Functional MR Imaging in Chest Malignancies




With recent advances in MR imaging, its application in the thorax has been feasible. The performance of both morphologic and functional techniques in the evaluation of thoracic malignances has improved not only differentiation from benign etiologies but also treatment monitoring based on a multiparametric approach. Several MR imaging–derived parameters have been described as potential biomarkers linked with prognosis and survival. Therefore, an integral approach with a nonradiating and noninvasive technique could be an optimal alternative for evaluating those patients.


Key points








  • Functional MR imaging is an emerging technique in the thorax providing an integral assessment of several characteristics of chest neoplasms by means of prognostic and reproducible quantitative parameters.



  • Functional MR imaging represents an alternative to fludeoxyglucose F 18 ( 18 FDG)-PET/CT in the differentiation of benign and malignant entities, allowing the accurate staging and treatment response monitoring.



  • Although there is only initial experience, the performance of functional MR imaging in the heart and chest wall can enhance its utility in the thorax.






Introduction


Several tumor characteristics may be assessed using functional imaging techniques ( Table 1 ). Currently, the most widely accepted tumor metabolism imaging is performed with 18 FDG-PET/CT.



Table 1

Tumor characteristics and functional imaging techniques




































Technique Biophysical Basis Quantitative Parameters
DWI–MR imaging Cellularity and tortuosity of interstitial space ADC
DKI Tissue microstructure D app ; K app
IVIM Blood flow and tissue diffusivity f, D, D*
DCE–MR imaging Blood flow and permeability K trans , V e , V p , K ep , AUC
BOLD Hypoxia Tissue R2 relaxivity
ASL Blood flow Flow quantification
Spectroscopy Metabolism Ratio of choline to other metabolites

Abbreviations: ASL, arterial spin labeling; AUC, area under the curve; D*, perfusion contribution to signal decay; D app , estimation of diffusion coefficient using nongaussian model; f, perfusion fraction; K app , apparent diffusional kurtosis; K ep , elimination rate; K tran s , transfer rate; V e , extracellular volume; V p , vascular space volume.

Modified from Luna A, Cunha GM, Sanchez-Sanchez R, et al. Overview of functional Imaging techniques for liver malignancies in current clinical practice or in a very near future. In: Luna A, Vilanova JC, Da Cruz LC, et al, editors. Functional imaging in oncology: clinical applications. vol. 2. Berlin: Springer-Verlag; 2014. p. 953.


Although the application of thoracic MR imaging is technically demanding, its great advantage is the possibility of integrating several morphologic and functional techniques in the evaluation of different tumor characteristics. From these techniques, several quantitative and reproducible parameters can be calculated, some of which serve as widely accepted prognostic biomarkers. This article reviews currently available techniques and some of the more promising potential applications of thoracic functional MR imaging.




Introduction


Several tumor characteristics may be assessed using functional imaging techniques ( Table 1 ). Currently, the most widely accepted tumor metabolism imaging is performed with 18 FDG-PET/CT.



Table 1

Tumor characteristics and functional imaging techniques




































Technique Biophysical Basis Quantitative Parameters
DWI–MR imaging Cellularity and tortuosity of interstitial space ADC
DKI Tissue microstructure D app ; K app
IVIM Blood flow and tissue diffusivity f, D, D*
DCE–MR imaging Blood flow and permeability K trans , V e , V p , K ep , AUC
BOLD Hypoxia Tissue R2 relaxivity
ASL Blood flow Flow quantification
Spectroscopy Metabolism Ratio of choline to other metabolites

Abbreviations: ASL, arterial spin labeling; AUC, area under the curve; D*, perfusion contribution to signal decay; D app , estimation of diffusion coefficient using nongaussian model; f, perfusion fraction; K app , apparent diffusional kurtosis; K ep , elimination rate; K tran s , transfer rate; V e , extracellular volume; V p , vascular space volume.

Modified from Luna A, Cunha GM, Sanchez-Sanchez R, et al. Overview of functional Imaging techniques for liver malignancies in current clinical practice or in a very near future. In: Luna A, Vilanova JC, Da Cruz LC, et al, editors. Functional imaging in oncology: clinical applications. vol. 2. Berlin: Springer-Verlag; 2014. p. 953.


Although the application of thoracic MR imaging is technically demanding, its great advantage is the possibility of integrating several morphologic and functional techniques in the evaluation of different tumor characteristics. From these techniques, several quantitative and reproducible parameters can be calculated, some of which serve as widely accepted prognostic biomarkers. This article reviews currently available techniques and some of the more promising potential applications of thoracic functional MR imaging.




Functional MR imaging in the chest: technical considerations


Diffusion-Weighted Imaging


Diffusion-weighted imaging (DWI) focuses on the evaluation of brownian motion of water motion in biologic tissues, which has been related to tissue cellularity and architecture (see Table 1 ; Table 2 ). Thoracic application is technically demanding, requiring the application of echo planar imaging (EPI) readout and parallel acquisition strategy. The main hurdle is the presence of motion artifacts and geometric distortion secondary to B0 inhomogeneities and accumulation of phase error during echo train length.



