- Email: [email protected]
A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade
Clinical Radiology xxx (2016) e1ee7
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade L. Kong a, c, H. Chen b, c, Y. Yang a, L. Chen a, * a b
Department of Anesthesiology, Anhui Provincial Cancer Hospital, Hefei 230031, China Department of Anesthesiology, Nanjing General Hospital of Nanjing Military Command, Nanjing 210002, China
art icl e i nformat ion Article history: Received 30 January 2016 Received in revised form 23 July 2016 Accepted 25 October 2016
AIM: To investigate the ability of arterial spin labelling (ASL) perfusion parameters to distinguish high-grade from low-grade gliomas. MATERIALS AND METHODS: The PubMed and EMBASE databases were systematically searched for relevant articles published up to September 2015. Studies that evaluated both high- and low-grade gliomas using ASL were included. The random effect model was used to calculate the standardised mean difference (SMD) of maximum mean absolute tumour blood flow values (aTBFmax, aTBFmean) and maximum mean relative tumour blood flow (rTBFmax, rTBFmean) between high- and low-grade gliomas. RESULTS: Nine studies encompassing 305 patients with high- and low-grade gliomas, met all inclusion and exclusion criteria and were included in the study. Compared with low-grade gliomas, high-grade gliomas had a significant increase in all ASL perfusion values: aTBFmax (SMD¼0.70, 95% confidence interval [CI]: 0.22e1.19, p¼0.0046); aTBFmean (SMD¼0.86, 95% CI: 0.2e1.52, p¼0.01); rTBFmax (SMD¼1.08, 95% CI: 0.54e1.63, p¼0.0001) and rTBFmean (SMD¼0.88, 95% CI: 0.35e1.4, p¼0.0011). CONCLUSIONS: The current study results indicate that tumour blood flow from ASL differs significantly with respect to the glioma grade. Despite some limitations, there is evidence that ASL may be useful to distinguish high- and low-grade gliomas. Further larger-scale studies are necessary to examine the utility of ASL to distinguish tumour grade. Ó 2016 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction Gliomas are the most common primary brain tumours in adults,1 and a precise diagnosis is important because the adjuvant therapy after surgery and prognosis differ
* Guarantor and correspondent: L. Chen, Department of Anesthesiology, Anhui Provincial Tumor Hospital, No. 107, East Huanhu Road, Hefei, 230031, China. Tel.: þ86 551 65897820. E-mail address: [email protected] (L. Chen). c These authors contributed equally to this work.
considerably according to tumour grade.2 Conventional morphological imaging technologies are usually limited in grading gliomas. For example, enhancement is not a reliable factor for determining tumoural grade.3,4 Perfusion imaging is the commonly used advanced imaging method for the evaluation of brain tumours.5,6 Currently, two major types of imaging methods are available to evaluate brain tumour perfusion. One is dynamic susceptibility contrast (DSC) perfusion imaging and the other is arterial spin labelling (ASL). One practical advantage of ASL technique is that it relies on endogenous tracers, such as water molecules, and can be
http://dx.doi.org/10.1016/j.crad.2016.10.016 0009-9260/Ó 2016 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
e2
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
repeated many times in patients with renal failure.7,8 Several recent clinical studies have highlighted ASL techniques as helpful in the differential diagnosis of brain tumours; for example, low-grade gliomas from high-grade gliomas, and gliomas from other type of tumours.9e13 Clinically established methods for quantitative perfusion measurements usually include absolute tumour blood flow (aTBF) and relative tumour blood flow (rTBF), which is the normalisation of the aTBF values in relation to mean values in normal appearing regions. Some studies have investigated the value of TBF to distinguish high- from low-grade gliomas, but the findings have been incongruent. For example, Warmuth et al.14 demonstrated that mean (aTBFmean) and maximum aTBF (aTBFmax) values were significantly higher in high-grade than in low-grade brain tumours. Cebeci et al.15 reported significant positive correlations between rCBV from DSC and rCBF from ASL, and they considered relative values from ASL were more reliable. In contradistinction, Roy et al.16 demonstrated that ASL-derived absolute CBF values were not significantly different between histopathologically proven high- and low-grade glioma, even after normalising these values from the contralateral regions. They concluded that despite the advances in the technical developments of ASL, the currently available ASL method still suffers from interpatient variability.16 To the authors’ knowledge, the efficiency of ASL to distinguish between high-grade and low-grade gliomas has not been evaluated quantitatively. Therefore, the main objective of the present study was to conduct a quantitative meta-analysis of the existing literature to determine the statistical consensus of aTBF, and rTBF in distinguishing the tumour grade of gliomas.
grade IIIeIV) and low-grade gliomas (WHO gIeII); (2) histopathological analysis was used as the reference standard; (3) patients had no radiotherapy, surgery, or chemotherapy before ASL; (4) studies reported aTBF values and/or the rTBF values were available for effective calculation; (5) the number of patients should be at least eight; (6) no data overlapped between studies, if studies had the same or overlapping data, only the largest study was included in the final analysis. Studies were excluded based on the following criteria: animal studies, abstracts, reviews, case report, letters, editorials, comments, and conference proceedings. Articles that did not provide adequate information to allow the calculation of values were also excluded.
Materials and methods
rTBFmean¼ aTBFmean/reference value.
Search strategy
Reference regions may be normal-appearing grey matter, white matter, the contralateral mirrored region, or the global mean value. Data were recorded at the patient level, when possible. Each reviewer extracted study information on a standardised Microsoft Excel spreadsheet. Discrepancies were resolved by consensus. The same two reviewers independently assessed the quality of each article using the modified Quality Assessment of Diagnosis Accuracy Studies (QUADAS-2) score tool,18 which consisted of 14 questions answered ‘‘yes”, ‘‘no”, or ‘‘unclear”. Disagreements between the two investigators were resolved by consensus, and if disagreement persisted, a third reviewer made the ultimate decision.
This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.17 A systematic search was performed in the PubMed and EMBASE for relevant publications from inception to September 2015. The search terms were as follows: (“arterial spin labelling” or “arterial spin labelling” or “arterial spin labelled” or “ASL”) AND (“glioma” or “brain neoplasm” or “brain tumour” or “brain tumour”). There was no language restriction. Additionally, the references of selective articles were searched manually to identify potentially relevant studies that were not identified in the previous searches. Two authors independently performed the search, reviewed all eligible articles, and excluded obviously irrelevant studies by reading titles, abstracts, and keywords. The full text of each article was obtained when one or both reviewers were unsure or recommended review of the full text of the article.
