Skip to main content
NCBI home page
Medicine logoLink to Medicine
. 2019 May 13;98(19):e15580. doi: 10.1097/MD.0000000000015580

Based on arterial spin labeling helps to differentiate high-grade gliomas from brain solitary metastasis

A systematic review and meta-analysis

Min Fu a,∗, Fang Han b, Changchao Feng a, Tao Chen c, Xiaobo Feng d
Editor: Lu Wang
PMCID: PMC6531209  PMID: 31083237

Abstract

Background:

We first introduced this meta-analysis was to assess the accuracy of arterial spin labeling (ASL) in the differentiating high-grade gliomas (HGG) from brain solitary metastases (BSM).

Methods:

The PubMed, Web of Knowledge, and Cochrane Libraries and China National Knowledge Infrastructure databases were searched up to August 31, 2018. The pooled weighted sensitivity and specificity, summary receiver operating characteristic curve (SROC), sensitivity analysis, and threshold effect analysis were performed on Stata version 12.0 and Meta-Disc version 1.4. Deeks’ funnel plot asymmetry test was performed to assess publication bias.

Results:

Of 5 eligible articles, of the 346 lesions from 346 patients, 274 were HGG, and 72 were BSM. The forest pooled sensitivity of 0.88 (95% confidence interval [CI]: 0.65, 0.96) and specificity of 0.85 (95% CI: 0.74, 0.92) of ASL were reported in this meta-analysis. The pooled area under the curve of SROC was 0.92 (95% CI: 0.89, 0.94). Sensitivity analysis demonstrated that the pooled estimates were reliable. No evident publication bias was obtained (P = .38).

Conclusion:

The parameters derived from ASL with high accuracy in differentiating HGG from BSM. However, results must be interpreted with caution due to the small sample size considered. Large sample prospective studies were necessary to assess and confirm its clinical value.

Keywords: ASL, gliomas, meta-analysis, metastasis, MRI

1. Introduction

High-grade gliomas (HGG) and metastases are the most common malignant tumors in the brain. Preoperative distinction between HGG and metastases was of vital importance as the clinical course, treatment and prognosis of HGG and metastatic tumor are totally different.[1] The precise diagnosis and subsequently proper treatment significantly affected patient's survival. The main problem was that differentiation between HGG and metastases on conventional MR sequences, which may show as solitary strongly enhancing brain tumors surrounded by a T2-hyperintense edema.[2] Different biological behaviors and pathological characteristics were helpful to differentiating HGG from solitary metastases.[3] The growth of HGG is invasive, HGG infiltrated the surrounding tissues through the nerve fiber bundles, and therefore the peritumoral area was accompanied by a large number of tumor cells.[4] Yet, metastases were transformed through the bloodstream to the brain parenchyma by extracerebral tumors.[5] Compression of the venous veins was attributed to peritumoral edema. Therefore, the peritumoral of the brain metastatic was presented as vasogenic edema and absence of tumor cells.[6]

An advanced magnetic resonance imaging (MRI) technique such as arterial spin labeling (ASL) perfusion MRI)provides physiologic and hemodynamic information related to tumoral vascularity.[7] ASL was a promising non-contrast-enhanced imaging technique for evaluating brain tumors, which was now well established in clinical. ASL was widely used in brain tumors diagnosis and grading.[8–10] Several studies have reported the tumor blood flow derived from ASL was significantly higher in HGG than in LGG potential higher sensitivity relative to conventional imaging techniques.[10,11] It provided high-resolution morphological information regarding the lesions. ASL had also demonstrated diagnostic value in differentiating HGG from brain solitary metastases.[12]

Several studies have reported that related cerebral blood flow (rCBF) in HGG was significantly higher than in metastasis (P < .05),[13–15] While previous reports demonstrated that rCBF values may not be sufficient for discriminating metastatic tumors from HGG.[16,17] Meta-analysis has been recognized as an effective method to answer a wide variety of clinical questions by summarizing and reviewing previously published quantitative research.[18] So far, the contribution of ASL on the discrimination of HGG from brain solitary metastases based on meta-analysis has been unpublished. We first introduced this meta-analysis was to assess the accuracy of ASL in the differentiating HGG from brain solitary metastases.

