: (1) MD, which is equivalent to ADC derived from DWI sequence, and (2) FA, FA may separate more clearly true abscesses from tumour cysts with high signal intensity on DWI. We found high FA values in the cavities of all 10 abscess cysts. FA values ranged between those of perifocal edema and NAWM. In contrast FA was much lower in 15 out of 16 tumour

222

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

Fig. 2. 61-Year-old patient with an abscess in the right frontal lobe. The abscess shows the typical image pattern with a large central cyst on conventional T2 weighted (a) and ring enhancement on post-contrast T1 weighted image (b). The abscess cavity is hyperintense on DWI-EPI scan (c) and shows restricted diffusion on ADC map (d) and as well on MD parameter image derived from DTI scan (0.50 × 10−3 mm2 /s) (e). The FA parameter map reveals FA elevation (0.22) (f).

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

223

Fig. 3. 68-Year-old patient suffering from glioblastoma on the left side. Conventional MR images (T2 weighted (a) and post-contrast T1 weighted (b)) show a tumour with a central cyst, ring enhancement, a small contrast enhancing satellite structure and perifocal edema. The tumour cyst is hypointense in DWI-EPI sequence (c) and reveals no signs of diffusion restriction either on ADC map (d) nor on MD parameter image derived from DTI sequence (0.28 × 10−3 mm2 /s) (e). FA is not elevated (0.077) (f).

224

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

cysts using a cut-off value of <0.14 and showing FA values quite similar to CSF measured in lateral ventricles. One metastatic cyst revealed a FA value of 0.14, which overlapped with the lower threshold of abscesses. But in this case diffusion was not restricted and MD showed a high value (>0.9 × 10−3 mm2 /s), so that neoplastic origin could be diagnosed correctly. Nath et al. [15] found a similar FA cut-off value of 0.12. They observed in 40 patients with brain abscesses FA values >0.12, whereas FA was <0.12 in 10 out of 13 patients with cystic neoplastic tumours. Markedly increased FA in abscess cavities compared to low FA values in cavities of cystic astrocytomas and neurocysticercosis were reported also by Gupta et al. [14]. The DTI parameter FA is an index value of diffusion anisotropy, which reduces the three-dimensional shape of anisotropy to an one-dimensional parameter yielding values between 0 and 1 [12,13]. Hence high FA values are found, e.g. in white matter, where the orientation of diffusion ellipsoids follows the local course of fiber tracts. Very low FA on the other hand indicates free isotropic diffusion like in CSF in the ventricles. Our results and published data [14,15] show, that FA values of abscess pus are somewhat lower as those measured in NAWM and much higher than those of CSF, suggesting a certain structure in abscess cavities. Recently it has been shown, that increased FA in abscesses is mainly caused by adhesive neuroinflammatory molecules reflecting an upregulation of anti-inflammatory processes [19,22]. In contrast to abscesses the content of cysts and necrotic compartments of primary and secondary brain tumours consists of tissue debris and fewer inflammatory cells. Cyst fluid is almost clearer and more serous than pus. Hence, there are less diffusion barriers in tumour cysts than in abscess cavities. This explains a more isotropic diffusion leading to MD values, which are almost as high as CSF in lateral ventricles and low FA values. There is a diagnostic problem in a small number of neoplastic cysts presenting with atypical DWI and ADC findings, which overlap with those from abscesses [6–9,17,18]. In our series, two tumour cysts showed restricted diffusion with low ADC and MD, and could not be differentiated precisely from abscesses, but they revealed low FA values between 0.10 and 0.13, which were below the cut-off value of 0.14 and within the range for CSF in the lateral ventricles (0.06–0.13). The histopathological findings of stereotactic biopsy and resection specimen revealed tumour coagulation necrosis in one case and large areas of necrosis in the other. Thus densely packed cell debris together with proteinaceous ingredients in tumour cysts may alter diffusion in a small number of patients but does not have the potential to increase FA as high as pus in abscess cavities. So the DTI parameter FA may help to distinguish densely packed tumour cysts with restricted diffusion from abscesses. A further rare finding is increased FA in cavities of neoplastic tumours. Nath et al. [15] described in 3 out of 13 patients with tumour cysts FA values >0.12 and explained, that these high FA values may be due to the presence of haemorrhage. It has been shown, that haemorrhagic brain lesions can cause increased FA [23]. In clinical routine haemorrhage in cysts can be recognised by its heterogeneous signal appearance on T2 weighted B0 images of the DWI sequence, by using additional susceptibility sensitive T2* weighted sequences or by measuring density on CT scans. In the present study no haemorrhage was detected in cysts of brain tumours or abscesses. Additionally there are cases of cerebral abscess cavities showing hypointensity on DWI and high ADC values which overlapped with findings of neoplastic cysts. Mishra et al. [17] reported, that pus culture from 5 out of 8 patients with high ADC was sterile and that an antibiotic drug therapy had been initiated already in these patients. In the series of Reddy et al. [9] 4 brain abscesses in 3 patients appeared isointense to hypointense on DWI and high ADC values were false negative and hence overlapped with those of tumour

