Phase-resolved MRI for measurement of pulmonary perfusion and ventilation defects in comparison with dynamic contrast-enhanced MRI and 129Xe MRI

Abstract

Introduction

This meta-analysis aims to evaluate the agreement and correlation between phase-resolved functional lung MRI (PREFUL MRI) and dynamic contrast-enhanced (DCE) MRI in evaluating perfusion defect percentage (QDP), as well as the agreement between PREFUL MRI and ¹²⁹Xe MRI in assessing ventilation defect percentage (VDP).

Method

A systematic search was conducted in the Medline, Embase and Cochrane Library databases to identify relevant studies comparing QDP and VDP measured by DCE MRI and ¹²⁹Xe MRI compared with PREFUL MRI. Meta-analytical techniques were applied to calculate the pooled weighted bias, limits of agreement (LOA) and correlation coefficient. The publication bias was assessed using Egger’s regression test, while heterogeneity was assessed using Cochran’s Q test and Higgins I² statistic.

Results

A total of 399 subjects from 10 studies were enrolled. The mean difference and LOA were −2.31% (−8.01% to 3.40%) for QDP and 0.34% (−4.94% to 5.62%) for VDP. The pooled correlations (95% CI) were 0.65 (0.55 to 0.73) for QDP and 0.72 (0.61 to 0.80) for VDP. Furthermore, both QDP and VDP showed a negative correlation with forced expiratory volume in 1 s (FEV₁). The pooled correlation between QDP and FEV₁ was −0.51 (−0.74 to −0.18), as well as between VDP and FEV₁ was −0.60 (−0.73 to −0.44).

Conclusions

PREFUL MRI is a promising imaging for the assessment of lung function, as it demonstrates satisfactory deviations and LOA when compared with DEC MRI and ¹²⁹Xe MRI.

PROSPERO registration number

CRD42023430847.

WHAT IS ALREADY KNOWN ON THIS TOPIC.

WHAT THIS STUDY ADDS

• The PREFUL MRI demonstrates high accuracy and good consistency in quantifying pulmonary perfusion defects and ventilation defects, showing a strong correlation with existing clinically reliable methods. It holds promise as a long-term pulmonary function assessment method for patients with chronic lung diseases.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The application of PREFUL MRI in pulmonary diseases provides patients with a convenient and non-invasive method for lung functional imaging, enabling early diagnosis and treatment assessment of pulmonary conditions.

Introduction

Pulmonary diseases are closely linked to structural changes in the lungs, making morphological evaluation crucial for patient assessment. However, during the early stages, the disease process predominantly affects pulmonary function while exhibiting limited modifications in pulmonary morphology.¹ Spirometry is widely used to assess the ventilatory function, which provides a global assessment of lung function but exhibits limited sensitivity to early functional changes.² Lung functional imaging provides the capability to identify regional lung lesions, enabling enhanced evaluation of lung diseases prior to the manifestation of global functional and morphological alterations.³ Chronic obstructive pulmonary disease (COPD) is a heterogeneous disease, and patients with different phenotypes respond differently to therapeutic drugs. Functional lung imaging contributes to the analysis of disease phenotypes in patients with COPD, providing potential benefits for drug screening and optimising treatment.⁴

The development of hyperpolarised noble gas MRI imaging as a viable technique for ventilation imaging has been underway.^{5 6} Such using ¹²⁹Xe ventilation MRI, it is possible to calculate the ventilation defect percentage (VDP), which represents the proportion of the lung that is non-ventilated.⁷ VDP has been proven to be a reliable and sensitive clinical outcome measure for assessing longitudinal lung disease changes in patients with cystic fibrosis (CF).^{8 9} Nevertheless, the clinical adoption of this technique has been limited due to its reliance on specialised polariser equipment. Dynamic contrast-enhanced (DCE) MRI is commonly used to assess pulmonary perfusion and volume, enabling quantification of the perfusion defect percentage (QDP).^{10 11} DCE MRI has demonstrated its capability to offer quantitative assessments and has been successfully employed in the evaluation of various diseases, including COPD, CF and chronic thromboembolic pulmonary hypertension (CTEPH).11,13 Furthermore, DCE MRI has been demonstrated to discriminate idiopathic pulmonary fibrosis (IPF) and serve as a biomarker for the progression of IPF disease.¹⁴ However, DCE MRI requires breath holding and injection of a gadolinium-containing contrast agent, and the long-term effects of gadolinium deposition in the brain are unknown.¹⁵