2. Methods

2.1. Search strategy

The PubMed, Web of Knowledge, and Cochrane Libraries and China National Knowledge Infrastructure (CNKI) databases were searched using the following terms: “arterial spin labeling or ASL” and (brain metastasis) OR (brain metastases) OR (metastatic brain tumor∗) OR (cerebral metastasis) OR (cerebral metastases) OR solitary metasta∗)) AND (glioma) OR (glioblastoma). English and Chinese language publications were limited to our search. This study was also limited to human studies. In addition, we reviewed references of relevant guidelines, reviews, and meta-analyses identified in PubMed.

2.2. Eligibility criteria and data extraction

Studies were required to meet the following inclusion criteria:

  • (1)

    population: patients had enhancing brain lesion;

  • (2)

    ASL was applied to differentiate HGG from brain metastasis;

  • (3)

    reference standard: histopathological diagnosis and clinical follow-up were considered as reference standard;

  • (4)

    the eligible studies must have sufficient data to reconstruct 2 × 2 tables for sensitivity and specificity.

Conference abstracts, reviews, letters, case reports, animal studies, editorials, and short surveys that did not provide sufficient information for reconstruction of 2 × 2 tables were excluded.

2.3. Quality assessment and data extraction

Study quality was assessed based on the 14 items of quality assessment of diagnostic accuracy studies by 2 reviewers independently.[19] Two authors independently worked on extracting the following data: baseline characteristics of study (first author, year of publication, study design, number, and age of patients); MRI parameters (ASL technical, field strength, diagnostic threshold for differentiating HGG from brain metastasis). Additionally, reference standard (histopathology, follow-up) data were also collected. Finally, the values of true-positive (TP), false positive (FP), true negative (TN), and false negative (FN) were calculated. Any controversy existed in this study; a third reviewer assessed all involved problems. Consensus on all disagreements was a prerequisite for data analysis.

2.4. Statistical analysis

The pooled weighted sensitivity and specificity, summary receiver operating characteristic curve, sensitivity analysis, and threshold effect analysis were performed on Stata version 12.0 (College Station, TX) and Meta-Disc version 1.4 (Unit of Clinical Biostatistics, Madrid, Spain). The inconsistency index (I-squared, I2) was done to evaluate the possibility of heterogeneity among individual studies. If I2 > 50%, it indicated that heterogeneity exists in this study.[20] Publication bias analysis was assessed by using the Deeks’ funnel plot test. P < .05 was recognized as statistical significance.

3. Results

The search process was summarized in Figure 1. Based on the results of computer search and manual cross-checking of reference lists, 47 citations (32 English and 15 Chinese) were retrieved. A total of 21 articles remained by using Endnote citation manager to remove duplicate articles. After the initial evaluation, we found 21 eligible articles (15 English and 6 Chinese). After reading the full texts and abstracts, we excluded 12 of the 17 relevant articles for the following reasons: the aim of studies was to estimate the diagnostic performance of ASL of brain tumor (n = 4); Review of clinical application of ASL in the diagnosis of central nervous system diseases (n = 3); Among the various parameters, the peritumoral relative cerebral blood volume as the measurement parameter was used in all studies. Articles did not provide adequate data that could be used to calculate TP, FP, TN, and FN values of rCBF (n = 5); finally, 5 articles (4 English and 1 Chinese) met all inclusion and selected for data analysis.[13–17]

Figure 1.

Figure 1

Open in a new tab

Flow chart of studies selection in this meta-analysis.

The basic study characteristics for eligible studies were presented in Table 1. In total, 346 patients were reviewed from the 5 papers. Of the 346 lesions from 346 patients, 274 were HGG, and 72 were brain metastasis tumors. The mean age of the patients was variable ranging from 50 to 63.7 years. In 5 studies, 2 articles were prospective, 3 were retrospective, and the sample size ranged from 38 to 128. Four studies placed the region of interest (ROI) in the surrounding area of the tumor and 1 in the tumoral area. Four studies reported using 3.0-T MRI field strength, while only 1 studies adopted 1.5 T MRI systems. Reference standard was described in all studies with histopathology.