Inclusion and excluded criteria The inclusion criteria were as follows: (1) ASL was used to measure perfusion values of both high-grade (WHO
Data extraction and quality assessment The final articles were assessed independently by the same two authors. For each included study basal characteristics (authors, year of publication, and country of origin), patient characteristics (mean age, sex, and number and type of gliomas), and technical aspects (imaging field strength, ASL techniques, reference standard, and the method for TBF measurement) were noted. The aTBFmean, aTBFmax, mean rTBF (rTBFmean), and maximum rTBF (rTBFmax) values were tabulated as mean values and SDs. aTBFmax was obtained by placing a region of interest (ROI) in the apparent maximum TBF region and average several ROIs within the tumour to create the aTBFmean. rTBF values were evaluated by normalising aTBF values to reference regions: rTBFmax ¼ aTBFmax/reference value and
Statistical analysis The mean difference (MD) of aTBFmean and aTBFmax and the standardised mean differences (SMDs) of aTBFmean, aTBFmax, rTBFmean, and rTBFmax, between high-grade and low-grade gliomas were calculated and used as effect-size statistics. The SMD is the standardised difference between two means and can be calculated as the difference between the high- and low-grade glioma groups divided by the
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
pooled standard deviation (SD). The overall effect size was presented with the mean and 95% confidence interval (CI). The extent of heterogeneity was assessed by the chisquared value test and the inconsistency index (I2). Low, moderate, and high I2 values were considered to be 25%, 50%, and 75%, respectively.19 Publication bias was assessed by the funnel plot asymmetry test. Subgroup analysis was performed using a more homogeneous set of studies with similar variables. Subgroups were only constructed when at least three studies could be included. The abovementioned statistical analysis was performed with R (version 3.2, http://www.r-project.org) and the metafor package (http://cran.r-project.org/web/packages/metafor/ index.html) was used to implement a random-effects model.
Results Literature identification This search produced 90 articles from PubMed and 179 articles from EMBASE. A total of 236 articles remained after using EndNote citation manager (Thomson Reuters, New York, NY, USA) to remove duplicates. After screening titles and abstracts, 20 articles were considered potentially eligible and were retrieved in full text for further assessment. After a full-text review, nine studies were considered eligible for this review.11,12,14,16,20e24 The selection process is demonstrated in the flowchart in Fig 1.
e3
Study characteristics and methodological quality assessment Nine studies, comprising 197 patients with high-grade gliomas and 108 patients with low-grade gliomas, met all the inclusion and exclusion criteria and were included in the present meta-analysis. The characteristics of the included studies are presented in Table 1. Among them, six studies were prospective studies, and three were retrospective studies, with sample sizes ranging from 9 to 64. Four studies used both a TBF and rTBF values, four studies used only rTBF; one study used only aTBF. Seven studies compared ASL with other imaging techniques, such as, magnetic resonance spectroscopy (MRS), dynamic susceptibility-weighted MRI (DSC), dynamic contrastenhanced MRI (DCE), diffusion tensor imaging (DTI). Two studies used ASL to distinguish gliomas from metastases, central nervous system (CNS) lymphomas, or meningioma. The revised ‘‘QUADAS-2” instrument was used to evaluate each included study. Of 14 questions, no study met all of them. Most studies were of high (moderate) quality, which satisfied the majority of standards. All nine studies had histopathology (biopsy or surgical resection) as the sole reference standard, but risks of bias in “flow and timing” were generally high. The assessment of study quality by QUADAS-2 is graphically summarised in Fig 2.
Quantitative analysis aTBFmax and aTBFmean Of the nine included studies, five studies used aTBF11,12,14,16,24 The pooled data revealed a significant inmax. crease of aTBFmax in high-grade gliomas compared with low-grade gliomas (MD¼45.08 ml/min/100 g, 95% CI: 22.31e67.85, p¼0.002; SMD¼0.70, 95% CI: 0.22-1.19, p¼0.0046), with a high level of heterogeneity (I2¼65%, p¼0.04 for MD and I2¼50%, p¼0.08 for SMD; Fig 3a,c). Four studies11,12,14,24 reported values of aTBFmean. The pooled data showed that aTBFmean was significantly higher in high-grade gliomas than low-grade gliomas (MD¼29.18 ml/min/100 g, 95% CI: 2.04e56.33, p¼0.02; SMD¼0.86, 95% CI: 0.2e1.52, p¼0.01), with a high level of heterogeneity (I2¼84%, p¼0.0004 for MD and I2¼56%, p¼0.08 for SMD; Fig 3b,d).
Figure 1 Flow diagram of the study selection process.
rTBFmax and rTBFmean Five studies11,16,20,22,23 used the rTBFmax. The rTBFmax in high-grade gliomas increased significantly compared to low-grade gliomas (SMD¼1.08, 95% CI: 0.54e1.63, p¼0.0001) with high heterogeneity (I2¼61%, p¼0.03; Fig 4a). Five studies11,12,14,20,21 used the rTBFmean. The pooled data revealed that rTBFmean in high-grade gliomas was higher significantly than low-grade gliomas (SMD¼0.88, 95% CI: 0.35e1.40, p¼0.0011; Fig 4b); heterogeneity was moderate (I2¼39%, p¼0.15).
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
e4
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
Table 1 Characteristics of included studies for meta-analysis. Author
Year
No. of patients (n)
M/F
Mean age (yr) (range)
Study type
Histology
FS (T)
ASL technique
Parameters
Warmuth Wolf Weber Chawla Lehmann Weber Roy Yeom Lin
2003 2005 2006 2007 2010 2010 2013 2014 2015
18 26 54 35 9 57 64 18 24
9/9 18/8 N 21/14 8/1 34/23 N N 19/5
38.2(9e64) 47(20e66) N 45(20e68) 53(19e82) N N N 42(16e65)
P P P R P P P R R
LGG(9);HGG(9) LGG(7); HGG(19) LGG(9) vs. HGG(46) LGG(13);HGG(22) LGG(5) vs. HGG(4) LGG(15) vs. HGG(42) LGG(26); HGG(38) LGG(14);HGG(4) LGG(11); HGG(13)
1.5 3 1.5 3 3 1.5 3 3 3
pASL cASL pASL cASL pcASL pASL pcASL pcASL pcASL
aTBFmean; aTBFmax; rTBFmean aTBFmean; aTBFmax; rTBFmean; rTBFmax rTBFmean; rTBFmax aTBFmax; aTBFmean; rTBFmean rTBFmean rTBFmax aTBFmax; rTBFmax rTBFmax aTBFmean; aTBFmax
HGG, high-grade glioma; LGG, low-grade glioma; FS, Field strength; R, retrospective; P, prospective; aTBFmean, average of absolute tumour blood flow value; aTBFmax, maximum of absolute tumour blood flow value; rTBFmean, aTBFmean normalised to normal appearing tissue; rTBFmax, aTBFmax normalised to normal appearing tissue; cASL, continuous arterial spin labelling; pASL, pulsed arterial spin labelling; pcASL, pseudo-continuous ASL; N, not reported.
Sensitivity analysis and publication bias In the sensitivity analysis, no significant change in the results was evident when excluding studies one by one for all perfusion values. The funnel plot was found to be symmetrical, suggesting a low likelihood of publication bias.
Subgroup analyses In stratified analyses, high-grade gliomas exhibited a significant increase in all perfusion parameters compared with low-grade gliomas in subgroups of publication year (before 2011), field strength (3 T), reference (grey matter), and reference (global). Subgroup analyses were only performed when at least three studies could be included. These results are shown in Table 2.