Fourier decomposition (FD) MRI has been shown to provide quantitative assessment of pulmonary ventilation under free breathing and without the need for contrast media.¹⁶ Recently, a postprocessing technique, phase-resolved functional lung (PREFUL) imaging, has been introduced to enhance temporal resolution and enable the quantitative evaluation of regional perfusion and ventilation dynamics.¹⁷ Distinguishing from FD MRI, PREFUL is different in several aspects: first, PREFUL performs perfusion/ventilation filtering with broad high-pass/low-pass filtering, while FD employs peak integration; second, PREFUL computes ventilation amplitudes in the time domain, contrasting with FD’s frequency domain computation; additionally, PREFUL sorts images based on their phase to reconstruct cardiac/ventilation cycles with enhanced temporal resolution.^{17 18} Extensive investigations have demonstrated consistent measurement outcomes between PREFUL MRI and DCE MRI in detecting regional pulmonary QDP across various lung diseases.^{11 19 20} Similarly, there is concordance between PREFUL MRI and ¹²⁹Xe MRI in assessing regional VDP.²¹ Additionally, prior studies have revealed a significant correlation between PREFUL MRI and spirometry-derived forced expiratory volume in 1 s (FEV₁).^{13 20 21} Although multiple studies have evaluated the accuracy of PREFUL MRI in assessing lung function, each study included only a small number of patients, necessitating caution in making generalisations and broader interpretations.

Therefore, the primary objective of this meta-analysis was to systematically assess the agreement and correlation of PREFUL MRI with DCE MRI and ¹²⁹Xe MRI of lung function parameters. Additionally, we also assessed the correlation of QDP and VDP derived from PREFUL MRI with the FEV₁ derived from spirometry.

Methods

This study followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses.²² This study was registered in PROSPERO with ID number CRD42023430847.

Search strategy

A systematic search was conducted in Medline, Embase and Cochrane Library databases to identify relevant articles that compared QDP and VDP as measured by PREFUL MRI, DCE MRI and ¹²⁹Xe MRI. The search was limited to articles published until 27 October 2023. Two authors independently carried out the search process, and any discrepancies were resolved through mutual agreement and consensus.

An exhaustive search of the literature was conducted using a combination of keywords, either in MeSH or ‘free text’ terms: ‘phase-resolved functional lung magnetic resonance imaging’ OR ‘PREFUL MRI’ OR ‘free‐breathing proton magnetic resonance imaging’ OR ‘free-breathing ¹H magnetic resonance imaging’ OR ‘dynamic contrast‐enhanced magnetic resonance imaging’ OR ‘DCE MRI’ OR ‘¹²⁹Xe magnetic resonance imaging’ OR ‘¹²⁹Xe MRI’ AND ‘perfusion defect percentage’ OR ‘QDP’ OR ‘ventilation defect percentage’ OR ‘VDP’ OR ‘forced expiratory volume in the first second’ OR ‘FEV₁’. For a full comprehensive search strategy, see online supplemental file 1.