cysts. Lai et al. [18] described in 1 out of 21 patients with pyogenic abscesses hypointensity on DWI and high ADC. This patient had received antibiotic therapy for already 21 days. The cause of increasing diffusion in abscess cysts might be due to changes in pus composition and probably reflects increasing pus liquefaction as result of adequate antibiotic therapy. Nath et al. [15] reported about 10 out of 40 patients with abscesses, which showed high FA values even in the presence of increased diffusion and stated, that this effect might be secondary to antibiotic therapy due to the presence of oriented inflammatory cells secondary to up-regulation of adhesion molecules. How can DTI be used effectively in daily clinical routine MRI diagnostic? Considering not to waste valuable MRI investigation time, is DTI needed in every case presenting with a ring enhancing lesion? A possible multimodal diagnostic pathway in clinical routine could be to integrate first DWI as a basic diffusion sequence in a MRI protocol, which consists further of conventional T1 and T2 weighted sequences including intravenous administration of Gadolinium contrast and a perfusion study. We use additional DTI mainly in cases, where DWI shows hyperintense signal and low ADC within a cystic rim enhancing lesion in order to differentiate infectious from non-infectious etiology. In the case of hypointense cyst content on DWI an abscess seems to be unlikely except abscesses, which are under antibiotic therapy. If in such a case there are clinical evidences of an infectious etiology (e.g. fewer, inflammatory laboratory constellation, signs of meningism) or an antibiotic therapy has been initiated, we perform additional DTI sequence for differential diagnostic reasons. Haemorrhage in tumour cysts, which may cause increase of FA as well and thus overlap with findings in abscesses, can be identified normally due to their heterogeneous signal intensities on susceptibility sensitive T2 weighted B0 images of the diffusion-weighted EPI sequence. In such a case we perform an additional T2* weighted GE sequence. Further studies are necessary to evaluate such a diagnostic approach for daily clinical routine work. A further attempt to improve MR based characterisation of intracranial cystic lesions may be to combine DWI or DTI with proton magnetic resonance spectroscopy (H-MRS) or with MRperfusion [15,18,24,25]. Cerebral abscesses show beside elevation of lipids characteristic peaks of cytosolic aminoacids and lactate with or without additional resonances of acetate, alanine and succinate which originate from granulocytes in pus and arise from decay of infectious microorganisms. Cystic tumours are characterised by elevation of choline (Ch) and lactate. Nath et al. [15] found, that DTI and H-MRS were complementary and could improve the specificity in differentiating abscesses from cystic tumours. Drawback of additional H-MRS is, that availability of H-MRS for clinical routine diagnostic is limited. Although there is a good availability of MRI units for diagnostic imaging, the number of MRI’s equipped with H-MRS technology is limited. A further problem of H-MRS is the possibility of spectra contamination with signals from extralesional compounds especially in small lesions. At least H-MRS is more time consuming than a simple DTI sequence and therefore will lengthen MR examinations, what a lot of seriously ill patients probably would not tolerate. MR-perfusion shows decreased cerebral blood volume (CBV) in abscess rims compared to NAWM, whereas CBV of ring enhancing tissue of glioblastomas and metastases is significantly increased [24,25]. Furthermore CBV correlates in gliomas with malignancy grade, gives valuable informations for planning surgical procedures and may be helpful in predicting survival. Disadvantage of MR-perfusion can be, that in cases with very thin enhancing tissue rims, these cannot be identified correctly on CBV parameter maps. However, MR-perfusion can be incorporated easily in MRT routine protocols and enables combined with DWI, DTI and H-MRS respectively a multimodal approach to differential diagnosis of cystic brain lesions.