Discussion Accurate tumour grading is essential to determine the choice of therapeutic approach and to the assessment of prognosis.2,25 Conventional MRI techniques are often unreliable on tumour grading. With the advent of advanced imaging techniques, the ability to accurately grade tumours has been a subject of interest. Previous studies26,27 have demonstrated that information about vessel permeability may contribute to the grading of gliomas. Perfusion MRI is one of the most effective methods for quantifying the glioma grading; however, current reports mostly contain data from small samples and the results are inconsistent. Thus,
the present meta-analysis was performed with the aim of resolving the incongruities by increasing sample size and improving the accuracy of evaluation of effect size. In the present meta-analysis, the utility of ASL perfusion values to distinguish high-grade gliomas from low-grade was examined. Pooled analysis revealed that high-grade gliomas had a significant increase in all measured values relative to low-grade gliomas. These findings remained largely unchanged when sensitivity analysis and allsubgroup analyses was performed, suggesting that these results were very unlikely due to chance. To the authors’ knowledge, this is the first meta-analysis to assess the overall values of ASL perfusion values in glioma grading. The aTBF is widely used in the assessment of glioma grading. It allows comparison of patients with each other or comparison of values in individual patients during the course of treatment; however, Warmuth et al.14 concluded that aTBF is not a crucial factor for tumour grading. The present results indicated that the TBF (maximum and mean) value was significantly higher in high-grade gliomas than in low-grade gliomas, suggesting that TBF is a potential indicator of malignancy in gliomas. To correct for age-dependent and patient-dependent variations, rTBF is calculated by using absolute perfusion values normalised to the normal-appearing tissue. Several studies revealed that rTBF is useful for characterising gliomas and can be used to discriminate glioblastomas and other lesions.28,29 Previous studies have used contralateral white matter,30 grey matter,31,32 or the cerebellum23 as the €rnum et al.9 believed that white internal references. Ja
Figure 2 Quality assessment with revised QUADAS-2 tool. Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
e5
Figure 3 The comparison of aTBF values between high-grade and low-grade gliomas. (a) SMD of aTBFmax; (b) SMD of aTBFmean; (c) MD of aTBFmax; (d) MD of aTBFmean.
matter perfusion measurements with ASL may have disadvantages and cerebellar perfusion usually is unaffected by pathology in the brain. Among nine studies included in this meta-analysis, three studies used grey matter as the internal references, two studies used white matter, three
studies used contralateral mirrored tissue, and one used the global averaged value. The difference between the rTBF between high- and low-grade gliomas was further analysed, which resulted in a larger effect size (1.08 for rTBFmax, 0.88 for rTBFmean). The results of the present study suggest that
Figure 4 The comparison of rTBF values between high-grade and low-grade gliomas. (a) rTBFmax; (b) rTBFmean. Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
e6
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
Table 2 Subgroup analysis for studies investigating perfusion values. Perfusion values
Subgroup
No. of studies
SMD
95% CI
p-Value
I2
aTBFmax aTBFmean rTBFmax
Publication year (2011) Field strength (3 T) Publication year (2010) Reference (GM)a Publication year (2011) Reference (M) b
3 3 3 3 3 3
0.88 0.57 1.16 1.10 1.18 1.12
0.39e1.37 0.09e1.04 0.90e1.41 0.83e1.37 0.60e1.76 0.25e1.99
0.0005 0.02 <0.00001 <0.0001 <0.0001 0.01
0% 6% 82% 89% 18% 42%
rTBFmean
aTBFmax, maximum of absolute tumour blood flow value; aTBFmean, average of absolute tumour blood flow value; rTBFmax, aTBFmax normalised to normal appearing tissue; rTBFmean, aTBFmean normalised to normal appearing tissue. a Reference (GM): contralateral grey matter as reference when rTBF calculated. b Reference (M): contralateral mirrored region as reference when rTBF evaluated.
rTBF may be a better index to grade gliomas, but there were too few studies to undertaken subgroup analysis to decide which internal reference was better. Another index of tumour perfusion is the relative signal intensity (SI), which is determined as a percentage of maximal SI within the tumour and contralateral normal brain tissue. Several studies32e34 have revealed that the relative SI may be useful in distinguishing between highand low-grade gliomas, but there were too few studies to extract effective data to include in the present metaanalysis. Moreover, the method used to define the regions of interest was shown to influence the perfusion values. The manual method is operator dependent and may be limited due to partial volume effects. The other method is histogram analyses of perfusion, which was first introduced by € demann et al.,35 but it is unclear which method is supeLu rior.30,36 In the present meta-analysis, all nine studies used the manual method. Moderate or high heterogeneity was observed for some of the perfusion parameters tested. One source of heterogeneity was sampling error, but this is impossible to control for. In the subgroup of publication year and field strength (3 T), heterogeneity decreased greatly. It is speculated that these factors may account for the heterogeneity among studies; however, the results should be interpreted with caution because only three studies were included. Several limitations of the present meta-analysis are worth noting. First, only nine studies were included in the meta-analysis and most of these studies have small sample sizes. There has been extensive perfusion research related to predicting the glioma grade, much of it using the DSC MRI technique.37 DSC is a robust technique that is in widespread use, while ASL is only just starting to be used outside of major research centres. With more extensive use of 3 T systems and new, improved ASL sequences, much of the previous limitations will be overcome and ASL will become a standard sequence in neuroradiology.38 Second, the presence of heterogeneity (design type, gliomas grades, and other clinical characteristics) affects the results of the present study. Different field strength (1.5 or 3 T), MRI manufacturers, data acquisition, and postprocessing techniques can lead to different results. Different studies use different approaches to measure tumour perfusion; in particular, the assessment and
placement of ROIs were subjective. The choice of nontumour tissue and the definition of tumour ROI varied between studies. In conclusion, despite the above-mentioned limitations, the present meta-analysis provides evidence that ASL has the ability to differentiate high- from low-grade gliomas; however, the results of the meta-analysis were drawn from studies with small samples. Future studies with larger sample sizes using the ASL technique are particularly needed.
