2.1 Literature Retrieval Results
Fifty-four studies were selected for inclusion after reading the full text, of which 43 were in English and 11 were in Chinese. Patients include both adults and children. The studies were conducted in the following countries: China (n = 24), India (n = 2), Italy (n = 3), Spain (n = 1), Turkey (n = 1), Sweden (n = 2), Japan (n = 3), Norway (n = 1), the United States (n = 5), Canada (n = 1), Korea (n = 2), France (n = 1), Germany (n = 4). Denmark (n = 1), Belgium (n = 1), Brazil (n = 1), Australia (n = 1). Seven studies reported two methods. Of those studies including quantitative data and Continuous Variable Forest Map, 20 was in ASL, 22 in DSC, 15 in DKI of the studies. Of those studies including fourfold table data for meta-analysis of diagnostic tests, 19 was in ASL, 19 in DSC, 16 in DKI. The retrieval process is shown in Fig. 1.
2.2 Analysis
2.2.1 rCBF in ASL
Twenty studies assessing the difference of rCBF between HGG and LGG were included. Heterogeneity test showed that χ2 = 66.79, I༒= 72%, P < 0.001, indicating substantial heterogeneity. Therefore, the random effect model was applied to estimate the overall rCBF. The overall rCBF was 1.45 (1.12, 1.77), P < 0.001 (Fig. 2).
2.2.2 rCBV in DSC-MRI
Twenty-two studies assessing the difference of rCBV between HGG and LGG were included. Heterogeneity test showed that χ2 = 74.23, I༒= 72%, P < 0.001, indicating substantial heterogeneity. Therefore, the random effect model was applied to estimate the overall rCBV. The overall rCBV was 1.37 (1.08, 1.66), P < 0.001 (Fig. 3).
2.2.3 MK in DKI
Fifteen studies assessing the difference of MK between HGG and LGG were included. Heterogeneity test showed that χ2 = 46.39, I༒= 70%, P < 0.001, indicating substantial heterogeneity. Therefore, the random effect model was applied to estimate the overall MK. MK The overall MK was 1.57 (1.21, 1.93), P < 0.001 (Fig. 4).
2.2.4 Diagnostic Value
Sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic ratio and area under curve were summarized according to the studies including fourfold table (Table 2). The results showed that rCBF had the highest diagnostic ratio of 71 (31,163). According to the SROC curve (Fig. 5), the largest AUC was rCBF, followed by MK and rCBV
The incidence of glioma is about 45% of all intracranial tumors [63]. Fagan diagram of diagnostic grading of glioma by rCBF, rCBV and MK is shown in Fig. 6. If the prior probability is 45%, the posterior probability of rCBF, rCBV and MK is 88%, 80% and 83% respectively. The higher the DOR value, the better the discrimination ability. The DOR value of rCBF is 71 (31,163), indicating an overall high accuracy.
(As shown in Figure A, assuming that the patient's prior probability is 45%, the posterior probability is 88%, and the PLR is 9.)
Table 2
The values of rCBF,rCBV and MK
index | n | Sen (95%CI) | Spe (95%CI) | PLR (95%CI) | NLR (95%CI) | DOR (95%CI) | AUC (95%CI) |
rCBF | 19 | 0.88 (0.83,0.92) | 0.91 (0.84,0.94) | 9.3 (5.4,16.0) | 0.13 (0.09,0.20) | 71 (31,163) | 0.95 (0.93,0.97) |
rCBV | 19 | 0.92 (0.83,0.96) | 0.81 (0.73,0.88) | 5.0 (3.3,7.4) | 0.10 (0.05,0.22) | 50 (20,129) | 0.91 (0.89,0.94) |
MK | 16 | 0.88 (0.82,0.92) | 0.86 (0.78,0.91) | 6.2 (4.1,9.3) | 0.14 (0.10,0.21) | 44 (26,75) | 0.93 (0.91,0.95) |
2.2.5 Meta-regression
The results of meta-regression are shown in Table 3. The results showed that among the five covariates in ASL study, region, year of study, number of patients and QUADAS-2 score were all important factors contributing to heterogeneity except for field strength. Among the six covariates in DSC-MRI study, region, year of study, number of patients, field strength and QUADAS-2 score, none had significant impact on heterogeneity. Among the five covariates in DKI study, the year of study, age of patients, number of patients and QUADAS-2 score all had no significant impact on heterogeneity except for region.
