Diagnostic accuracy of Doppler ultrasound, CT and MRI in Budd Chiari syndrome: systematic review and meta-analysis

Abstract

Objective:

To evaluate the sensitivity, specificity, and diagnostic odds ratio (DOR) of Doppler ultrasound, CT, and MRI in the diagnosis of Budd Chiari syndrome (BCS).

Methods:

We performed a literature search in PubMed, Embase, and Scopus to identify articles reporting the diagnostic accuracy of Doppler ultrasound, CT, and MRI (either alone or in combination) for BCS using catheter venography or surgery as the reference standard. The quality of the included articles was assessed by using the Quality Assessment of Diagnostic Accuracy Studies-2 tool.

Results:

11 studies were found eligible for inclusion. Pooled sensitivities and specificities of Doppler ultrasound were 89% [95% confidence interval (CI), 81–94%, I² = 24.7%] and 68% (95% CI, 3–99%, I² = 95.2%), respectively. Regarding CT, the pooled sensitivities and specificities were 89% (95% CI, 77–95%, I² = 78.6%) and 72% (95% CI, 21–96%, I² = 91.4%), respectively. The pooled sensitivities and specificities of MRI were 93% (95% CI, 89–96%, I² = 10.6%) and 55% (95% CI, 5–96%, I² = 87.6%), respectively. The pooled DOR for Doppler ultrasound, CT, and MRI were 10.19 (95% CI: 1.5, 69.2), 14.57 (95% CI: 1.13, 187.37), and 20.42 (95% CI: 1.78, 234.65), respectively. The higher DOR of MRI than that of Doppler ultrasound and CT shows the better discriminatory power. The area under the curve for MRI was 90.8% compared with 88.4% for CT and 86.6% for Doppler ultrasound.

Conclusion:

Doppler ultrasound, CT and MRI had high overall diagnostic accuracy for diagnosis of BCS, but substantial heterogeneity was found. Prospective studies are needed to investigate diagnostic performance of these imaging modalities.

Advances in knowledge:

MRI and CT have the highest meta-analytic sensitivity and specificity, respectively for the diagnosis of BCS. Also, MRI has the highest area under curve for the diagnosis of BCS.

Introduction

Budd Chiari syndrome (BCS), also known as hepatic venous outflow tract obstruction, is a heterogeneous disease. It results from the venous obstruction at the level of hepatic veins (HVs), inferior vena cava (IVC) or both.^1,2 The impaired venous drainage causes hepatocyte ischemia and dysfunction. If this is not reversed, hepatocyte death ensues. This injury and cell death pathway may eventually progress to cirrhosis. This progression can be prevented if it is diagnosed at an early stage.^3,4 The clinical features of BCS are myriad and may range from asymptomatic disease early in the course, particularly in patients with involvement of one hepatic vein, to signs of chronic liver disease with ascites and other signs of hepatic dysfunction. Thus, an early diagnosis of this entity relies on radiological evaluation.

The non-invasive modalities for diagnosis of BCS are Doppler ultrasound, CT, and MRI. Doppler ultrasound allows anatomical as well as hemodynamic evaluation of suspected lesions.⁵ However, due to operator dependence and subjective nature, many centers routinely perform CT or MRI for confirmation of diagnosis. MRI is the modality of choice for evaluation of liver parenchyma as well as the venous system, including hepatic veins and IVC.⁶ However, MRI is not available in all the centers. Image quality may deteriorate in the presence of ascites.²

A few studies have shown that CT is also equally accurate in the diagnosis of BCS.^7–10 Because of heterogeneity in the reported data, there is no clarity as to which modality has the highest diagnostic accuracy. A recent meta-analysis explored the diagnostic accuracy of MRI.¹¹ However, this meta-analysis was limited by the fact that it included only recent studies. Additionally, most of the included studies (four out of the six studies) were in the Chinese language and represented a limited demographic population.¹¹ To the best of our knowledge, there is no other meta-analysis assessing the diagnostic accuracy of all the three primary imaging modalities (Doppler ultrasound, CT, and MRI) in BCS. The purpose of this meta-analysis was to evaluate the sensitivity, specificity, and diagnostic odds ratio (DOR) of Doppler ultrasound, CT, and MRI in the diagnosis of BCS using catheter venography or surgery as the reference standard.

Methods and materials

This meta-analysis complied with the Preferred Reporting Items for Systematic reviews and Meta-Analysis guidelines.¹² The protocol of the meta-analysis was registered with the international prospective registry of systematic reviews (CRD42019131377).