References 1. Ostrom QT, Gittleman H, Fulop J, et al. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2008e2012. Neuro-oncology 2015;17:iv1e62. 2. Tian H, Gou Y, Pan Y, et al. Quality appraisal of clinical practice guidelines on glioma. Neurosurg Rev 2015;38:39e47. 3. Lev MH, Ozsunar Y, Henson JW, et al. Glial tumor grading and outcome prediction using dynamic spin-echo MR susceptibility mapping compared with conventional contrastenhanced MR: confounding effect of elevated rCBV of oligodendrogliomas. AJNR Am J Neuroradiol 2004;25:214e21. 4. Rees J. Advances in magnetic resonance imaging of brain tumours. Curr Opin Neurol 2003;16:643e50. 5. Ellika SK, Jain R, Patel SC, et al. Role of perfusion CT in glioma grading and comparison with conventional MR imaging features. AJNR Am J Neuroradiol 2007;28:1981e7. 6. Law M, Yang S, Wang H, et al. Glioma grading: sensitivity, specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. AJNR Am J Neuroradiol 2003;24:1989e98. 7. Pollock JM, Tan H, Kraft RA, et al. Arterial spin-labeled MR perfusion imaging: clinical applications. Magn Reson Imaging Clin N Am 2009;17:315e38. 8. Watts JM, Whitlow CT, Maldjian JA. Clinical applications of arterial spin labelling. NMR Biomed 2013;26:892e900. €rnum H, Steffensen EG, Knutsson L, et al. Perfusion MRI of brain tu9. Ja mours: a comparative study of pseudo-continuous arterial spin labelling and dynamic susceptibility contrast imaging. Neuroradiology 2010;52:307e17. 10. Hirai T, Kitajima M, Nakamura H, et al. Quantitative blood flow measurements in gliomas using arterial spin-labelling at 3T: intermodality agreement and inter-and intraobserver reproducibility study. AJNR Am J Neuroradiol 2011;32:2073e9. 11. Wolf RL, Wang J, Wang S, et al. Grading of CNS neoplasms using continuous arterial spin labeled perfusion MR imaging at 3 tesla. J Magn Reson Imaging 2005;22:475e82. 12. Chawla S, Wang S, Wolf RL, et al. Arterial spin-labelling and MR spectroscopy in the differentiation of gliomas. AJNR Am J Neuroradiol 2007;28:1683e9.
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7 € pf V, Preusser M, et al. Non-invasive assessment of 13. Furtner J, Scho intratumoral vascularity using arterial spin labelling: a comparison to susceptibility-weighted imaging for the differentiation of primary cerebral lymphoma and glioblastoma. Eur J Radiol 2014;83:806e10. 14. Warmuth C, Gunther M, Zimmer C. Quantification of blood flow in brain tumors: comparison of arterial spin labelling and dynamic susceptibility-weighted contrast-enhanced MR imaging. Radiology 2003;228:523e32. 15. Cebeci H, Aydin O, Ozturk-Isik E, et al. Assessment of perfusion in glial tumors with arterial spin labelling; comparison with dynamic susceptibility contrast method. Eur J Radiol 2014;83:1914e9. 16. Roy B, Awasthi R, Bindal A, et al. Comparative evaluation of 3-dimensional pseudocontinuous arterial spin labelling with dynamic contrastenhanced perfusion magnetic resonance imaging in grading of human glioma. J Comp Assist Tomogr 2013;37:321e6. 17. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009;151:264e9. 18. Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155:529e36. 19. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003;327:557e60. 20. Weber MA, Zoubaa S, Schlieter M, et al. Diagnostic performance of spectroscopic and perfusion MRI for distinction of brain tumors. Neurology 2006;66:1899e906. 21. Lehmann P, Monet P, De Marco G, et al. A comparative study of perfusion measurement in brain tumours at 3 Tesla MR: arterial spin labelling versus dynamic susceptibility contrast-enhanced MRI. Eur Neurol 2010;64:21e6. € ttenberg J, et al. Biopsy targeting gliomas: do 22. Weber MA, Henze M, Tu functional imaging techniques identify similar target areas? Invest Radiol 2010;45:755e68. 23. Yeom KW, Mitchell LA, Lober RM, et al. Arterial spin-labeled perfusion of pediatric brain tumors. AJNR Am J Neuroradiol 2014;35:395e401. 24. Lin Y, Li J, Zhang Z, et al. Comparison of intravoxel incoherent motion diffusion-weighted MR imaging and arterial spin labelling MR imaging in gliomas. BioMed Res Int 2015:2015. 25. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;407:249e57. 26. Roberts H, Roberts T, Brasch R, et al. Quantitative measurement of microvascular permeability in human brain tumors achieved using dynamic contrast enhanced MR imaging: correlation with histologic grade. AJNR Am J Neuroradiol 2000;21:891e9.
e7
€ demann L, Grieger W, Wurm R, et al. Comparison of dynamic contrast27. Lu enhanced MRI with WHO tumor grading for gliomas. Eur Radiol 2001;11:1231e41. 28. Yoo RE, Choi SH, Cho HR, et al. Tumor blood flow from arterial spin labelling perfusion MRI: a key parameter in distinguishing high-grade gliomas from primary cerebral lymphomas, and in predicting genetic biomarkers in high-grade gliomas. J Magn Reson Imaging 2013;38:852e60. 29. Yamashita K, Yoshiura T, Hiwatashi A, et al. Differentiating primary CNS lymphoma from glioblastoma multiforme: assessment using arterial spin labelling, diffusion-weighted imaging, and 18F-fluorodeoxyglucose positron emission tomography. Neuroradiology 2013;55:135e43. 30. Emblem KE, Nedregaard B, Nome T, et al. Glioma grading by using histogram analysis of blood volume heterogeneity from MR-derived cerebral blood volume maps. Radiology 2008;247:808e17. 31. Ulmer S, Helle M, Jansen O, et al. Intraoperative dynamic susceptibility contrast weighted magnetic resonance imaging (iDSC-MRI) d technical considerations and feasibility. Neuroimage 2009;45:38e43. 32. Noguchi T, Yoshiura T, Hiwatashi A, et al. Perfusion imaging of brain tumors using arterial spin-labelling: correlation with histopathologic vascular density. AJNR Am J Neuroradiol 2008;29:688e93. € pf V, Schewzow K, et al. Arterial spin-labelling assessment 33. Furtner J, Scho of normalized vascular intratumoral signal intensity as a predictor of histologic grade of astrocytic neoplasms. AJNR Am J Neuroradiol 2014;35:482e9. 34. Kim M, Kim HS, Kim JH, et al. Diagnostic accuracy and interobserver variability of pulsed arterial spin labelling for glioma grading. Acta Radiol 2008;49:450e7. € demann L, Warmuth C, Plotkin M, et al. Brain tumor perfusion: 35. Lu comparison of dynamic contrast enhanced magnetic resonance imaging using T1, T2, and T2* contrast, pulsed arterial spin labelling, and H2(15)O positron emission tomography. Eur J Radio 2009;70:465e74. 36. Young R, Babb J, Law M, et al. Comparison of region-of-interest analysis with three different histogram analysis methods in the determination of perfusion metrics in patients with brain gliomas. J Magn Reson Imaging 2007;26:1053e63. 37. Paulson ES, Schmainda KM. Comparison of dynamic susceptibilityweighted contrast-enhanced MR methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 2008;249:601e13. 38. Telischak NA, Detre JA, Zaharchuk G. Arterial spin labelling MRI: clinical applications in the brain. J Magn Reson Imaging 2015;41:1165e80.