Table 2

Combination of finding in diffusion-weighted imaging and related differential diagnosis




























Signal Intensity High b Value Signal Intensity ADC Map Interpretation
Hypercellular tumors. Rarely, liquid or viscous abscess or blood products
T2 shine-through; liquefactive necrosis
Fluid, necrosis, low cellularity lesions, well-differentiated adenocarcinomas
Fibromuscular tissues, fat, susceptibility artifacts (T2 dark-through effect)
Mature fibrosis with lower water content

White, black and gray cells correspond to hyperintense, hypointense and isointense – mild hypointense behaviour for both high b value and ADC map.

Adapted from Padhani AR, Koh DM, Collins DJ. Whole-body diffusion-weighted MR imaging in cancer: current status and research directions. Radiology 2011;261(3):708.


Motion compensation techniques are often necessary to improve image quality. Respiratory triggering is usually preferred to a single breath hold. For avoiding pulsation artifacts, cardiac triggering (which increases acquisition time) may be useful, especially for the assessment of small lesions located near the heart ( Table 3 ).



Table 3

Diffusion-weighted imaging sequences performed at our centers at 1.5T and 3T MR imaging



























































Techniques Sequence Type/Parallel Accelerating Factor B Values (s/mm 2 ) TR/TE (ms) Resolution (mm 3 ) Synchronization Fat Suppression Technique Image Evaluation Quantitative Parameter Acquisition Time
DWI – 3T (IVIM + DKI) SS EPI/factor 2 0, 50, 100, 500, 1000, 1500 5000/55 2.5 × 2.5 × 7 Respiratory triggered Spectral fat suppression Monoexponential, IVIM, and DKI model ADC
D, D*, f
D app ; K app
4 min 41 s
DWI – 1.5T (IVIM + DKI) SS EPI/factor 2 0, 50, 100, 500, 1000, 1500 1400/100 3 × 3 × 7 Respiratory triggered Spectral fat suppression Monoexponential, IVIM and DKI model ADC
D, D*, f
D app ; K app
5 min 48 s
DWI cardiac – 3T SS EPI/factor 2 0, 50, 150, 300 1000/44 2.58 × 2.54 × 10 Breath hold; ECG based in diastole Spectral fat suppression Qualitative/semiquantitative ADC
ADC ratio
SI ratio
1 min 45 s
DWI cardiac – 1.5T SS EPI/factor 2 0, 50, 300 2250/96 2.6 × 2.54 × 8 Breath hold; ECG based in diastole Spectral fat suppression Qualitative/semiquantitative ADC
ADC ratio
SI ratio
2 min 25 s

Abbreviations: D*, perfusion contribution to diffusion signal decay; D app , estimation of diffusion coefficient using nongaussian model; K app , apparent diffusional kurtosis; SS EPI, single-shot echo planar imaging; TE, echo time; TR, repetition time.


Apparent diffusion coefficient (ADC) represents the exponential signal decay of water molecules on a voxel by voxel basis. Although ADC may be calculated using only 2 b values, the more b values included in its measurement, the more accurate it is.


Recently, advanced models of DWI have been proposed to better understand the behavior of water motion in the different tissues. The intravoxel incoherent motion (IVIM) model of diffusion signal decay has been shown to better fit than monoexponential analysis, especially in the evaluation of well vascularized organs. First, diffusion decay signal shows a rapid attenuation at low b values (b<100–150 s/mm 2 ) secondary to bulk motion of water molecules in capillaries (perfusion effects on diffusion). With higher b values (more than 100 s/mm 2 ), a slower decay of signal occurs due to the real diffusion of water molecules ( Fig. 1 ). IVIM-derived parameters are, theoretically, more reliable markers of tissue diffusivity than ADC and can separate both compartments of diffusion signal decay ( Box 1 ). In addition, ADC assumes a gaussian diffusion behavior, which does not always exactly fit to the real signal decay of diffusion signal. Diffusional kurtosis imaging (DKI) quantifies the deviation of tissue diffusion from a gaussian pattern by measuring diffusion with ultrahigh b values greater than 1500 s/mm 2 (see Fig. 1 ). IVIM and DKI models have been recently explored in the evaluation of chest malignancies and show some promise over conventional ADC measurements.




Fig. 1


Diffusion signal decay analyzed using a bicompartmental model and nongaussian diffusion fitting of DWI signal decay on a patient with lung cancer. Graphic represents the differences between the diffusion signal decay using a monoexponential model (ADC measurement) and the better fit to the real behavior of diffusion signal decay using IVIM and DKI models. ( Right ) Color parametric maps are derived from IVIM (f and D) and DKI (K app and D app ), which can be calculated with a sequence, including only 6 b values ( top ).


Box 1





  • IVIM-derived parameters



  • True diffusion of tissue H 2 O molecules (D): not influenced by movement of water molecules within the capillaries



  • Perfusion contribution to diffusion signal (f): fractional volume of flowing water molecules within the capillaries.



  • Perfusion contribution to signal decay (D*): amount of nondiffusional random movements of water molecules




  • DKI-derived parameters



  • D app estimation of the diffusion coefficient in the direction parallel to the orientation of diffusion sensitizing gradients.