Table 3
| Variable | Subgroup | n | Overall estimate of meta-regression |
Sensitivity(95% CI) | p | Specificity(95% CI) | p |
ASL | Region | China | 14 | 0.89(0.84,0.94) | 0.01 | 0.89(0.83,0.95) | 0.71 |
others | 5 | 0.86(0.77,0.95) | 0.94(0.88,1.00) |
Year | 2008–2014 | 8 | 0.87(0.80,0.93) | 0.00 | 0.88(0.80,0.96) | 0.01 |
2015–2019 | 11 | 0.89(0.83,0.95) | 0.92(0.87,0.98) |
Number of patients | ≤ 40 | 10 | 0.90(0.84,0.95) | 0.00 | 0.93(0.87,0.99) | 0.01 |
༞40 | 9 | 0.87(0.81,0.93) | 0.88(0.81,0.95) |
Field strength | 1.5T | 2 | 0.90(0.79,1.00) | 0.21 | 0.96(0.90,1.00) | 0.13 |
3.0T | 17 | 0.88(0.83,0.93) | 0.89(0.84,0.94) |
QUADAS-2 score | ≤ 10 | 7 | 0.93(0.89,0.98) | 0.00 | 0.94(0.89,1.00) | 0.01 |
༞10 | 12 | 0.84(0.79,0.90) | 0.88(0.81,0.94) |
DSC-MRI | Region | China | 4 | 0.94(0.83,1.00) | 0.77 | 0.88(0.74,1.00) | 0.14 |
others | 15 | 0.91(0.84,0.99) | 0.80(0.72,0.88) |
Year | 2006–2014 | 10 | 0.95(0.89,1.00) | 0.07 | 0.80(0.69,0.90) | 0.20 |
2015–2017 | 9 | 0.87(0.74,0.99) | 0.83(0.73,0.93) |
Age | ≤ 45 | 9 | 0.94(0.87,1.00) | 0.31 | 0.85(0.75,0.95) | 0.03 |
༞45 | 10 | 0.90(0.80,1.00) | 0.79(0.68,0.89) |
Number of patients | ≤ 40 | 10 | 0.95(0.88,1.00) | 0.33 | 0.89(0.82,0.95) | 0.56 |
༞40 | 9 | 0.89(0.79,1.00) | 0.72(0.62,0.83) |
Field strength | 1.5T | 4 | 0.98(0.93,1.00) | 0.23 | 0.76(0.60,0.92) | 0.00 |
3.0T | 15 | 0.90(0.82,0.98) | 0.83(0.75,0.91) |
QUADAS-2 score | ≤ 10 | 10 | 0.94(0.88,1.00) | 0.14 | 0.86(0.77,0.95) | 0.01 |
༞10 | 9 | 0.88(0.77,0.99) | 0.77(0.66,0.88) |
DKI | Region | China | 9 | 0.89(0.83,0.95) | 0.01 | 0.89(0.82,0.95) | 0.01 |
others | 7 | 0.86(0.77,0.94) | 0.81(0.69,0.92) |
Year | 2010–2015 | 4 | 0.85(0.75,0.95) | 0.14 | 0.89(0.79,0.99) | 0.06 |
2016–2019 | 12 | 0.89(0.84,0.94) | 0.85(0.77,0.96) |
Age | ≤ 48 | 9 | 0.85(0.78,0.92) | 0.14 | 0.87(0.79,0.94) | 0.02 |
༞48 | 7 | 0.91(0.85,0.97) | 0.84(0.75,0.93) |
Number of patients | ≤ 40 | 11 | 0.83(0.77,0.89) | 0.32 | 0.88(0.81,0.94) | 0.00 |
༞40 | 5 | 0.93(0.89,0.98) | 0.80(0.70,0.90) |
QUADAS-2 score | ≤ 10 | 10 | 0.84(0.78,0.91) | 0.18 | 0.87(0.80,0.94) | 0.02 |
༞10 | 6 | 0.92(0.87,0.97) | 0.83(0.72,0.93) |
2.2.6 Subgroup analysis
In ASL, subgroup analysis was carried out based on region and technique; in DSC-MRI, subgroup analysis was carried out based on region and magnetic resonance field strength; in DKI, subgroup analysis was carried out based on region. The results of subgroup analysis are shown in Table 4.