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
Contents lists available at ScienceDirect
Clinical Radiology journal homepage: www.clinicalradiologyonline.net
A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade L. Kong a, c, H. Chen b, c, Y. Yang a, L. Chen a, * a b
Department of Anesthesiology, Anhui Provincial Cancer Hospital, Hefei 230031, China Department of Anesthesiology, Nanjing General Hospital of Nanjing Military Command, Nanjing 210002, China
art icl e i nformat ion Article history: Received 30 January 2016 Received in revised form 23 July 2016 Accepted 25 October 2016
AIM: To investigate the ability of arterial spin labelling (ASL) perfusion parameters to distinguish high-grade from low-grade gliomas. MATERIALS AND METHODS: The PubMed and EMBASE databases were systematically searched for relevant articles published up to September 2015. Studies that evaluated both high- and low-grade gliomas using ASL were included. The random effect model was used to calculate the standardised mean difference (SMD) of maximum mean absolute tumour blood flow values (aTBFmax, aTBFmean) and maximum mean relative tumour blood flow (rTBFmax, rTBFmean) between high- and low-grade gliomas. RESULTS: Nine studies encompassing 305 patients with high- and low-grade gliomas, met all inclusion and exclusion criteria and were included in the study. Compared with low-grade gliomas, high-grade gliomas had a significant increase in all ASL perfusion values: aTBFmax (SMD¼0.70, 95% confidence interval [CI]: 0.22e1.19, p¼0.0046); aTBFmean (SMD¼0.86, 95% CI: 0.2e1.52, p¼0.01); rTBFmax (SMD¼1.08, 95% CI: 0.54e1.63, p¼0.0001) and rTBFmean (SMD¼0.88, 95% CI: 0.35e1.4, p¼0.0011). CONCLUSIONS: The current study results indicate that tumour blood flow from ASL differs significantly with respect to the glioma grade. Despite some limitations, there is evidence that ASL may be useful to distinguish high- and low-grade gliomas. Further larger-scale studies are necessary to examine the utility of ASL to distinguish tumour grade. Ó 2016 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Introduction Gliomas are the most common primary brain tumours in adults,1 and a precise diagnosis is important because the adjuvant therapy after surgery and prognosis differ
* Guarantor and correspondent: L. Chen, Department of Anesthesiology, Anhui Provincial Tumor Hospital, No. 107, East Huanhu Road, Hefei, 230031, China. Tel.: þ86 551 65897820. E-mail address: [email protected] (L. Chen). c These authors contributed equally to this work.
considerably according to tumour grade.2 Conventional morphological imaging technologies are usually limited in grading gliomas. For example, enhancement is not a reliable factor for determining tumoural grade.3,4 Perfusion imaging is the commonly used advanced imaging method for the evaluation of brain tumours.5,6 Currently, two major types of imaging methods are available to evaluate brain tumour perfusion. One is dynamic susceptibility contrast (DSC) perfusion imaging and the other is arterial spin labelling (ASL). One practical advantage of ASL technique is that it relies on endogenous tracers, such as water molecules, and can be
http://dx.doi.org/10.1016/j.crad.2016.10.016 0009-9260/Ó 2016 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
e2
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
repeated many times in patients with renal failure.7,8 Several recent clinical studies have highlighted ASL techniques as helpful in the differential diagnosis of brain tumours; for example, low-grade gliomas from high-grade gliomas, and gliomas from other type of tumours.9e13 Clinically established methods for quantitative perfusion measurements usually include absolute tumour blood flow (aTBF) and relative tumour blood flow (rTBF), which is the normalisation of the aTBF values in relation to mean values in normal appearing regions. Some studies have investigated the value of TBF to distinguish high- from low-grade gliomas, but the findings have been incongruent. For example, Warmuth et al.14 demonstrated that mean (aTBFmean) and maximum aTBF (aTBFmax) values were significantly higher in high-grade than in low-grade brain tumours. Cebeci et al.15 reported significant positive correlations between rCBV from DSC and rCBF from ASL, and they considered relative values from ASL were more reliable. In contradistinction, Roy et al.16 demonstrated that ASL-derived absolute CBF values were not significantly different between histopathologically proven high- and low-grade glioma, even after normalising these values from the contralateral regions. They concluded that despite the advances in the technical developments of ASL, the currently available ASL method still suffers from interpatient variability.16 To the authors’ knowledge, the efficiency of ASL to distinguish between high-grade and low-grade gliomas has not been evaluated quantitatively. Therefore, the main objective of the present study was to conduct a quantitative meta-analysis of the existing literature to determine the statistical consensus of aTBF, and rTBF in distinguishing the tumour grade of gliomas.
grade IIIeIV) and low-grade gliomas (WHO gIeII); (2) histopathological analysis was used as the reference standard; (3) patients had no radiotherapy, surgery, or chemotherapy before ASL; (4) studies reported aTBF values and/or the rTBF values were available for effective calculation; (5) the number of patients should be at least eight; (6) no data overlapped between studies, if studies had the same or overlapping data, only the largest study was included in the final analysis. Studies were excluded based on the following criteria: animal studies, abstracts, reviews, case report, letters, editorials, comments, and conference proceedings. Articles that did not provide adequate information to allow the calculation of values were also excluded.
Materials and methods
rTBFmean¼ aTBFmean/reference value.
Search strategy
Reference regions may be normal-appearing grey matter, white matter, the contralateral mirrored region, or the global mean value. Data were recorded at the patient level, when possible. Each reviewer extracted study information on a standardised Microsoft Excel spreadsheet. Discrepancies were resolved by consensus. The same two reviewers independently assessed the quality of each article using the modified Quality Assessment of Diagnosis Accuracy Studies (QUADAS-2) score tool,18 which consisted of 14 questions answered ‘‘yes”, ‘‘no”, or ‘‘unclear”. Disagreements between the two investigators were resolved by consensus, and if disagreement persisted, a third reviewer made the ultimate decision.
This meta-analysis followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement.17 A systematic search was performed in the PubMed and EMBASE for relevant publications from inception to September 2015. The search terms were as follows: (“arterial spin labelling” or “arterial spin labelling” or “arterial spin labelled” or “ASL”) AND (“glioma” or “brain neoplasm” or “brain tumour” or “brain tumour”). There was no language restriction. Additionally, the references of selective articles were searched manually to identify potentially relevant studies that were not identified in the previous searches. Two authors independently performed the search, reviewed all eligible articles, and excluded obviously irrelevant studies by reading titles, abstracts, and keywords. The full text of each article was obtained when one or both reviewers were unsure or recommended review of the full text of the article.