Literature search

The following databases were searched: PubMed, Embase, and Scopus to examine the use of Doppler ultrasound, CT, and MRI in the diagnosis of BCS. Two reviewers screened the title and abstracts independently, and disagreement were resolved in consensus regarding the inclusion of a potential article. There was no time or language restriction for the search. Articles were searched up to 15 March 2019. The following terms were used for the search: (BCS or Budd Chiari syndrome or hepatic venous outflow tract obstruction) and (ultrasound or USG or ultrasound or Doppler or CT or computed tomography or CTA or computed tomography angiography or CTV or computed tomography venography or MR or MRI or magnetic resonance imaging or MRA or magnetic resonance angiography or MRV or magnetic resonance venography). All the identified studies were retrieved, and their references were also checked for other relevant publications. Studies were first screened by examining their titles and abstracts. The full texts of potentially eligible studies were retrieved for further review.

Inclusion and exclusion criteria

Criteria for inclusion of articles were as follows: (a) prospective or retrospective articles should have evaluated the diagnostic accuracy of Doppler ultrasound, CT, and MRI (either alone or in combination) for the diagnosis of BCS; (b) the reference standard should have consisted of catheter venography and/or surgery; (c) the absolute numbers of true-positive, true-negative, false-positive, and false-negative findings should have been directly reported or derivable from reported data in the paper (for studies that were deemed eligible but had some missing data, the corresponding authors of those studies were contacted via email). We included all the studies reporting ≥5 patients.

Review articles, case series (less than five patients), case reports, pictorial essays, letter to the editor, unpublished data, conference abstracts, and proceedings on the topic of interest were excluded. Two independent reviewers screened all the studies and read the full text of eligible studies after completing the search. The disagreement regarding the inclusion of an individual study was settled in consensus.

Data extraction and quality assessment

Articles were separated into those that focused on Doppler ultrasound, CT, MRI alone, or in combination. Extracted data included the following: study type; imaging modality employed for diagnosis; reference standard; veins involved; patient demographic information; numbers of true-positive, true-negative, false-positive, and false-negative. The methodological details of the procedure namely equipment details and the technical details were also extracted. Both the readers assessed the methodologic and reporting quality of each study independently by using the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool.¹³ Disagreements were settled by mutual consensus.

Statistical analysis

A 2 × 2 contingency table was extracted or reconstructed for each imaging modality in the included studies. Sensitivity and specificity estimates were calculated from the contingency tables for each individual study. The pooled sensitivities, specificities, and 95% confidence intervals (CIs) were calculated using random-effects method with Wilson Score confidence interval for individual studies. The pooled positive likelihood ratio (PLR), negative likelihood ratio (NLR), and DORs were also calculated. We did a bivariate diagnostic random-effects meta-analysis and hierarchical summary receiver operating characteristics (HSROC) model. The pooling of the sensitivity and specificity considers both the factors separately and gives the summary estimate individually. On the other hand, the bivariate random effects meta-analysis considers the joint distribution of sensitivity and specificity and thereby enables for across study correlation. The random effects assume that the variation seen in the effect size for each study is because of both within-study as well as between-study variation, this is as against the fixed effect which assumes that the variation in the effect size is because of the within-study variation.¹⁴ Area under the curve (AUC) were computed for each diagnostic test. For the pooled analysis, heterogeneity was quantified using the I² test statistic, including 95% CI, Cochrane Q statistic and its associated p-value of heterogeneity (Likelihood-ratio test). The I² statistic reports the percentage of total variation across studies that is due to heterogeneity rather than chance. Though number of factors determine heterogeneity, on a rough estimate a I² statistic greater than 50% implies substantial heterogeneity.¹⁵ For all the calculations, the default accepted continuity correction of 0.5 was applied, except for bivariate modelling for which the continuity correction was kept as 0, but all the 0 in the data set were uniformly taken as one for Doppler ultrasound, CT, and MRI. Continuity correction is the value which is added to the cells with zero values. It helps in reduction of bias towards estimation.¹⁶ Wilson method was used for the calculation of CI for both pooled analysis as well as the summary results. Test of equality was computed among the studies for sensitivity and specificity for Doppler, ultrasound and CT scan separately. Test of equality is based on the proportional test. It tests whether the proportions are equal among the studies. For example, the test of equality for sensitivity for Doppler assesses whether among the studies reporting Doppler, are the sensitivities statistically same. R statistical software (v. 3.5.1) was used for statistical analysis. In addition to the base package of R, mada (0.5.8) and meta (4.9.3) package was additionally used.