Study selection

The inclusion criteria for study selection were determined according to the following criteria: (1) studies provided detailed information on the MRI scan parameters, including the scan sequence, image processing software and processing method; (2) studies included comprehensive descriptions of the study methodology and demographic data of the participants; (3) Bland-Altman analysis was employed to assess the consistency between the different MRI imaging methods; (4) Spearman or Pearson coefficient was used to evaluate the correlation between the two modalities; (5) studies included the provision of QDP or VDP parameters for PREFUL MRI, QDP parameter for DCE MRI, VDP parameter for ¹²⁹Xe MRI and FEV₁ measurement obtained through spirometry. Exclusion criteria for the meta-analysis involved the exclusion of animal studies and papers written in languages other than English. Additionally, abstracts, letters, editorials and meta-analyses were excluded due to lacking Bland-Altman or Pearson analysis data. The meta-analysis included both prospective and retrospective studies, as long as they met the inclusion criteria. The study population was not restricted, encompassing both adults and children, as well as patients and healthy volunteers, as long as they fulfilled the inclusion criteria.

Data extraction

The following data were extracted by two investigators: (1) information related to the articles, such as authors, publication journal, the total number of subjects and demographic and disease characteristics of participants; (2) MRI acquisition sequences and protocols; (3) postprocessing analysis methods employed for the calculation of QDP and VDP, along with the dedicated software used for postprocessing; (4) outcomes derived from the Bland-Altman analysis, including the mean difference (MD) and limits of agreement (LOA) for QDP between PREFUL MRI and DCE MRI, as well as VDP between PREFUL MRI and ¹²⁹Xe MRI; (5) Spearman or Pearson correlation coefficient (r) values of QDP and VDP between the two imaging methods, and the correlation between FEV₁ and QDP and VDP. The Spearman coefficients reported in the publications were transformed into Pearson coefficients.^{23 24} Furthermore, Fisher’s r-to-z transformation was applied to obtain the Pearson coefficient values.²⁵

Quality assessment

The assessment of bias risk in each included study was conducted using the modified Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool via Review Manager V.5.3.²⁶ This tool is frequently employed for the assessment of the quality of diagnostic studies.

Statistical analysis

The MD, LOA and r values from each study were weighted based on the sample size using a random-effects model. A modified forest plot was created to visualise the meta-analysis results for bias and LOA of each parameter. Pooled correlation coefficients for each parameter were displayed in a forest plot. Heterogeneity among the included studies was assessed using the Cochrane Q test and expressed as I². Moderate to severe heterogeneity was considered present if the p value was less than 0.05 for the Q test or if I² exceeded 50%. Publication biases were evaluated using Egger’s test and represented through funnel plots. All statistical analyses were conducted using R (V.4.2.2, R Foundation for Statistical Computing, Vienna, Austria) with the meta package and Review Manager (V.5.3, The Cochrane Collaboration 2021, The Nordic Cochrane Centre, Copenhagen, Denmark).

Results

Study selection

Following the systematic search, a total of 105 citations from Medline, 456 from Embase and 6 from Cochrane were identified. After removing duplicates, 508 potentially eligible articles remained. Through the screening of titles and abstracts, 413 studies were excluded for irrelevant topics, animal models, case reports, reviews or guidelines. A total of 95 articles underwent full-text review. After a thorough evaluation based on the inclusion criteria, an additional 85 publications were excluded for reasons such as being non-English, focusing on other parameters, having unextractable data or potentially containing duplicate patient data. Finally, 10 studies met the inclusion criteria and were included in the meta-analysis. The flow chart illustrating the process of literature search and study selection can be found in figure 1.1319,21 27

Figure 1. Flow chart depicting the study review process in accordance with the guidelines provided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA).

Study characteristics

The meta-analysis comprised a total of 10 studies, including 399 participants: 101 with CF, 150 with COPD, 66 with CTEPH and 82 healthy volunteers. QDP was analysed in 4 studies, VDP in 4 studies and FEV₁ in 8 studies. Table 1 presents a summary of the characteristics of the enrolled studies and details regarding MRI image acquisition. Among the studies, a 1.5-tesla (T) system was used in all except for Munidasa et al²⁹ and Couch et al,²⁸ which used a 3 T system. The most commonly used acquisition sequence for PREFUL MRI was the spoiled gradient echo (SPGR), and all DCE MRI acquisitions employed the time-resolved angiography with stochastic trajectories sequence.