Inclusion and excluded criteria The inclusion criteria were as follows: (1) ASL was used to measure perfusion values of both high-grade (WHO
Data extraction and quality assessment The final articles were assessed independently by the same two authors. For each included study basal characteristics (authors, year of publication, and country of origin), patient characteristics (mean age, sex, and number and type of gliomas), and technical aspects (imaging field strength, ASL techniques, reference standard, and the method for TBF measurement) were noted. The aTBFmean, aTBFmax, mean rTBF (rTBFmean), and maximum rTBF (rTBFmax) values were tabulated as mean values and SDs. aTBFmax was obtained by placing a region of interest (ROI) in the apparent maximum TBF region and average several ROIs within the tumour to create the aTBFmean. rTBF values were evaluated by normalising aTBF values to reference regions: rTBFmax ¼ aTBFmax/reference value and
Statistical analysis The mean difference (MD) of aTBFmean and aTBFmax and the standardised mean differences (SMDs) of aTBFmean, aTBFmax, rTBFmean, and rTBFmax, between high-grade and low-grade gliomas were calculated and used as effect-size statistics. The SMD is the standardised difference between two means and can be calculated as the difference between the high- and low-grade glioma groups divided by the
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
pooled standard deviation (SD). The overall effect size was presented with the mean and 95% confidence interval (CI). The extent of heterogeneity was assessed by the chisquared value test and the inconsistency index (I2). Low, moderate, and high I2 values were considered to be 25%, 50%, and 75%, respectively.19 Publication bias was assessed by the funnel plot asymmetry test. Subgroup analysis was performed using a more homogeneous set of studies with similar variables. Subgroups were only constructed when at least three studies could be included. The abovementioned statistical analysis was performed with R (version 3.2, http://www.r-project.org) and the metafor package (http://cran.r-project.org/web/packages/metafor/ index.html) was used to implement a random-effects model.
Results Literature identification This search produced 90 articles from PubMed and 179 articles from EMBASE. A total of 236 articles remained after using EndNote citation manager (Thomson Reuters, New York, NY, USA) to remove duplicates. After screening titles and abstracts, 20 articles were considered potentially eligible and were retrieved in full text for further assessment. After a full-text review, nine studies were considered eligible for this review.11,12,14,16,20e24 The selection process is demonstrated in the flowchart in Fig 1.
e3
Study characteristics and methodological quality assessment Nine studies, comprising 197 patients with high-grade gliomas and 108 patients with low-grade gliomas, met all the inclusion and exclusion criteria and were included in the present meta-analysis. The characteristics of the included studies are presented in Table 1. Among them, six studies were prospective studies, and three were retrospective studies, with sample sizes ranging from 9 to 64. Four studies used both a TBF and rTBF values, four studies used only rTBF; one study used only aTBF. Seven studies compared ASL with other imaging techniques, such as, magnetic resonance spectroscopy (MRS), dynamic susceptibility-weighted MRI (DSC), dynamic contrastenhanced MRI (DCE), diffusion tensor imaging (DTI). Two studies used ASL to distinguish gliomas from metastases, central nervous system (CNS) lymphomas, or meningioma. The revised ‘‘QUADAS-2” instrument was used to evaluate each included study. Of 14 questions, no study met all of them. Most studies were of high (moderate) quality, which satisfied the majority of standards. All nine studies had histopathology (biopsy or surgical resection) as the sole reference standard, but risks of bias in “flow and timing” were generally high. The assessment of study quality by QUADAS-2 is graphically summarised in Fig 2.
Quantitative analysis aTBFmax and aTBFmean Of the nine included studies, five studies used aTBF11,12,14,16,24 The pooled data revealed a significant inmax. crease of aTBFmax in high-grade gliomas compared with low-grade gliomas (MD¼45.08 ml/min/100 g, 95% CI: 22.31e67.85, p¼0.002; SMD¼0.70, 95% CI: 0.22-1.19, p¼0.0046), with a high level of heterogeneity (I2¼65%, p¼0.04 for MD and I2¼50%, p¼0.08 for SMD; Fig 3a,c). Four studies11,12,14,24 reported values of aTBFmean. The pooled data showed that aTBFmean was significantly higher in high-grade gliomas than low-grade gliomas (MD¼29.18 ml/min/100 g, 95% CI: 2.04e56.33, p¼0.02; SMD¼0.86, 95% CI: 0.2e1.52, p¼0.01), with a high level of heterogeneity (I2¼84%, p¼0.0004 for MD and I2¼56%, p¼0.08 for SMD; Fig 3b,d).
Figure 1 Flow diagram of the study selection process.
rTBFmax and rTBFmean Five studies11,16,20,22,23 used the rTBFmax. The rTBFmax in high-grade gliomas increased significantly compared to low-grade gliomas (SMD¼1.08, 95% CI: 0.54e1.63, p¼0.0001) with high heterogeneity (I2¼61%, p¼0.03; Fig 4a). Five studies11,12,14,20,21 used the rTBFmean. The pooled data revealed that rTBFmean in high-grade gliomas was higher significantly than low-grade gliomas (SMD¼0.88, 95% CI: 0.35e1.40, p¼0.0011; Fig 4b); heterogeneity was moderate (I2¼39%, p¼0.15).
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
e4
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
Table 1 Characteristics of included studies for meta-analysis. Author
Year
No. of patients (n)
M/F
Mean age (yr) (range)
Study type
Histology
FS (T)
ASL technique
Parameters
Warmuth Wolf Weber Chawla Lehmann Weber Roy Yeom Lin
2003 2005 2006 2007 2010 2010 2013 2014 2015
18 26 54 35 9 57 64 18 24
9/9 18/8 N 21/14 8/1 34/23 N N 19/5
38.2(9e64) 47(20e66) N 45(20e68) 53(19e82) N N N 42(16e65)
P P P R P P P R R
LGG(9);HGG(9) LGG(7); HGG(19) LGG(9) vs. HGG(46) LGG(13);HGG(22) LGG(5) vs. HGG(4) LGG(15) vs. HGG(42) LGG(26); HGG(38) LGG(14);HGG(4) LGG(11); HGG(13)
1.5 3 1.5 3 3 1.5 3 3 3
pASL cASL pASL cASL pcASL pASL pcASL pcASL pcASL
aTBFmean; aTBFmax; rTBFmean aTBFmean; aTBFmax; rTBFmean; rTBFmax rTBFmean; rTBFmax aTBFmax; aTBFmean; rTBFmean rTBFmean rTBFmax aTBFmax; rTBFmax rTBFmax aTBFmean; aTBFmax
HGG, high-grade glioma; LGG, low-grade glioma; FS, Field strength; R, retrospective; P, prospective; aTBFmean, average of absolute tumour blood flow value; aTBFmax, maximum of absolute tumour blood flow value; rTBFmean, aTBFmean normalised to normal appearing tissue; rTBFmax, aTBFmax normalised to normal appearing tissue; cASL, continuous arterial spin labelling; pASL, pulsed arterial spin labelling; pcASL, pseudo-continuous ASL; N, not reported.
Sensitivity analysis and publication bias In the sensitivity analysis, no significant change in the results was evident when excluding studies one by one for all perfusion values. The funnel plot was found to be symmetrical, suggesting a low likelihood of publication bias.
Subgroup analyses In stratified analyses, high-grade gliomas exhibited a significant increase in all perfusion parameters compared with low-grade gliomas in subgroups of publication year (before 2011), field strength (3 T), reference (grey matter), and reference (global). Subgroup analyses were only performed when at least three studies could be included. These results are shown in Table 2.