Table.4. Subgroup analysis
| Subgroup | Category | n | p | Sen (95%CI) | Spe (95%CI) | PLR (95% CI) | NLR (95% CI) | DOR (95% CI) | AUC (95% CI) |
ASL | Region | CHINA | 14 | 0.30 | 0.88 (0.83,0.92) | 0.89 (0.81,0.94) | 8.4 (4.5,15.4) | 0.13 (0.09,0.20) | 63 (25,157) | 0.94 (0.92,0.96) |
other | 5 | 0.35 | 0.89 (0.68,0.97) | 0.94 (0.84,0.98) | 14.0(5.1,38.2) | 0.11(0.03,0.40) | 123(18,846) | 0.96 (0.94,0.98) |
Technique | 3DPCASL | 10 | 0.33 | 0.87 (0.82,0.91) | 0.88 (0.81,0.93) | 7.6 (4.4,13.1) | 0.14 (0.10,0.21) | 53 (24,116) | 0.92 (0.90,0.94) |
PASL | 8 | 0.46 | 0.93(0.75,0.98) | 0.93(0.80,0.98) | 14.2(4.2,48.1) | 0.08(0.02,0.31) | 183(19,1754) | 0.98(0.96,0.99) |
DSC-MRI | Region | CHINA | 4 | 0.23 | 0.91(0.82,0.96) | 0.90(0.55,0.99) | 9.6(1.5,60.4) | 0.10(0.05,0.21) | 95 (11,835) | 0.89(0.82,0.96) |
other | 15 | 0.00 | 0.92(0.80,0.97) | 0.80(0.70,0.87) | 4.6(3.1,6.9) | 0.10(0.04,0.26) | 46 (15,144) | 0.90(0.87,0.92) |
Field strength | 3.0 T | 15 | 0.00 | 0.88(0.79,0.94) | 0.82(0.73,0.89) | 5.0(3.2,7.7) | 0.14(0.08,0.26) | 35 (17,73) | 0.91(0.88,0.93) |
1.5 T | 4 | 0.01 | 1.00(0.12,1.00) | 0.77(0.57,0.90) | 4.4(2.1,9.1) | 0.00(0.00,10.67) | 2976(0,3125) | 0.91(0.89,0.93) |
DKI | Region | CHINA | 9 | 0.00 | 0.90(0.81,0.95) | 0.90(0.79,0.95) | 8.6(4.3,17.2) | 0.11(0.06,0.21) | 75 (35,160) | 0.95(0.93,0.97) |
other | 7 | 0.46 | 0.85(0.76,0.91) | 0.79(0.69,0.87) | 4.1(2.7,6.2) | 0.20(0.12,0.32) | 21 (10,44) | 0.88(0.85,0.91) |
2.3 Publication bias
Deek's test was used to evaluate publication bias for studies containing fourfold tables. P > 0.1 indicated that there was no publication bias. Among them, 19 studies were included for ASL. Deek's test showed P = 0.85 indicating no significant publication bias (Deek's funnel plot Fig. 7a). 19 studies were included for DSC-MRI. Deek's test showed P = 0.45 indicating no significant publication bias (Deek's funnel plot Fig. 7b). 16 studies were included for DKI. Deek's test showed P = 0.12 indicating no significant publication bias (Deek's funnel plot Fig. 7c).
2.4 Sensitivity Analysis
Sensitivity analysis is an important method to deal with heterogeneity and publication bias. There was no significant difference in meta-analysis results of rCBF and rCBV after excluding the literature one by one, suggesting that rCBF and rCBV were more stable to be included in the literature. In DKI, after excluding Falk Delgado's research, χ2 decreased from 46.39 to 28.11, and I2 from 70–54%, suggesting significant heterogeneity brought by this study.