Table 1. Characteristics of included studies.

First author	Year	Journal	Study design	Study subject	Total	Age (year)	Sex (M:F)	PREFUL MRI	DCE MRI	129Xe MRI
Moher Alsady32	2023	JMRI	Prospective	CTEPH	45	72±9*	–	1.5 T (Avanto, Siemens); SPGR sequence	1.5 T (Avanto, Siemens); TWIST sequence
Behrendt19	2022	Pulmonary Circulation	Prospective	CF	16	30 (2–36)†	5:11	1.5 T (Avanto, Siemens); SPGR sequence	1.5 T (Avanto, Siemens); TWIST sequence
Behrendt13	2020	JMRI	Retrospective	CF	14	46–76‡	3:11	1.5 T (Avanto or Aera, Siemens); SPGR sequence	1.5 T (Avanto or Aera, Siemens); TWIST sequence
				COPD	20	5–22‡	17:3
				CTEPH	21	43–82‡	12:9
Kaireit20	2019	JMRI	Prospective	COPD	47	66 (57–70)†	21:26	1.5 T (Avanto, Siemens); 3D-SPGR sequence	1.5 T (Avanto, Siemens); TWIST sequence
Munidasa29	2023	MRM	Prospective	CF	15	15 (13–16)†	8:7	3 T (Prisma, Siemens); 3D-SPGR sequence		3 T (Prisma, Siemens); FSGRE sequence
				Healthy	7	15 (12–15)†	3:4
Marshall21	2023	JMRI	Prospective	CF	31	23.3±10.2*	13:18	1.5 T (Avanto, Siemens); SPGR sequence		1.5 T (Avanto, Siemens); SSFP sequence
				Healthy	6	25.8±4.4*	0:6
Kaireit20	2019	JMRI	Retrospective	CF	8	55 (20–70)†	20:14	1.5 T (Avanto, Siemens); SPGR sequence		1.5 T (Avanto, Siemens);SOSGR sequence
				COPD	20
				Healthy	6
Couch28	2021	AR	Prospective	CF	17	13.0±2.7*	–	3 T (Prisma, Siemens); SPGR sequence		3 T (Prisma, Siemens); FSGRE sequence
				Healthy	10
Voskrebenzev31	2022	Radiology: Cardiothoracic Imaging	Prospective	COPD	50	64 (46–78)†	35:15	1.5 T (Avanto, Siemens); SPGR sequence
Klimeš30	2021	JMRI	Prospective	COPD	13	59 (50–71)†	5:8	1.5 T (Avanto, Siemens); 3D-SPGR sequence
				Healthy	53	27 (22–33)†	24:29

^*dData showedn as mean±standard deviationSD.

^†Data presented as median (25th percentile to –75th percentiles), .

^‡dData expressed as min to max, .

ARAcademic RadiologyCFcystic fibrosisCOPDchronic obstructive pulmonary diseaseCTEPHchronic thromboembolic pulmonary hypertensionDCEdynamic contrast-enhancedFfemaleFSGREfast spoiled gradient recalled echoJMRIJournal of Magnetic Resonance ImagingMmaleMRMMagnetic Resonance in MedicinePREFULphase-resolved functional lungSOSGRsingle shot gradient echoSPGRspoiled gradient echoSSFPsteady-state free precessionTteslaTWISTtime‐resolved angiography with stochastic trajectories

Study quality

The results of the quality assessment, conducted using the QUADAS-2 checklist, are presented in figure 2. Five studies reported enrolling consecutive patients, while in five studies, the method of patient enrolment was not clearly explained. The risk of bias in the ‘index test’ or ‘reference standard’ domains was assessed as ‘unclear’ in six studies due to the absence of information regarding whether the MRI results were interpreted with knowledge of each other. The majority of studies (89%) demonstrated a low risk of bias in the ‘flow’ and ‘timing’ domains. Applicability concerns across all domains were rated as ‘low’.

