Role of chest MRI for the diagnosis of malignant pulmonary nodules: a systematic review and a meta-analysis

Full Text

BACKGROUND

The solitary pulmonary nodule (SPN) is a single, delimited, rounded lesion with a diameter of <3 cm [1, 2]. SPN is completely surrounded by unchanged pulmonary tissue and is unrelated to atelectasis, the root of the lung, or mediastinum. This mass could be caused by benign processes such as hamartoma, infectious lesions, granulomatous inflammation, or malignant processes (primary lung cancer, metastatic disease, or lymphoma). Malignancy of nodules is assumed until proven otherwise [2].

Currently, computed tomography (CT) is the gold standard for assessing SPN and monitoring patients who are at risk of cancer [3]. CT, despite its many advantages, has a major disadvantage: high radiation exposure, which certainly increases in dynamic monitoring. With the development and improvement of hardware and software, the search for new alternative visualization methods becomes evident. Over the last two decades, research into the potential use of magnetic resonance imaging (MRI) for the diagnosis of chest diseases has resulted in a separate research area devoted to SPN detection by MRI. The advantage of MRI is the lack of ionizing radiation exposure and the optional quantitative assessment of the revealed changes even without the use of contrast agents.

The aim of the study was to compare the capabilities of chest MRI to standard CT for detecting malignant SPN.

METHODS

This work was designed according to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) standard [4].

Study type

Inclusion criteria:

• (I) Prospective case-control studies, retrospective case-control studies, prospective cohort studies, and retrospective cohort studies;
• (II) Studies that included a description of the use of MRI to detect malignant SPN;
• (III) CT as the reference method; and
• (IV) Availability of sensitivity and specificity data.

Exclusion criteria:

• (I) The full text of the paper was not available.
• (II) The study did not involve humans.
• (III) The study involved children.
• (IV) Case reports, case series studies, systematic reviews, and meta-analyses.
• (V) Combination of positron emission tomography (PET) and CT (PET/CT) or PET and MRI (PET/MRI); contrast-enhanced studies.
• (VI) Involvement of patients with pulmonary tuberculosis and other inflammatory lung diseases.

Participants

Patients over 18 years of age.

The review excluded patients for whom diagnostic data could not be obtained using standard reference methods (standard chest CT).

Interventions

Studies that assessed the ability of MRI and standard CT to detect SPN suspicious for malignancy.

Results

Primary results: numerical values of lung MRI sensitivity and specificity to assess the detection of malignant SPN.

Secondary results: identification of the most optimal MR pulse sequences.

Sources of information

The databases PubMed and Google Scholar were searched until April 7, 2021.

Search

Since PubMed takes approximately a month to assign the term MeSH to a published paper, two types of queries were used in the PubMed database, that is, MeSH library terms and keywords to search among recent articles:

“Magnetic Resonance imaging” [Mesh] or “MRI” and “Computed tomography” or “CT” and “Lung neoplasms” [Mesh] or “Solitary Pulmonary Nodule” [Mesh] and “Sensitivity” and “Specificity”;

“Lung MRI” or “chest MRI” and “Computed tomography” or “CT” and “lung cancer” or “Solitary Pulmonary Nodule”.

The query “MRI, CT, lung cancer, specificity, sensitivity” was used to search in the Google Scholar database.

Data collection and data items

Using the Google Spreadsheet service, a data extraction table was created. The authors had simultaneous and unrestricted access to the document. The following data was extracted from the selected papers by two researchers (O.Yu. Panina and Ye.A. Grik): title of the article, journal (or service for posting preprints), publication date, DOI, MRI protocol, MRI magnetic induction value, types of lesions revealed, sensitivity, specificity, and standard deviation for MRI and CT. After calculating the sensitivity and specificity indicators for each pulse sequence (PS) separately, the most effective values were included in the meta-analysis.

Three other researchers (E.S. Akhmad, Yu.N. Vasilyeva, and Yu.A. Vasilyev) verified the extracted data. All disagreements were resolved through discussion among the authors.