Results

Literature search and article selection

The study selection process is detailed in the flowchart (Figure 1). The search yielded a total of 13,476 articles (PubMed - 3286 articles; Embase - 5375 articles; Scopus - 4815). After screening titles and abstracts, and removing duplicates, the full texts of 40 articles were reviewed. Table 1 shows the excluded articles for which the full texts were available. Finally, 11 articles were selected: 3 articles on Doppler ultrasound; 2 articles on CT; 3 articles on MRI; 1 article on Doppler ultrasound and CT; and 2 articles on Doppler ultrasound, CT and MRI. The details of these studies are given in Table 2.

Figure 1.

Flowchart of study inclusion.

Table 1.

Reason for the exclusion of available studies that were appropriate for the diagnosis of BCS

First author	Year	Modality	Study type	Reason for exclusion
Mori17	1989	CT	Retrospective	Inability to construct 2 × 2 table
Lim8	1992	Doppler ultrasound and CT	Retrospective	Inability to construct 2 × 2 table
Lin18	2003	MRI	Prospective	Had zero value for 3 out of the four namely TP, FP, FN and TN.
Meng19	2007	CT	Retrospective	Inability to construct 2 × 2 table
Faraoun20	2015	Doppler ultrasound, CT and MRI	Prospective	Inadequate reference standard
Zhang9	2015	Doppler ultrasound and CT	Retrospective	Inability to construct 2 × 2 table
Lu21	2015	MRV	Prospective	Inability to construct 2 × 2 table
Yang22	2017	MRA	Prospective	Inability to construct 2 × 2 table
Lu23	2011	MRV	Lack of information	Incomplete data
Shen24	2013	MRI	Lack of information	Incomplete data
Qin25	2015	MRI	Lack of information	Incomplete data
Pu26	2015	MRI	Lack of information	Incomplete data

BCS, Budd Chiari syndrome; FN, false negative; FP, false positive; MRA, magnetic resonance angiography;TN, true negative; TP, true positive.

Table 2.

Demographic details and diagnostic performances of the included studies

First author	Year	Study type	Modality	Reference standard	Veins involved	No. of patients enrolled	TP	FN	FP	TN	Sensitivity (%)	Specificity (%)	Positive likelihood ratio	Negative likelihood ratio	Diagnostic odd ratio
Chawla27	1999	Prospective	Doppler	DSA	IVC	37	15	3	0	19	83	100	32.63	0.19	172.71
Millener28	1993	Prospective	Doppler	DSA	IVC	20	17	3	0	0	85	50	1.67	0.33	5
Bolondi29	1991	Retrospective	Doppler	DSA	HV	8	7	1	0	0	83	50	1.76	0.33	5
Liu7	2016	Retrospective	Multiphase CT	DSA	IVC	329	215	15	6	93	93	94	14.35	0.07	199.99
Virmani10	2009	Prospective	Multiphase CT	DSA	IVC	25	21	0	0	4	100	100	9.7	0.03	387
Ren30	2007	Prospective	MRI/MRA	DSA	IVC	50	38	3	1	8	93	89	6.11	0.1	62.33
Wu31	2014	Prospective	MRI/MRA	DSA	HV + IVC	35	32	0	1	2	100	67	2.63	0.02	108.33
Song32	2018	Prospective	MRI/MRA	DSA	HV + IVC	52	45	7	0	0	87	50	1.72	0.28	6.07
Gupta33	1987	Retrospective	Doppler	DSA	HV + IVC	18	8	1	4	5	89	56	1.89	0.27	6.93
			CT			14	7	1	2	4	78	57	1.71	0.44	3.86
Zhou34	2014	Retrospective	Doppler	DSA	HV + IVC	338	302	24	12	0	93	0*	0.96	1.95	0.49
			CT			126	101	17	8	0	86	0*	0.9	2.65	0.34
			MRI			157	131	8	18	0	94	0*	0.96	2.31	0.42
Miller35	1993	Retrospective	Doppler	DSA	HV	40	14	4	3	19	78	86	5.02	0.28	17.95
			CT			36	10	6	6	14	62	7	2	0.55	3.6
			MRI			30	12	1	2	15	92	88	6.43	0.12	51.67

^*-No continuity correction applied; DSA: digital subtraction angiography; FN: false negative; FP: false positive; MRA: magnetic resonance angiography;^* TN: true negative;TP: true positive.