Table 1.

Characteristics of studies included in the meta-analysis.

3.

Open in a new tab

The forest plots of sensitivities and specificities among included studies were shown in Figure 2. The combined sensitivity of 0.88 (95% confidence interval [CI]: 0.65, 0.96) and specificity of 0.85 (95% CI: 0.74, 0.92) were reported for the classification of HGG and brain metastasis by ASL in this meta-analysis. The pooled area under the curve (AUC) was 0.92 (95% CI: 0.89, 0.94). Our meta-analysis demonstrated that ASL appears to be a more promising in differentiating HGG from a solitary metastasis. A pooled homogeneity sensitivity of I2 = 83.31% and specificity of I2 = 71.6%, respectively. Meta-regression analysis was performed to explore the sources of heterogeneity among the included articles. The covariates included country, field strength, ROI placement, study design, percentage of HGG among eligible studies. However, no significant difference was found statistically (P > .05).

Figure 2.

Figure 2

Open in a new tab

Forest map of sensitivity and specificity of ASL. ASL = arterial spin labeling.

3.1. Sensitivity analysis and publication bias

Sensitivity analysis was constructed to determine whether to revise of the inclusion criteria affected the outcomes finally.[21] We evaluated the effect of each study on the pooled results by excluding single study. The pooled data were analyzed again when each study was excluded. The result indicated that there were no significant changes of the pooled weighted sensitivity and specificity, which validated the reliability and rationality of our analysis. Deeks’ funnel plot asymmetry test was performed to assess publication bias. Finally, no evident publication bias was obtained through the Figure 3 (P = .38).

Figure 3.

Figure 3

Open in a new tab

Publication bias analyzed by Deek funnel plot.

4. Discussion

In this study, we analyzed the diagnostic value of ASL in the classification of HGG and brain metastasis using a meta-analysis to obtain a powerful conclusion. Systematic reviews and meta-analyses aim to collect and analyze individual data related a research question and therefore were considered as a gold standard reach to evidence synthesis. An overall combined sensitivity of 0.88 (95% CI: 0.65, 0.96) and specificity of 0.85 (95% CI: 0.74, 0.92) were calculated in our paper. The pooled AUC was shown in Figure 4 with 0.92 (95% CI: 0.89, 0.94). Our results demonstrated that rCBF values from ASL would be a powerful tool in the classification of HGG and brain metastasis.

Figure 4.

Figure 4

Open in a new tab

SROC of ASL in characterization of HGG and Solitary Metastatic Brain Tumor. ASL = arterial spin labelling, HGG = high grade gliomas, SROC = summary receiver operating characteristic curve.

These outcomes were well known in the studies have been reported.[22] As expected, patients with HGG showed significantly higher rCBF than those with metastasis in peritumoral areas. Additionally, a recent animal study has reported that the peritumoral edema of glioma not only included tumor cells, but also contained astrocytic swelling and microglial activation; our study demonstrated that rCBF values from ASL with high accuracy in differentiating HGG from brain metastases, which could be explained by the histopathologic findings. Similar finding was also reported in perfusion studies in the classification of GBM and brain metastasis tumors using ASL,[13] of 128 consecutive patients who were diagnosed as glioblastoma (n = 89) and metastasis tumor (n = 39) and underwent operation, the rCBF derived from ASL showed high diagnostic performance with a similar specificity of 84.6%, which was consistent with our findings. Although a lower sensitivity (42.7%) of rCBF was presented, possibly due to small sample size.

Notable heterogeneities were shown in Figure 2. In Figure 4, “shoulder-arm” shape did not seem to be displayed among individual studies, indicating that threshold effect was not a concern. The next step was done to explore the threshold effect in this study. In this meta-analysis, the Spearman correlation coefficient was 0.229 (P = .208), indicating that the heterogeneity was unlikely to be attributable to the threshold effect among eligible articles. We performed meta-regression analysis to understand the potential other factors significantly affected heterogeneity among individual studies. However, no significance was found statistically using other covariates, including country, field strength, study design, percentage of HGG (P > .05).