Risk of bias in selected studies

The authors used the QUADAS-2 (Quality Assessment of Diagnostic Accuracy Studies) checklist [5], which is recommended by the Agency for Healthcare Research and Quality (Cochrane Collaboration, US) for systematic reviews. Each of the selected papers was assessed based on four domains: patient selection, index test, reference test, and patient flow. For a detailed description of each domain and the judgment of the criteria used, see the Cochrane Handbook for Systematic Reviews of Interventions [6].

Statistical analysis

The selection of method for analyzing and grouping sensitivity and specificity data (random- or fixed-effects model) was performed according to the results of the study heterogeneity assessment. The χ² criterion and I² heterogeneity index were used to assess the statistical heterogeneity of the studies included in the meta-analysis. In the studies, statistically significant heterogeneity corresponds to p <0.10 in the χ² criterion and I >40%. Moreover, meta-analysis was performed using the RevMan 5.4.1 software package.

RESULTS

Based on search results in the PubMed and Google Scholar databases, 168 selected papers were imported into the Mendeley Reference Manager software library. Following a second check for inclusion and exclusion criteria and text review, 33 papers remained (Fig. 1).

 

Fig. 1. Flow diagram

 

A text review of the remaining 33 papers [7–38] revealed that two studies performed contrast-enhanced CT [23, 24], which is not an appropriate type of intervention. Furthermore, the sensitivity and specificity data for CT were completely absent in ten papers [20–22, 25–29, 31, 32], so these studies were excluded. The meta-analysis included papers in which a low-dose CT was used as a control study [9, 10, 16] rather than an exclusion criterion. Thus, the meta-analysis included 21 studies (see Fig. 1).

The selected studies included 1188 patients and contained information on lung MRI and CT scans. Sensitivity data for MRI and CT were presented in all papers; however, specificity rates were not reported in three articles [8, 16, 30]. The majority of the studies were performed on tomographs with magnetic field induction of 1.5 T (Table. 1).

 

Table 1. Characteristics of studies included in the meta-analysis

No.	Studies	Year	Magnetic field induction, T	MRI model and manufacturer	MRI PS
1	Both [7]	2005	1,5	Magnetom Vision, Siemens	VIBE, HASTE, T2TSE
2	Bruegel [8]	2007	1,5	Magnetom Sonata, Siemens	STIR
3	Chang [19]	2015	1,5	Intera Achieva, Philips	SS-TSE-HF
4	Cieszanowski [30]	2016	1,5	Magnetom Avanto, Siemens	T2TSE, T2-STIR, T2-HASTE
5	Dewes [33]	2016	3,0	Magnetom Prisma, Siemens	CAIPIRINHA-VIBE
6	Fatihoğlu [34]	2019	1,5	Magnetom Aera, Siemens	DWI (ADC)
7	Heye [35]	2012	1,5	Avanto, Siemens	VIBE, HASTE
8	Huang [39]	2020	1,5	Magnetom Aera, Siemens	UTE free-breathing
9	Koo [36]	2019	3,0	Magnetom Skyra, Siemens	T2FSE
10	Koyama [37]	2008	1,5	Intera, Philips	STIR
11	Koyama [38]	2015	1,5	Achieva, Philips	DWI (ADC)
12	Meier-Schroers [9]	2016	1,5	Ingenia, Philips	T2FSE
13	Meier-Schroers [10]	2019	1,5	Ingenia, Philips	T2STIR
14	Ohno [11]	2017	3,0	Vantage Titan, Canon Medical Systme	UTE
15	Regier [12]	2011	1,5	Achieva, Philips	DWI (ADC)
16	Satoh [13]	2008	1,5	Intera NovoDual, Philips	DWI (ADC)
17	Schaefer [14]	2006	1,0	Magnetom Expert, Siemens	PDWI
18	Schroeder [15]	2005	1,5	Magnetom Sonata, Siemens	HASTE
19	Sommer [16]	2014	1,5	Magnetom Avanto, Siemens	HASTE
20	Vogt [17]	2004	1,5	Magnetom Sonata, Siemens	HASTE
21	Yi [18]	2007	3,0	Achieva, Philips	T1WI 3D TFE*