W. Reiche et al. / Clinical Neurology and Neurosurgery 112 (2010) 218–225

Neurosurgical biopsy, aspiration and drainage either stereotactic, ultrasound guided or free-hand is considered to be first-line treatment of cerebral abscesses, because expected brain tissue damage will be less compared to open surgery with abscess excision [26]. Therefore differentiation of pyogenic brain abscess from cystic malignant tumour by appropriate application of DWI and DTI is important not only in diagnosis and follow-up of antibiotic therapy but also will influence neurosurgical approach to these lesions. In case of an abscess, the centre should be target with intentions to aspirate pus and to culture the causing pathogens [27]. In contrast, the optimal site for tissue diagnosis of a malignant cystic tumour is the contrast enhancing tumour ring. The drawback of this study is the limited number of patients especially in the abscess group. The reason for this is, that cerebral abscesses are rare in Western Europe compared to primary and secondary brain tumours. However, we think, that previous reports about DWI and DTI [9,14,15,17–19,22] coupled with our additional cases show the usefulness of the DTI metric FA for distinguishing infectious abscesses from those cystic ring enhancing lesions, which are hyperintense in DWI with low ADC and hence mimicked abscesses. In conclusion the constellation of reduced diffusion compared to NAWM and increased FA within a ring enhancing cyst strongly indicates a cerebral abscess. The majority of neoplastic cysts derived from glioblastomas and metastases reveal elevated diffusion and low FA. Reduced diffusion within tumour cysts due to densely packed cell debris and proteinaceous compounds is a rare finding. In such a case, low FA values could be a valuable parameter indicating no further orientation of diffusion and hence helping to distinguish neoplastic cysts mimicking abscesses from true infectious abscesses. Increased FA may appear also in haemorrhagic cysts. Such haemorrhage can be distinguished from abscess pus on susceptibility sensitive T2 weighted B0 images of the DWI sequence or additional T2* weighted GE sequence. References [1] Ebisu T, Tanaka C, Umeda M, Kitamura M, Naruse S, Higuchi T, et al. Discrimination of brain abscess from necrotic or cystic tumors by diffusion-weighted echo planar imaging. Magn Reson Imaging 1996;14:1113–6. [2] Tien RD, Felsberg GJ, Friedmann H, Brown M, MacFall J. MR imaging of high-grade cerebral gliomas: value of diffusion-weighted echoplanar pulse sequences. Am J Roentgenol 1994;162:671–7. [3] Kim YJ, Chang KH, Song IC, Kim HD, Seong SO, Kim YH, et al. Brain abscess and necrotic or cystic brain tumor: discrimination with signal intensity on diffusion-weighted MR imaging. Am J Roentgenol 1998;171:1487–90. [4] Desprechins B, Stadnik T, Koerts G, Shabana W, Braucq C, Ostreaux M. Use of diffusion-weighted MR imaging in differential diagnosis between intracerebral necrotic tumors and cerebral abscesses. Am J Neuroradiol 1999;20:1252– 7. [5] Noguchi K, Watanabe N, Nagayoshi T, Kanazawa T, Toyoshima S, Shimizu M, et al. Role of diffusion-weighted echo-planar MRI in distinguishing between brain abscess and tumor: preliminary report. Neuroradiology 1999;41:171–4.