Figure 2. Quality assessment of included studies by modified Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Each bar shows the percentage of studies with high (red), unclear (yellow) and low (green) risks of bias and applicability of concerns.

Publication bias

Figure 3 displays the funnel plots for each parameter. The funnel plots for all parameters exhibited a relatively symmetric shape, suggesting no significant publication bias (p>0.05), except for the agreement of QDP, which showed remarkable publication bias (p=0.017).

Figure 3. Funnel plots were employed to identify potential publication bias. Each circle in the plots represents a study included in the analysis. The p values derived from Egger’s test, which measures the asymmetry of the funnel plot, are presented. (A) Agreement of QDP. (B) Agreement of VDP. (C) Correlation of QDP. (D) Correlation of VDP. (E) Correlation of QDP with FEV₁. (F) Correlation of VDP with FEV₁. FEV₁, forced expiratory volume in 1 s; MD, mean difference; QDP, perfusion defect percentage; VDP, ventilation defect percentage.

Agreement and correlation for measurements of QDP

Among the included studies, three reported the MD and LOA for QDP as assessed by Bland-Altman analysis between PREFUL MRI and DCE MRI. One of these studies analysed three different groups of patient results.¹³ The MD of QDP was −2.31%, with a pooled LOA ranging from −8.01% to 3.40% (figure 4A). Significant heterogeneity was observed (I²=82%, p<0.01). In addition, the pooled correlation coefficient for QDP was 0.65 (95% CI 0.55 to 0.73), with low heterogeneity (I²=0%, p=0.44). The result of the modified forest plot is shown in figure 5A.

Figure 4. (A) Forest plots for agreement between PREFUL MRI and DCE MRI for QDP. (B) Forest plots for agreement between PREFUL MRI and ¹²⁹Xe MRI for VDP. CF, cystic fibrosis; COPD, chronic obstructive pulmonary disease; CTEPH, chronic thromboembolic pulmonary hypertension; DCE, dynamic contrast-enhanced; LOA, limits of agreement; MD, mean difference; PREFUL, phase-resolved functional lung; QDP, perfusion defect percentage; VDP, ventilation defect percentage.

Figure 5. (A) Forest plots for correlation between PREFUL MRI and DCE MRI for QDP. (B) Forest plots for correlation between PREFUL MRI and ¹²⁹Xe MRI for VDP. (C) Forest plots for correlation between QDP and FEV₁. (D) Forest plots for correlation between VDP and FEV₁. CF, cystic fibrosis; COPD, chronic obstructive pulmonary disease; CTEPH, chronic thromboembolic pulmonary hypertension; DCE, dynamic contrast-enhanced; FEV₁, forced expiratory volume in 1 s; PREFUL, phase-resolved functional lung; QDP, perfusion defect percentage; VDP, ventilation defect percentage.

Agreement and correlation for measurements of VDP

Four studies investigated the correlation of VDP between PREFUL MRI and ¹²⁹Xe MRI. However, only three studies reported the agreement for VDP assessment using Bland-Altman analysis. The overall MD for VDP was 0.34%, and the pooled LOA ranged from −4.94% to 5.62%. Notably, there was significant heterogeneity observed in the finding (I²=89%, p<0.01). Furthermore, the pooled correlation coefficient for VDP was 0.72 (95% CI 0.61 to 0.80), with mild heterogeneity (I²=40%, p=0.14). Figures4B 5B provide forest plots of these results.

Correlation between PREFUL MRI and FEV₁ of spirometry

The meta-analysis including four studies revealed a pooled correlation coefficient of −0.51 (95% CI −0.74 to −0.18) between QDP and FEV₁. The analysis also indicated significant heterogeneity among the studies (I²=83%, p<0.01). Moreover, in seven studies, the pooled correlation coefficient between VDP and FEV₁ was −0.60 (95% CI −0.73 to −0.44), with a significant heterogeneity (I²=70%, p<0.01). All the results are shown in figure 5C,D.