Discussion Accurate tumour grading is essential to determine the choice of therapeutic approach and to the assessment of prognosis.2,25 Conventional MRI techniques are often unreliable on tumour grading. With the advent of advanced imaging techniques, the ability to accurately grade tumours has been a subject of interest. Previous studies26,27 have demonstrated that information about vessel permeability may contribute to the grading of gliomas. Perfusion MRI is one of the most effective methods for quantifying the glioma grading; however, current reports mostly contain data from small samples and the results are inconsistent. Thus,
the present meta-analysis was performed with the aim of resolving the incongruities by increasing sample size and improving the accuracy of evaluation of effect size. In the present meta-analysis, the utility of ASL perfusion values to distinguish high-grade gliomas from low-grade was examined. Pooled analysis revealed that high-grade gliomas had a significant increase in all measured values relative to low-grade gliomas. These findings remained largely unchanged when sensitivity analysis and allsubgroup analyses was performed, suggesting that these results were very unlikely due to chance. To the authors’ knowledge, this is the first meta-analysis to assess the overall values of ASL perfusion values in glioma grading. The aTBF is widely used in the assessment of glioma grading. It allows comparison of patients with each other or comparison of values in individual patients during the course of treatment; however, Warmuth et al.14 concluded that aTBF is not a crucial factor for tumour grading. The present results indicated that the TBF (maximum and mean) value was significantly higher in high-grade gliomas than in low-grade gliomas, suggesting that TBF is a potential indicator of malignancy in gliomas. To correct for age-dependent and patient-dependent variations, rTBF is calculated by using absolute perfusion values normalised to the normal-appearing tissue. Several studies revealed that rTBF is useful for characterising gliomas and can be used to discriminate glioblastomas and other lesions.28,29 Previous studies have used contralateral white matter,30 grey matter,31,32 or the cerebellum23 as the €rnum et al.9 believed that white internal references. Ja
Figure 2 Quality assessment with revised QUADAS-2 tool. Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
e5
Figure 3 The comparison of aTBF values between high-grade and low-grade gliomas. (a) SMD of aTBFmax; (b) SMD of aTBFmean; (c) MD of aTBFmax; (d) MD of aTBFmean.
matter perfusion measurements with ASL may have disadvantages and cerebellar perfusion usually is unaffected by pathology in the brain. Among nine studies included in this meta-analysis, three studies used grey matter as the internal references, two studies used white matter, three
studies used contralateral mirrored tissue, and one used the global averaged value. The difference between the rTBF between high- and low-grade gliomas was further analysed, which resulted in a larger effect size (1.08 for rTBFmax, 0.88 for rTBFmean). The results of the present study suggest that
Figure 4 The comparison of rTBF values between high-grade and low-grade gliomas. (a) rTBFmax; (b) rTBFmean. Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
e6
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7
Table 2 Subgroup analysis for studies investigating perfusion values. Perfusion values
Subgroup
No. of studies
SMD
95% CI
p-Value
I2
aTBFmax aTBFmean rTBFmax
Publication year (2011) Field strength (3 T) Publication year (2010) Reference (GM)a Publication year (2011) Reference (M) b
3 3 3 3 3 3
0.88 0.57 1.16 1.10 1.18 1.12
0.39e1.37 0.09e1.04 0.90e1.41 0.83e1.37 0.60e1.76 0.25e1.99
0.0005 0.02 <0.00001 <0.0001 <0.0001 0.01
0% 6% 82% 89% 18% 42%
rTBFmean
aTBFmax, maximum of absolute tumour blood flow value; aTBFmean, average of absolute tumour blood flow value; rTBFmax, aTBFmax normalised to normal appearing tissue; rTBFmean, aTBFmean normalised to normal appearing tissue. a Reference (GM): contralateral grey matter as reference when rTBF calculated. b Reference (M): contralateral mirrored region as reference when rTBF evaluated.
rTBF may be a better index to grade gliomas, but there were too few studies to undertaken subgroup analysis to decide which internal reference was better. Another index of tumour perfusion is the relative signal intensity (SI), which is determined as a percentage of maximal SI within the tumour and contralateral normal brain tissue. Several studies32e34 have revealed that the relative SI may be useful in distinguishing between highand low-grade gliomas, but there were too few studies to extract effective data to include in the present metaanalysis. Moreover, the method used to define the regions of interest was shown to influence the perfusion values. The manual method is operator dependent and may be limited due to partial volume effects. The other method is histogram analyses of perfusion, which was first introduced by € demann et al.,35 but it is unclear which method is supeLu rior.30,36 In the present meta-analysis, all nine studies used the manual method. Moderate or high heterogeneity was observed for some of the perfusion parameters tested. One source of heterogeneity was sampling error, but this is impossible to control for. In the subgroup of publication year and field strength (3 T), heterogeneity decreased greatly. It is speculated that these factors may account for the heterogeneity among studies; however, the results should be interpreted with caution because only three studies were included. Several limitations of the present meta-analysis are worth noting. First, only nine studies were included in the meta-analysis and most of these studies have small sample sizes. There has been extensive perfusion research related to predicting the glioma grade, much of it using the DSC MRI technique.37 DSC is a robust technique that is in widespread use, while ASL is only just starting to be used outside of major research centres. With more extensive use of 3 T systems and new, improved ASL sequences, much of the previous limitations will be overcome and ASL will become a standard sequence in neuroradiology.38 Second, the presence of heterogeneity (design type, gliomas grades, and other clinical characteristics) affects the results of the present study. Different field strength (1.5 or 3 T), MRI manufacturers, data acquisition, and postprocessing techniques can lead to different results. Different studies use different approaches to measure tumour perfusion; in particular, the assessment and
placement of ROIs were subjective. The choice of nontumour tissue and the definition of tumour ROI varied between studies. In conclusion, despite the above-mentioned limitations, the present meta-analysis provides evidence that ASL has the ability to differentiate high- from low-grade gliomas; however, the results of the meta-analysis were drawn from studies with small samples. Future studies with larger sample sizes using the ASL technique are particularly needed.
References 1. Ostrom QT, Gittleman H, Fulop J, et al. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2008e2012. Neuro-oncology 2015;17:iv1e62. 2. Tian H, Gou Y, Pan Y, et al. Quality appraisal of clinical practice guidelines on glioma. Neurosurg Rev 2015;38:39e47. 3. Lev MH, Ozsunar Y, Henson JW, et al. Glial tumor grading and outcome prediction using dynamic spin-echo MR susceptibility mapping compared with conventional contrastenhanced MR: confounding effect of elevated rCBV of oligodendrogliomas. AJNR Am J Neuroradiol 2004;25:214e21. 4. Rees J. Advances in magnetic resonance imaging of brain tumours. Curr Opin Neurol 2003;16:643e50. 5. Ellika SK, Jain R, Patel SC, et al. Role of perfusion CT in glioma grading and comparison with conventional MR imaging features. AJNR Am J Neuroradiol 2007;28:1981e7. 6. Law M, Yang S, Wang H, et al. Glioma grading: sensitivity, specificity, and predictive values of perfusion MR imaging and proton MR spectroscopic imaging compared with conventional MR imaging. AJNR Am J Neuroradiol 2003;24:1989e98. 7. Pollock JM, Tan H, Kraft RA, et al. Arterial spin-labeled MR perfusion imaging: clinical applications. Magn Reson Imaging Clin N Am 2009;17:315e38. 8. Watts JM, Whitlow CT, Maldjian JA. Clinical applications of arterial spin labelling. NMR Biomed 2013;26:892e900. €rnum H, Steffensen EG, Knutsson L, et al. Perfusion MRI of brain tu9. Ja mours: a comparative study of pseudo-continuous arterial spin labelling and dynamic susceptibility contrast imaging. Neuroradiology 2010;52:307e17. 10. Hirai T, Kitajima M, Nakamura H, et al. Quantitative blood flow measurements in gliomas using arterial spin-labelling at 3T: intermodality agreement and inter-and intraobserver reproducibility study. AJNR Am J Neuroradiol 2011;32:2073e9. 11. Wolf RL, Wang J, Wang S, et al. Grading of CNS neoplasms using continuous arterial spin labeled perfusion MR imaging at 3 tesla. J Magn Reson Imaging 2005;22:475e82. 12. Chawla S, Wang S, Wolf RL, et al. Arterial spin-labelling and MR spectroscopy in the differentiation of gliomas. AJNR Am J Neuroradiol 2007;28:1683e9.