Note. MRI PS: pulse sequences of magnetic resonance imaging

 

Risk of bias

Eleven studies adequately reported the index test and reference test data [7, 8, 10–13, 15–18, 33]. The primary sources of errors were the index test (MRI) and its interpretation (Fig. 2). Some studies lacked complete data to assess the risk of bias, for example, whether the index test results were interpreted without knowing the results of the reference test, or whether interpretation of the reference test was impossible without knowing the results of the index test. Moreover, the risk of bias was characteristic of studies that were more likely to be published if an effect was found as opposed to those that did not. However, participants in all studies met the protocol criteria for this review.

 

Fig. 2. Histogram of the risk of bias

 

The meta-analysis revealed statistically significant heterogeneity (p <0.00001) using the χ² criterion and the heterogeneity index (I²=99%) for sensitivity and specificity. In this case, the random-effects method was used to analyze the data.

Diagnostic accuracy of chest MRI

In each of the 21 studies, MRI was compared to the reference method. Sensitivity for MRI ranged from 70.4% to 100%, while the specificity ranged from 60.6% to 100% (Fig. 3). The mean MRI sensitivity was 88.3%, while the mean MRI specificity was 71.3%. In studies where the standard deviation parameters for sensitivity and specificity were not specified, the calculation was performed by estimating the values of the indicators [40].

 

Fig. 3. Forest plot of grouped data for specificity (a) and sensitivity (b) [40]

Note. SMD: standardized mean difference; CI: confidence interval

 

Table 2 shows the characteristics that were included in the meta-analysis of studies with the highest sensitivity and specificity values for MRI, which are comparable to CT.

DISCUSSION

The results of this meta-analysis show that MRI has lower mean specificity and sensitivity when compared to CT. In the majority of studies included in the meta-analysis, the sensitivity and specificity of chest CT for detecting SPN was 100%. Since CT was used as a reference test, only three studies had lower rates. For chest MRI, 5 of the 21 studies had 100% sensitivity, and only 2 studies had 100% sensitivity and specificity.

When the results of the meta-analysis were examined, high sensitivity rates were observed in studies that calculated overall sensitivity and specificity rates for the entire MR protocol rather than separately for each PS (see Table 2). This phenomenon demonstrates the peculiarity of the meta-analysis for MRI as a method in which the signal characteristics are assessed in conjunction with the scanning protocol. These examples may indicate a lack of research into MRI capabilities in the differential diagnosis of SPN, the need for studies of current PS, and the careful tuning of routine PS on a tomograph. This approach will allow for greater MRI efficiency in detecting SPN and studying their characteristics, which is especially important in lung cancer diagnosis.

 

Table 2. Characteristics of studies with the highest sensitivity and specificity values

No.	Author, year of study	Sensitivity (general index)	Specificity (general index)	MRI PS	Magnetic field induction, T
1	Both, 2005 [7]	100	100	VIBE, HASTE, T2TSE	1,5
2	Cieszanowski, 2016 [30]	100	-	T2TSE, T2-STIR, T2-HASTE	1,5
3	Meier-Schroers, 2016 [9]	100	100	T2FSE	1,5
4	Regier, 2011 [12]	97	92,3	DWI (ADC)	1,5
5	Heye, 2012 [35]	100	100	VIBE, HASTE	1,5
6	Schaefer, 2006 [14]	100	75	PDWI	1,5

Note. MRI PS: pulse sequences of magnetic resonance imaging

 

Lung cancer continues to be the leading cause of death worldwide, including in the Russian Federation, and it is a serious social and economic problem [41, 42]. The presence of cancer among detected SPN ranges from 10% to 70% [2]. In some countries, low-dose CT is performed in high-risk groups as part of screening. Currently, the coverage of the screening program remains low, and the criteria for inclusion of patients are limited to ensure its economic viability. Thus, many patients will still be diagnosed after the onset of symptoms, rather than at an early stage of disease development and at a high cost of misdiagnosis [41]. Furthermore, radiologists and clinicians continue to face difficulties in monitoring and managing ambiguous SPN. Accordingly, only a comprehensive approach is always used for diagnosis, patient routing, and selection of optimal management and treatment strategies [43].