QUADAS-2 assessment

The results of QUADAS-2 assessment are presented in Table 3. Overall, 10 studies were at risk of bias, and 7 studies raised concerns regarding applicability. The most common domain at high risk of bias was the method of patient selection (consecutive or random) and lack of listing of the exclusion criteria. Flow and timing were at high or unclear risk in five studies. All the studies had a detailed description of the technical parameters of the Doppler ultrasound, CT, and MRI (Table 4). Hence, there was no bias related to the description of the index test. Two studies had a high risk of bias due to the concern that the interpretation of the index test results was made in the presence of results of the reference standard. None of the studies had a bias related to the reference standard.

Table 3.

Quality of studies

	Risk of bias				Applicability concerns
	Patient selection	Index test	Reference standard	Flow and timing	Patient selection	Index test	Reference standard
Chawla27
Millener28
Bolondi29
Liu7
Virmani10
Ren30
Wu31
Song32
Gupta33
Zhou34
Miller35

-low risk, - unclear risk, - high risk

Table 4.

Methodological and technical details of the included studies

Author	Imaging method	Machine details	Technical details
Chawla27	Doppler	3.5 MHz (GE-RT 3600, GE Co., Rancho, Cordova)	Color flow and spectral waveforms.
Millener28	Doppler	2.5–3.5 MHz transducer (the machine details were not provided)	Color flow evaluation. Spectral waveforms not specifically evaluated.
Bolondi29	Doppler	2.5–3.5MHz transducer (Ansaldo-Hitachi EUB 40, EUB 450, Tokyo, Japan	Color flow and spectral waveform
Liu7	CT	6-row detector (Brilliance, Philips), 16-row detector (Sensation 16, Siemens; Light Speed 16, GE Medical systems) and 64-row detector (Light Speed VCT or Discovery CT 750 HD, GE Healthcare)	Non-contrast and triphasic CT; arterial (25 s), portal venous (60 s) and delayed (180 s) 100 ml of ultravist 370 at the rate of 3 mL/ s. Beam collimation 2.5 mm, Pitch 0.75, reconstruction interval 0.75–1 mm, tube current 100–600 mAs, tube voltage 120 kVp
Virmani10	CT	16-row detector (Sensation 16, Siemens)	Two acquisitions-first with automatic triggering after an attenuation of 100 HU in the infrarenal IVC and second in the portal venous phase after 60 ml of ultravist 370 at a rate of 4 ml s−1. Beam collimation 2.5 mm, Pitch 0.75, reconstruction interval 0.75 mm, gantry rotation time 0.5 s, tube current 200–250 mAs, tube voltage 120 kVp
Ren30	MRI	1.5 T (Toshiba Visart/Ex, Japan)	MR angiography with 2D and 3D fresh blood imaging. No intravenous contrast injection.
Wu31	MRI	1.5 T (Signa, GE)	Inflow inversion recovery and Fast imaging employing steady state acquisition. No intravenous contrast injection.
Song32	MRI	3T (Magnetom Verio, Siemens)	Pre-contrast and post-contrast 3D volume interpolated body examination.
Gupta33	Doppler	3.5 MHz (Diasonic 400 CV) or % MHz (Toshiba SAL 50A)	Color flow evaluation.
	CT	Siemens Somatom 2	Non-contrast followed by single phase after injection of 100 ml Urograffin at the rate of 1 ml s−1. 8 mm thick sections.
Zhou34	Doppler	GE, USA (details of transducer not available)	Technical details not specified.
	CT	6-row detector (Brilliance, Philips), 16-row detector and 64-row detector (Light Speed, GE Healthcare)	Non-contrast and triphasic CT; arterial (25–30 s), portal venous (60 s) and IVC (180 s). Amount and rate of contrast injection not specified.
	MRI	1.5 or 3 T signa (GE Healthcare)	Pre-contrast and post-contrast dynamic 3D images.
Miller35	Doppler	2.5–3.5 MHz (Acuson scanners, Acuson Corp., Mountain View, CA)	Color flow and spectral Doppler.
	CT	GE CT 9800/8800	Non-contrast CT (n = 3), Contrast enhanced CT only (n = 1), non-contrast followed by contrast enhanced CT (n = 17). 150 ml of Conray 60/Isovue 300 at the rate of 2 ml s−1. Scanning after 40 s of contrast injection. 10 mm thick sections.
	MRI	1.5 T (Signa, GE)	No intravenous contrast injection. Short TR/TE (400–800 ms/20 ms) spin echo images, long TR (>1800 ms) with short (20–30 ms) and long TE (80–100 ms) and/or gradient echo images (GRASS).