These findings seemed to be encouraging, confirming the high diagnostic power of ASL in differentiating gliomas from solitary metastases, several differences and limitations in our meta-analysis should be noted. One limitation of our study was the sample size of included studies; however, articles were uncontrolled and there was no publication bias by Deeks’ asymmetry test. Another limitation of was only English and Chinese language publications were searched in PubMed, Web of Knowledge, and Cochrane Libraries, and CNKI databases. Besides, the reliability of a meta-analysis relied on the quality of study heterogeneity; we did not perform subgroup analysis because there were only 5 studies. We performed meta-regression analysis to explore the potential other factors significantly affected heterogeneity among individual studies, yet no evidence was shown to be a significant factor affecting study heterogeneity.

5. Conclusion

To the best of our knowledge, this was the first meta-analysis providing comprehensive insights into the classification of HGG and brain metastasis tumors using ASL. In our study of the rCBF parameters derived from ASL with high accuracy in differentiating HGG from brain metastases. Our study suggested that the ASL was useful for discriminating brain metastatic tumors from HGG. However, results must be interpreted with caution due to the small sample size considered. Large sample prospective studies were necessary to assess and confirm its clinical value.

Author contributions

Conceptualization: Min Fu.

Data curation: Fang Han, Tao Chen, Xiaobo Feng.

Formal analysis: Tao Chen.

Funding acquisition: Min Fu.

Methodology: Changchao Feng, Tao Chen.

Resources: Changchao Feng.

Software: Fang Han, Xiaobo Feng.

Writing – original draft: Min Fu.

Writing – review and editing: Min Fu, Xiaobo Feng.

Footnotes

Abbreviations: ASL = arterial spin labeling, AUC = area under the curve, BSM = brain solitary metastases, CNKI = China Knowledge Resource Integrated database, CI = confidence interval, DOR = diagnostic odds ratio, FN = false-negative, FP = false-positive, HGG = high-grade gliomas, rCBF = related cerebral blood flow, SEN = sensitivity, SPE = specificity, SROC = summary receiver operating characteristic curve, TP = True-positive, TN = true-negative.

Ethics approval and consent to participate: This study does not involve ethical issues.

All authors have no funding and conflicts of interest to disclose.