Discussion

To the best of our understanding, this meta-analysis is the first systematic investigation of pulmonary function assessment using PREFUL MRI in comparison to DCE MRI and ¹²⁹Xe MRI. The analysis includes a total of 10 studies involving 399 subjects. The findings indicate that the evaluation of QDP on PREFUL MRI demonstrates good agreement (MD, −2.31%; LOA, −8.01% to 3.40%) and moderate correlations (r=0.65; 95% CI 0.55 to 0.73) with DCE MRI, although notable heterogeneities are observed among the studies. Furthermore, the results reveal that VDP on PREFUL MRI exhibits favourable agreement (MD, 0.34%; LOA, −4.94% to 5.62%) and strong correlations (r=0.72; 95% CI 0.61 to 0.80) with ¹²⁹Xe MRI. These findings suggest that free-breathing, non-contrast agents PREFUL MRI may allow for pulmonary perfusion and ventilation assessment comparable to DCE MRI and ¹²⁹Xe MRI.

DCE MRI is a well-established and widely used technique for assessing lung perfusion. Recent studies have demonstrated that DCE MRI achieves comparable diagnostic accuracy when compared with planar scintigraphy and ventilation/perfusion scans.³³ Furthermore, investigations have shown significant correlations and spatial agreement between perfusion-weighted PREFUL MRI and DCE MRI measurements.³⁴ However, it is worth noting that previous research findings have been based on limited sample sizes. In our meta-analysis, we provide robust evidence supporting the clinical practice of PREFUL MRI. Notably, when considering the study by Behrendt et al,¹³ we observed a larger pooled MD and the LOA. This can be attributed to incomplete breath holding during DCE data acquisition, resulting in residual movement and artificially elevated perfusion values near the diaphragm.

The majority of studies used an SPGR sequence to acquire PREFUL algorithm images at 1.5 T MR, typically capturing 3–8 coronal slices. Postprocessing procedures involved registering each coronal slice image to a reference image in an intermediate respiratory position. Furthermore, automatic segmentation techniques were employed to delineate the lung parenchyma and eliminate pulmonary vessels with Otsu methods for variable thresholding.¹⁰ Through the combination of automatically determined regions of interest along with heart frequencies, the complete cardiac cycle was reconstructed using a phase-sorting algorithm. For each slice, the perfusion weighting phase was automatically selected, and the perfusion value was normalised using the signal from the whole blood voxel defined in the middle slice.¹³ Notably, the determination of the threshold plays a critical role in the calculation of QDP. In Pöhler et al’s study, the threshold of PREFUL MRI for healthy perfusion values was defined as 1% of the full blood voxel signal value.³⁵ However, in the research conducted by Behrendt et al,¹⁹ the threshold of PREFUL MRI was set at 2% and the threshold of DCE MRI was 1.75% for QDP analysis. Additionally, as taken in the study by Schiwek et al,¹⁰ the calculation method involving variable thresholds rather than fixed thresholds is being increasingly adopted for both PREFUL MRI and DCE MRI. These differences in threshold determination across studies could potentially account for variations in the MD observed in our meta-analysis.

Hyperpolarised gas MRI is a highly sensitive and specific imaging technique that directly visualises the distribution of ventilation within the lung.³⁶ On the other hand, PREFUL MRI is a method that uses regional lung motion during the breathing cycle to quantify regional ventilation (RV). It identifies ventilation defects, characterised by minimal signal changes or altered signal dynamics between inspiration and expiration. Although our meta-analysis demonstrates good agreement and strong correlation between PREFUL VDP and ¹²⁹Xe VDP, some variations persist, particularly in the study conducted by Couch et al.²⁸ These differences can be attributed, at least in part, to the influence of time-of-flight inflow effects, resulting in enhanced visibility of pulmonary vessels in single-slice, free-breathing ¹H images.