2D, two-dimensional; 3D, three-dimensional; HU, Hounsfield unit; IVC, inferior vena cava; TE, echo time; TR, repetition time.

Diagnostic performance evaluation of various imaging modalities

The pooled sensitivities and specificities for Doppler in the diagnosis of BCS were 89% (95% CI, 81–94%, I² = 24.7%) and 68% (95% CI, 3–99%, I² = 95.2%), respectively. Regarding CT, the pooled sensitivities and specificities were 89% (95% CI, 77–95%, I² = 78.6%) and 72% (95% CI, 21–96%, I² = 91.4%), respectively. The pooled sensitivities and specificities of MRI were 93% (95% CI, 89–96%, I² = 10.6%) and 55% (95% CI, 5–96%, I² = 87.6%),

These values are summarized in Table 5. Forest plots for the sensitivities and specificities of Doppler ultrasound, CT, and MRI are shown in Figure 2. The test of equality (p-value) of sensitivity and specificity among studies for Doppler, CT and MRI are: 0.082, <0.001; <0.001, <0.001; and 0.23, <0.001 respectively. The correlation (ρ) between the sensitivities and false positive rates for Doppler, CT and MRI are: 0.88 (95% CI, −0.24–0.99); −0.23 (95% CI, −0.92–0.82); 0.18 (95% CI, −0.84–0.92). The pooled DOR for Doppler ultrasound, CT, and MRI were 10.19 (95% CI, 1.5–69.2), 14.57 (95% CI, 1.13–187.37), and 20.427 (95% CI, 1.78–234.65), respectively. MRI yielded an AUC of 90.8% compared with 88.4% for CT and 86.6% for Doppler ultrasound. The HSROC curves for the three diagnostic test are shown in Figure 3.

Table 5.

Summary of sensitivity and specificity of Doppler, CT and MRI

Modality	Sensitivity	Specificity
Doppler	0.89 (0.81–0.94)	0.68 (0.03–0.99)
CT	0.89 (0.77–0.95)	0.72 (0.21–0.96)
MRI	0.93 (0.89–0.96)	0.55 (0.05–0.96)

Figure 2.

Forest plots demonstrate the sensitivities and specificities of Doppler ultrasound, CT, and MRI.

Figure 3.

Summary receiver operating characteristic curves for Doppler ultrasound, CT, and MRI.

Source of heterogeneity

There was significant heterogeneity for the specificity of Doppler ultrasound, CT, and MRI (I² = 95.2%, p (heterogeneity) <0.001; 91.4%, p < 0.001; 87.6%, p < 0.001 respectively). For sensitivity, there was significant heterogeneity for CT (78.6%, p < 0.001). There was non-significant heterogeneity for the sensitivity of Doppler and MRI (24.7%, p = 0.314; 10.6%, p = 0.104 respectively). The study by Zhou et al consistently yielded a lower-than-predicted specificity for all the modalities.³⁰ No outliers were identified for the sensitivity of CT.

Summary of technical and methodological details of the included studies (Table 4)

Doppler examinations were performed using transducer with frequency ranging between 2.5and 3.5 MHz. Evaluation included color and spectral Doppler evaluation. CT scans were performed on multidetector scanners with variable detector rows. Most of the studies were performed with multiphase protocols including non-contrast scans. The detailed parameters of CT scan were reported in two studies. MRI examinations were performed on 1.5 or 3T scanners. Two studies included examinations without intravenous contrast injection.^30,31

Discussion

This systematic review was performed to assess the diagnostic performance of the three imaging modalities (Doppler, CT and MRI) that are commonly employed for evaluation of patients with BCS. We found that the pooled sensitivities of all imaging modalities were high. However, the pooled specificities were comparatively lower. The highest pooled sensitivities were demonstrated by MRI (93%) while the highest pooled specificities were found for CT (72%). DOR of MRI was the highest ~20.42. Similarly, MRI yielded the highest AUC of 90.8% compared with 88.4% for CT and 86.6% for Doppler. The pooled sensitivity of Doppler was higher than that of CT, while pooled specificity was higher than that of MRI. The AUC of Doppler was slightly lower than CT. The higher DOR of MRI as compared to that of Doppler ultrasound and CT shows the better discriminatory power.