This meta-analysis revealed an alternative approach for assessing SPN that are suspicious for cancer. In addition, the emphasis was on standard studies that did not make use of contrast enhancement.

Limitations

There were several limitations to this study. For various reasons, the meta-analysis included data with lesions larger than 6 mm. First, nodule sizes greater than 6 mm are the most common in selected studies; second, nodules smaller than 6 mm have a fairly low risk of malignancy according to recent data from Fleischner Society [3]. In addition, the meta-analysis did not compare MRI with histologic data, which could be considered a limitation of the study.

CONCLUSIONS

MRI has the sensitivity and specificity needed for additional diagnosis of SPN that are suspicious for malignancy and revealed by CT. The mean MRI sensitivity was 88.3%, while the mean MRI specificity was 71.3%.

MRI is a non-ionizing imaging method that can be used as an additional technique in the evaluation of various PS while resolving controversial cases.

Further research into the most efficient PS, the feasibility of contrast enhancement, and new technological solutions for high-quality diagnostics of SPN is necessary.

ADDITIONAL INFORMATION

Funding source. The study had no sponsorship.

Competing interests. The authors declare that they have no competing interests.

Authors’ contribution. Yury A. Vasilev — research concept, writing the manuscript, analysis and expert assessment of research results; Оlga Yu. Panina — research concept, search for publications, processing of the results, writing the manuscript; Evgeniia A. Grik — search for publications, processing of the results, Kate A. Akhmad — processing of the results, systematization and final editing of the review; Yulia N. Vasileva — analysis and expert assessment of research results, systematization and final editing of the review. All authors made a substantial contribution to the conception of the work, acquisition, analysis, interpretation of data for the work, drafting and revising the work, final approval of the version to be published and agree to be accountable for all aspects of the work.

About the authors

Yuriy A. Vasilev

Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of Moscow Health Care; City Clinical Oncological Hospital No. 1

Email: dr.vasilev@me.com
ORCID iD: 0000-0002-0208-5218
SPIN-code: 4458-5608

MD, Cand. Sci. (Med)

Russian Federation, 24/1 Petrovka str., 127051, Moscow; Moscow

Olga Y. Panina

Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of Moscow Health Care; City Clinical Oncological Hospital No. 1; Moscow State University of Medicine and Dentistry named after A.I. Evdokimov

Email: o.panina@npcmr.ru
ORCID iD: 0000-0002-8684-775X
SPIN-code: 5504-8136
Scopus Author ID: 57219837311

Junior Scientist Researcher

Russian Federation, 24/1 Petrovka str., 127051, Moscow; Moscow; 20, p. 1, Delegatskaya str., Moscow, 127473

Evgeniia A. Grik

Moscow State University of Medicine and Dentistry named after A.I. Evdokimov

Email: evgeniyagrik@gmail.com
ORCID iD: 0000-0002-7908-3982
SPIN-code: 5558-7307

MD

Russian Federation, 20/1, Delegatskaya str., Moscow, 127473

Kate S. Akhmad

Research and Practical Clinical Center for Diagnostics and Telemedicine Technologies of Moscow Health Care

Email: e.ahmad@npcmr.ru
ORCID iD: 0000-0002-8235-9361
SPIN-code: 5891-4384
Russian Federation, 24/1, Petrovka street,127051 Moscow

Yulia N. Vasileva

Moscow State University of Medicine and Dentistry named after A.I. Evdokimov

Author for correspondence.
Email: drugya@yandex.ru
ORCID iD: 0000-0003-4955-2749
SPIN-code: 9777-2067

MD, Cand. Sci. (Med.)

Russian Federation, 20/1, Delegatskaya str., Moscow, 127473

References

Supplementary files