The computation of VDP in PREFUL MRI primarily relies on two ventilation-derived maps: the RV map and the flow-volume loop (FVL) correlation map. The RV map quantifies the signal change between end inspiration and end expiration while correcting for volume errors resulting from image registration.³⁷ On the other hand, the FVL correlation map cross-correlates each voxel FVL in the lung parenchyma with a healthy reference FVL to generate a voxel FVL cross-correlation map, which is subsequently used to calculate the FVL-based VDP value.³⁸ Generally, for VDP_RV, VDP is determined by identifying voxels with ventilation deficit, where all RV values below the 90th percentile multiplied by a factor of 0.4 are considered as such defects. Conversely, for VDP_FVL, a fixed threshold of 0.9 is used to identify ventilation defects.³⁸ In our meta-analysis, the VDP values derived from PREFUL MRI were calculated based on VDP_RV, which resulted in relatively consistent results among all included studies, with an MD of 0.34% (95% CI −4.94% to 5.62%) and a correlation coefficient of 0.73. The results indicate that the VDP measured by PREFUL and ¹²⁹Xe MRI was essentially consistent, with a difference of only 0.34% between the VDP values calculated by the two MRI imaging methods. This consistency in VDP outcomes may be attributed to the utilisation of VDP_RV as the basis for VDP calculation across the studies.

Previous studies have reported remarkable correlations between RV assessed by FD techniques and spirometry parameters, particularly FEV₁.^{35 39} Similarly, considerable correlations have been observed between PREFUL MRI-derived QDP and predicted FEV₁ for the whole lung.^{13 19 20} However, on combining the correlation coefficients from the included studies, our results found that the correlation between QDP and FEV₁ (r=−0.51; 95% CI −0.74 to −0.18) was relatively weaker compared with that between VDP and FEV₁ (r=−0.60; 95% CI −0.73 to −0.44). This observation can be explained by the fact that QDP represents a perfusion parameter while FEV₁ reflects a ventilation parameter. Additionally, it should be noted that PREFUL MRI, unlike global lung function measurements such as FEV₁, involves the calculation of QDP based on only 3–8 slices, resulting in incomplete coverage of the entire lung parenchyma.¹⁹

The results from our meta-analysis hold significant implications for the future clinical application of PREFUL MRI, considering the crucial role of lung function assessment in the diagnosis and prognosis of pulmonary diseases. Moreover, the measurement of ventilation and perfusion is of remarkable importance in monitoring the efficacy of pharmacological interventions in hyperinflated patients with COPD and predicting adverse outcomes such as chronic lung allograft dysfunction-related mortality or transplant failure in large prospective cohorts of lung transplant recipients.^{40 41} Finally, PREFUL MRI offers the advantage of simultaneously obtaining ventilation and perfusion parameters in a single examination, eliminating the need for breath-holding manoeuvres and contrast agent administration. This feature is particularly advantageous for patients with lung diseases who may have difficulties tolerating prolonged supine breath-holding examinations.

Several limitations are associated with this meta-analysis. First, the sample size was comparatively modest, consisting of only 10 studies. Moreover, there was variation in the literature enrolled when examining different lung function indicators. Second, the diverse range of disease states encompassed in this study, including both healthy volunteers and patients with various conditions such as CF, introduces potential limitations in extrapolating the results to specific populations. Third, substantial heterogeneity was observed in the meta-analysis, but due to the limited number of available studies, subgroup analyses could not be conducted to explore potential sources of heterogeneity. Lastly, the quantification of lung function parameters using PREFUL MRI does not directly measure FEV₁. Published studies have consistently shown a significant correlation between PREFUL lung function parameters and spirometry-based FEV₁. Therefore, this study does not directly compare PREFUL MRI with spirometry-based FEV₁.

Conclusions

PREFUL MRI is a promising innovative imaging for assessing lung function, as it exhibits satisfactory bias and LOA compared with DEC MRI and ¹²⁹Xe MRI. In the future, large-scale, multicentre, prospective studies will be needed to validate it.

Data availability statement

Data are available upon reasonable request.