These results suggest that Doppler is an excellent first-line imaging test for patients suspected to have BCS. The advantages of Doppler ultrasound are easy availability and lower cost. Doppler ultrasound also provides functional information regarding the stenosis.⁵ However, the subjective nature of assessment and lack of evaluation of the extrahepatic collateral pathways is a significant limitation. The results of this meta-analysis also suggest the higher diagnostic accuracy of MRI compared with CT and Doppler ultrasound. CT is more widely available, and most of the radiologists have excellent expertise in the interpretation of CT. It allows a comprehensive evaluation of the hepatic veins, IVC, and the intrahepatic as well as extrahepatic collateral pathways.² The results from the present meta-analysis suggest that CT may be an appropriate test in centers where MRI is unavailable or where the expertise for performance or interpretation of MRI is limited. Additionally, MRI may not be feasible in patients with massive ascites due to the artifacts and the inability of the patient to lie down for the duration of the scan.²

A possible explanation for the lower pooled specificities is the effect of the study by Zhou et al. As this retrospective study recruited patients with confirmed BCS, there were no true-negative cases on Doppler ultrasound/CT/MRI. A recent meta-analysis by Xu et al on the diagnostic accuracy of magnetic resonance angiography for BCS yielded a pooled sensitivity of 97.6% (95% CI, 95.1–99.0%) and a pooled specificity of 70.7% (95% CI, 54.5–83.9%).¹¹ The overall DOR was 94.053 (95% CI, 32.71–270.41), and the AUC was 0.972. These estimates are higher than our study. The possible explanation for this discrepancy is the fact that Xu et al did not include the study by Zhou et al which proved an outlier for calculation of specificities of all the imaging modalities in our meta-analysis.^11,30 However, like our results, this study showed that the pooled sensitivity is higher than the pooled specificity.

There were a few differences in the methodological aspects of the included studies, especially those reporting CT and MRI. The CT scans were performed across variable detector row configuration. Most of the CT studies used a multiphasic protocol, although there were differences in the timing of the acquisition of various phases.^7,10,34 Liu et al performed CT in the arterial, portal venous and hepatic venous phase.⁷ On the other hand, Virmani et al performed two phase, one with automatic triggering with the region of interest in the infrarenal IVC and the other in the portal venous phase.¹⁰ Both these studies reported a high sensitivity and specificity. However, Gupta et al performed non-contrast CT scans followed by a single phase after injection of intravenous contrast at a very low injection rate (1 ml s⁻¹).³³ The standard rate of contrast injection at present is 3–4 ml s⁻¹. Miller et al also performed a single-phase contrast CT in most patients.³⁵ In few patients, only a non-contrast CT was performed. This CT protocol could be partially responsible for lower diagnostic performance of CT in these studies. Three out of the five MRI studies were performed without the injection of intravenous gadolinium.^30,31,35 These studies relied on non-contrast angiography sequence to depict the venous abnormality. The relatively lower specificity of MRI in the study by Wu et al and Song et al may be explained by the fact that these studies included patients with combine type of BCS.^31,32 Additionally, Wu et al did not inject intravenous contrast.³¹ The lower specificity in the study by Song et al despite the use of intravenous contrast is partly due to inclusion of patients with cord like occlusion and membranous obstruction.³²

There were a few limitations to our study. There were a limited number of studies satisfying the inclusion criteria. BCS is a heterogeneous condition that can involve the HV or IVC or both (in different forms: short segment stenosis, long-segment stenosis, and occlusion). Moreover, the disease may be acute, sub acute, or chronic. Most of the included studies had assessed the diagnostic accuracy for IVC obstruction. Only one study assessed HV obstruction alone, and three studies assessed the HV + IVC obstruction. A subgroup analysis could not be performed due to a limited number of patients and lack of data. The quality assessment revealed that most of the studies were at risk of bias and indicates that better prospective studies need to be performed. Substantial heterogeneity of the included studies is another limitation.

In conclusion, MRI showed the highest sensitivity, and CT showed the highest specificity for the diagnosis of BCS. Though the AUC was the highest for MRI, overall pooled diagnostic accuracy results support the use of Doppler, CT, and MRI interchangeably and based on the local availability and expertise. However, this meta-analysis also highlights the need for prospective studies exploring the diagnostic performance of the imaging tests based on the site and type of venous obstruction as well as the duration of disease.