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016
L. Kong et al. / Clinical Radiology xxx (2016) e1ee7 € pf V, Preusser M, et al. Non-invasive assessment of 13. Furtner J, Scho intratumoral vascularity using arterial spin labelling: a comparison to susceptibility-weighted imaging for the differentiation of primary cerebral lymphoma and glioblastoma. Eur J Radiol 2014;83:806e10. 14. Warmuth C, Gunther M, Zimmer C. Quantification of blood flow in brain tumors: comparison of arterial spin labelling and dynamic susceptibility-weighted contrast-enhanced MR imaging. Radiology 2003;228:523e32. 15. Cebeci H, Aydin O, Ozturk-Isik E, et al. Assessment of perfusion in glial tumors with arterial spin labelling; comparison with dynamic susceptibility contrast method. Eur J Radiol 2014;83:1914e9. 16. Roy B, Awasthi R, Bindal A, et al. Comparative evaluation of 3-dimensional pseudocontinuous arterial spin labelling with dynamic contrastenhanced perfusion magnetic resonance imaging in grading of human glioma. J Comp Assist Tomogr 2013;37:321e6. 17. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 2009;151:264e9. 18. Whiting PF, Rutjes AW, Westwood ME, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155:529e36. 19. Higgins JP, Thompson SG, Deeks JJ, et al. Measuring inconsistency in meta-analyses. BMJ 2003;327:557e60. 20. Weber MA, Zoubaa S, Schlieter M, et al. Diagnostic performance of spectroscopic and perfusion MRI for distinction of brain tumors. Neurology 2006;66:1899e906. 21. Lehmann P, Monet P, De Marco G, et al. A comparative study of perfusion measurement in brain tumours at 3 Tesla MR: arterial spin labelling versus dynamic susceptibility contrast-enhanced MRI. Eur Neurol 2010;64:21e6. € ttenberg J, et al. Biopsy targeting gliomas: do 22. Weber MA, Henze M, Tu functional imaging techniques identify similar target areas? Invest Radiol 2010;45:755e68. 23. Yeom KW, Mitchell LA, Lober RM, et al. Arterial spin-labeled perfusion of pediatric brain tumors. AJNR Am J Neuroradiol 2014;35:395e401. 24. Lin Y, Li J, Zhang Z, et al. Comparison of intravoxel incoherent motion diffusion-weighted MR imaging and arterial spin labelling MR imaging in gliomas. BioMed Res Int 2015:2015. 25. Carmeliet P, Jain RK. Angiogenesis in cancer and other diseases. Nature 2000;407:249e57. 26. Roberts H, Roberts T, Brasch R, et al. Quantitative measurement of microvascular permeability in human brain tumors achieved using dynamic contrast enhanced MR imaging: correlation with histologic grade. AJNR Am J Neuroradiol 2000;21:891e9.
e7
€ demann L, Grieger W, Wurm R, et al. Comparison of dynamic contrast27. Lu enhanced MRI with WHO tumor grading for gliomas. Eur Radiol 2001;11:1231e41. 28. Yoo RE, Choi SH, Cho HR, et al. Tumor blood flow from arterial spin labelling perfusion MRI: a key parameter in distinguishing high-grade gliomas from primary cerebral lymphomas, and in predicting genetic biomarkers in high-grade gliomas. J Magn Reson Imaging 2013;38:852e60. 29. Yamashita K, Yoshiura T, Hiwatashi A, et al. Differentiating primary CNS lymphoma from glioblastoma multiforme: assessment using arterial spin labelling, diffusion-weighted imaging, and 18F-fluorodeoxyglucose positron emission tomography. Neuroradiology 2013;55:135e43. 30. Emblem KE, Nedregaard B, Nome T, et al. Glioma grading by using histogram analysis of blood volume heterogeneity from MR-derived cerebral blood volume maps. Radiology 2008;247:808e17. 31. Ulmer S, Helle M, Jansen O, et al. Intraoperative dynamic susceptibility contrast weighted magnetic resonance imaging (iDSC-MRI) d technical considerations and feasibility. Neuroimage 2009;45:38e43. 32. Noguchi T, Yoshiura T, Hiwatashi A, et al. Perfusion imaging of brain tumors using arterial spin-labelling: correlation with histopathologic vascular density. AJNR Am J Neuroradiol 2008;29:688e93. € pf V, Schewzow K, et al. Arterial spin-labelling assessment 33. Furtner J, Scho of normalized vascular intratumoral signal intensity as a predictor of histologic grade of astrocytic neoplasms. AJNR Am J Neuroradiol 2014;35:482e9. 34. Kim M, Kim HS, Kim JH, et al. Diagnostic accuracy and interobserver variability of pulsed arterial spin labelling for glioma grading. Acta Radiol 2008;49:450e7. € demann L, Warmuth C, Plotkin M, et al. Brain tumor perfusion: 35. Lu comparison of dynamic contrast enhanced magnetic resonance imaging using T1, T2, and T2* contrast, pulsed arterial spin labelling, and H2(15)O positron emission tomography. Eur J Radio 2009;70:465e74. 36. Young R, Babb J, Law M, et al. Comparison of region-of-interest analysis with three different histogram analysis methods in the determination of perfusion metrics in patients with brain gliomas. J Magn Reson Imaging 2007;26:1053e63. 37. Paulson ES, Schmainda KM. Comparison of dynamic susceptibilityweighted contrast-enhanced MR methods: recommendations for measuring relative cerebral blood volume in brain tumors. Radiology 2008;249:601e13. 38. Telischak NA, Detre JA, Zaharchuk G. Arterial spin labelling MRI: clinical applications in the brain. J Magn Reson Imaging 2015;41:1165e80.
Please cite this article in press as: Kong L, et al., A meta-analysis of arterial spin labelling perfusion values for the prediction of glioma grade, Clinical Radiology (2016), http://dx.doi.org/10.1016/j.crad.2016.10.016