References

  • [1].Patel CK, Vemaraju R, Glasbey J, et al. Trends in peri-operative performance status following resection of high grade glioma and brain metastases: the impact on survival. Clin Neurol Neurosurg 2018;164:67–71. [DOI] [PubMed] [Google Scholar]
  • [2].Kerkhof M, Ganeff I, Wiggenraad RGJ, et al. Clinical applicability of and changes in perfusion MR imaging in brain metastases after stereotactic radiotherapy. J Neuro-Oncol 2018;138:1–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [3].Alatakis S, Stuckey S, Siu K, et al. Gliosarcoma with osteosarcomatous differentiation: review of radiological and pathological features. J Clin Neurosci 2004;11:650–6. [DOI] [PubMed] [Google Scholar]
  • [4].Holly KS, Barker BJ, Murcia D, et al. High-grade gliomas exhibit higher peritumoral fractional anisotropy and lower mean diffusivity than intracranial metastases. Front Surg 2017;4:18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [5].Min SK, Park SH, Park ES, et al. Quantitative analysis in peritumoral volumes of brain metastases treated with stereotactic radiotherapy. J Neuroradiol 2018;45:310–5. [DOI] [PubMed] [Google Scholar]
  • [6].Pan HC, Sun MH, Chen CC, et al. Neuroimaging and quality-of-life outcomes in patients with brain metastasis and peritumoral edema who undergo Gamma Knife surgery. J Neurosurg 2008;109Suppl:90–8. [DOI] [PubMed] [Google Scholar]
  • [7].Ferré JC, Bannier E, Raoult H, et al. Arterial spin labeling (ASL) perfusion: techniques and clinical use. Diagn Interv Imaging 2013;94:1211–23. [DOI] [PubMed] [Google Scholar]
  • [8].Razek AAKA, El-Serougy L, Abdelsalam M, et al. Differentiation of residual/recurrent gliomas from postradiation necrosis with arterial spin labeling and diffusion tensor magnetic resonance imaging-derived metrics. Neuroradiology 2018;60:169–77. [DOI] [PubMed] [Google Scholar]
  • [9].Li L, Wang X, Zhang M, et al. Clinical utility of arterial spin labeling for preoperative grading of glioma: a diagnostic meta-analysis. Biosci Rep 2018;38: [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [10].Di N, Pang H, Dang X, et al. Perfusion imaging of brain gliomas using arterial spin labeling: correlation with histopathological vascular density in MRI-guided biopsies. Neuroradiology 2016;59:1–9. [DOI] [PubMed] [Google Scholar]
  • [11].Brendle C, Hempel JM, Schittenhelm J, et al. Glioma Grading and Determination of IDH Mutation Status and ATRX loss by DCE and ASL Perfusion. Clin Neuroradiol 2018;28:421–8. [DOI] [PubMed] [Google Scholar]
  • [12].Weber MA, Thilmann C, Lichy MP, et al. Assessment of irradiated brain metastases by means of arterial spin-labeling and dynamic susceptibility-weighted contrast-enhanced perfusion MRI: initial results. Invest Radiol 2004;39:277–87. [DOI] [PubMed] [Google Scholar]
  • [13].Sunwoo L, Yun TJ, You SH, et al. Differentiation of glioblastoma from brain metastasis: qualitative and quantitative analysis using arterial spin labeling MR imaging. Plos One 2016;11:e0166662. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [14].Lin L, Xue Y, Duan Q, et al. The role of cerebral blood flow gradient in peritumoral edema for differentiation of glioblastomas from solitary metastatic lesions. Oncotarget 2016;7:69051–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • [15].Weber MA, Zoubaa S, Schlieter M, et al. Diagnostic performance of spectroscopic and perfusion MRI for distinction of brain tumors. Digest World Core Med J 2006;66:1899–906. [DOI] [PubMed] [Google Scholar]
  • [16].Ding F, Quan GM, Yuan T, et al. Differentiation of high grade glioma and solitary metastatic brain tumor in ring-like enhancement using three-dimensional arterial spin labeling. Chin J Med Imaging 2014;22:899–903. [Google Scholar]
  • [17].Ganbold M, Harada M, Khashbat D, et al. Differences in high-intensity signal volume between arterial spin labeling and contrast-enhanced T1-weighted imaging may be useful for differentiating glioblastoma from brain metastasis. J Med Invest 2017;64:58. [DOI] [PubMed] [Google Scholar]
  • [18].Stewart LA, Clarke M, Rovers M, et al. Preferred reporting items for a systematic review and meta-analysis of individual participant data: the PRISMA-IPD statement. J Am Med Assoc 2015;313:1657–65. [DOI] [PubMed] [Google Scholar]
  • [19].Hatjimihail AT. Quality of diagnostic accuracy studies: the development, use, and evaluation of QUADAS. Evid Based Med 2006;11:189–189. [Google Scholar]
  • [20].Cohen JF, Chalumeau M, Cohen R, et al. Cochran's Q test was useful to assess heterogeneity in likelihood ratios instudies of diagnostic accuracy. J Clin Epidemiol 2015;68:299–306. [DOI] [PubMed] [Google Scholar]
  • [21].Copas J, Shi JQ. Meta-analysis, funnel plots and sensitivity analysis. Biostatistics 2000;1:247–62. [DOI] [PubMed] [Google Scholar]
  • [22].Ding F, Quan G, Yuan T, et al. Differentiation of high grade glioma and solitary metastatic brain tumor in ring-like enhancement using three-dimensional arterial spin labeling. Chin J Med Imaging 2014;22:899–903. [Google Scholar]

Articles from Medicine are provided here courtesy of Wolters Kluwer Health

RESOURCES