225

[6] Holtås S, Geijer B, Strömblad LG, Maly-Sundgren P, Burtscher IM. A ringenhancing metastasis with central high signal on diffusion-weighted imaging and low apparent diffusion coefficients. Neuroradiology 2000;42:824–7. [7] Hartmann M, Jansen O, Heiland S, Sommer C, Münkel K, Sartor K. Restricted diffusion within ring enhancement is not pathognomic for brain abscess. Am J Neuroradiol 2001;22:1738–42. [8] Tung GA, Evangelista P, Rogg JM, Duncan 3rd JA. Diffusion-weighted MR imaging of rim-enhancing brain masses: is markedly decreased water diffusion specific for brain abscess? Am J Roentgenol 2001;177:709–12. [9] Reddy JS, Mishra AM, Behari S, Husain M, Gupta V, Rastogi M, et al. The role of diffusion-weighted imaging in the differential diagnosis of intracranial cystic mass lesions: a report of 147 lesions. Surg Neurol 2006;66:246–50. [10] Herrnberger B. Diffusion tensor magnetic resonance imaging. Nervenheilkunde 2004;23:50–9. [11] Le Bihan D. Looking into the functional architecture of the brain with diffusion MRI. Nat Rev Neurosci 2003;4:469–80. [12] Basser PJ, Pierpaoli C. Microstructural and physiological features of tissues elucidated by quantitative-diffusion-tensor MRI. J Magn Reson B 1996;111:209–19. [13] Papadakis NG, Xing D, Houston GC, Smith JM, Smith MI, James MF, et al. A study of rotationally invariant and symmetric indices of diffusion anisotropy. Magn Reson Imaging 1999;17:881–92. [14] Gupta RK, Hasan KM, Mishra AM, Jha D, Husain M, Prasad KN, et al. High fractional anisotropy in brain abscesses versus other cystic intracranial lesions. Am J Neuroradiol 2005;26:1107–14. [15] Nath K, Agarwal M, Ramola M, Husain M, Prasad KN, Rathore RKS, et al. Role of diffusion tensor imaging matrics and in vivo proton magnetic resonance spectroscopy in the differential diagnsosis of cystic intracranial mass lesions. Magn Reson Imaging 2009;27:198–206. [16] Kreher BW, Hennig J, Il’yasov KA. DTI & FiberTools: a complete toolbox for DTI calculation, fiber tracking, and combined evaluation. Proc Intl Soc Magn Reson Med 2006;14:2758. [17] Mishra AM, Gupta RK, Jaggi RS, Reddy JS, Jha DK, Husain N, et al. Role of diffusion-weighted imaging and in vivo proton magnetic resonance spectroscopy in the differential diagnosis of ring-enhancing intracranial cystic mass lesions. J Comput Assist Tomogr 2004;28:540–7. [18] Lai PH, Hsu SS, Ding SW, Ko CW, Fu JH, Weng MJ, et al. Proton magnetic resonance spectroscopy and diffusion-weighted imaging in intracranial cystic mass lesions. Surg Neurol 2007;68(Suppl. 1):S25–36. [19] Mishra AM, Gupta RK, Saksena S, Prasad KN, Pandey CM, Rathore D, et al. Biological correlates of diffusivity in brain abscess. Magn Reson Med 2005;54:878–85. [20] Rusakov DA, Kullmann DM. Geometric and viscous components of the tortuosity of the extracellular space in the brain. Proc Natl Acad Sci USA 1998;95:8975–80. [21] Castillo M. Imaging brain abscess with diffusion-weighted and other sequences. Am J Neuroradiol 1999;20:1193–4. [22] Gupta RK, Nath K, Prasad A, Prasad KN, Husain M, Rathore RK, et al. In vivo demonstration of neuroinflammatory molecule expression in brain abscess with diffusion tensor imaging. Am J Neuroradiol 2008;29:326–32. [23] Haris M, Gupta RK, Husain N, Hasan KM, Husain M, Narayana PA. Measurement of DTI metrics in hemorrhagic brain lesions: possible implication in MRI interpretation. J Magn Reson Imaging 2006;24:1259–68. [24] Erdogan C, Hakyemez B, Yildirim N, Parlak M. Brain abscess and cystic brain tumor: discrimination with dynamic susceptibility contrast perfusionweighted MRI. J Comput Assist Tomogr 2005;29:663–7. [25] Chiang IC, Hsieh TJ, Chiu ML, Liu GC, Kuo YT, Lin WC. Distinction between pyogenic brain abscess and necrotic brain tumour using 3-tesla MR spectroscopy, diffusion and perfusion imaging. Br J Radiol 2009 May 26 [Epub ahead of print]. [26] Strowitzki M, Schwerdtfeger K, Steudel WI. Ultrasound-guided aspiration of brain abscesses through a single burr hole. Minim Invasive Neurosurg 2001;44:135–40. [27] Leuthardt EC, Wippold FJ, Oswood MC, Rich KM. Diffusion-weighted MR imaging in the preoperative assessment of brain abscesses. Surg Neurol 2002;58:395–402.