  • Apparent diffusional kurtosis (K app ): measures the deviation of the true diffusion from a gaussian pattern



Bicompartmental and nongaussian diffusion-weighted imaging–derived parameters


Dynamic Contrast-Enhanced–MR Imaging


Dynamic contrast-enhanced (DCE)–MR imaging provides information of tumor physiology, including blood flow, vascular volume, and permeability. DCE–MR imaging also provides imaging insight pertinent to tumor angiogenesis.


Free breathing, high temporal resolution, 3-D gradient-echo sequences are usually acquired during a 5-minute period of time. They have limited coverage and require the use of parallel imaging techniques. Motion and respiratory artifacts are common problems that require correction software during postprocessing.


Different strategies for analysis and quantification of DCE–MR imaging have been reported. The easiest and most used approach is to evaluate the variation of contrast enhancement during multiple time points of the lesion, obtaining time intensity curves (TICs). In the graphic representation of contrast dynamics, the first half is correlated with tumor angiogenesis and the second half with tumor interstitium. This approach does not separate perfusion and permeability. There are several TIC-derived descriptors that can be used for assessing perfusion ( Box 2 ).



Box 2





  • TIC-derived parameters



  • Contrast arrival time (s): time when contrast arrives to a pixel



  • Time to peak (s): time to maximum SI



  • Maximum Relative Signal Enhancement (RSE): maximum SI value related to initial non–contrast-enhanced intensity



  • Wash-in: represents the velocity of enhancement



  • Washout (WO): represents the velocity of enhancement loss



  • Initial area under the curve (IAUC): integral of the contrast concentration time curve for a specific time interval. It accounts for blood flow and motion, permeability, extracellular space, and microvessel density.




  • Quantitative derived parameters from a pharmacokinetic (compartmental) model



  • K trans (min −1 ): transfer rate of blood from plasma space to extracellular space (transfer constant)



  • K ep (min −1 ): transfer rate of blood from extracellular space to plasma space (elimination constant)



  • V e : volume of extracellular space inside a voxel



  • V p : volume of plasma space inside a voxel



Dynamic contrast-enhanced–MR imaging derived parameters


Once the contrast concentration is derived from each dynamic time point acquisition, the tissue status can be quantitatively assessed by calculating the contrast concentrations in the tissue and feeding artery. After applying a contrast kinetic model, quantitative hemodynamic parameters may be obtained. These are based on the use of compartments, defined as spaces where the contrast is evenly distributed. Two well-differentiated models are commonly used, monocompartmental and bicompartmental , according to if only the vascular volume (P) or also the extracellular compartment (E) is considered (see Box 2 ).


The bicompartimental model assumes that: P and E are compartments; E does not exchange tracer with the environment; the clearances for the outlets connecting P and E are equal and the clearance for the outlet of P to the environment equals the plasma flow. In some situations, a tissue with this configuration is reduced to a monocompartmental model:



  • 1.

    When one of the spaces has a negligible volume


  • 2.

    When the tracer extravasates with a slow rate (very low concentration at E)


  • 3.

    When the tracer extravasates with a high rate (single well-mixed space)



Compartmental models are more technically demanding and more time consuming but have demonstrated usefulness in monitoring treatment response and recurrence, because they provide reproducible measurements. Depending on which parametric approximation is used, derived parameters differ in their interpretation (see Box 2 ). Depending on whether or not the delivery of the contrast medium to the tissue is sufficient, K trans approximates the permeability surface area product or blood flow, respectively. Lack of standardization of different models and sequence designs have prevented them from being introduced in the clinical setting.




Functional MR imaging of the lung


Worldwide, lung cancer is the leading cause of cancer-related deaths. CT usually constitutes the first modality in evaluation and staging. It is often complemented with 18 FDG-PET/CT. Other functional modalities have emerged to increase the accuracy of detection, staging, and treatment monitoring. Morphologic MR was classically set as a second-line diagnostic procedure in the evaluation of patients with lung cancer. Recent improvements in hardware and functional techniques, however, have positioned chest MR as an alternative to 18 FDG-PET/CT.


Pulmonary Nodule Detection and Characterization


Functional MR imaging and pulmonary nodule detection


One of the most common challenges in pulmonary medicine is the detection and characterization of pulmonary nodules (PNs). CT is considered the best imaging technique for PN detection.


DWI has been shown useful in the management of PNs. DWI may detect 86.5% of nodules between 6 mm and 9 mm and 97% of PNs greater than 10 mm. In lesions less than 5 mm, the accuracy of DWI falls to only 43.8%, below the sensitivities of other MR imaging sequences, such as short-tau inversion recovery (STIR) (81.5% for lesions >3 mm).


Diffusion-weighted imaging and pulmonary nodule characterization


DWI has shown great usefulness in differentiating benign from malignant PNs ( Table 4 ). With an accuracy cited at 91%, it is comparable to other functional techniques for this differentiation (93%, 94%, and 94% for DCE-CT, 18 FDG-PET/CT, and DCE–MR imaging, respectively). DWI has similar limitations compared with PET in lesion characterization but with lower false-positive results ( Fig. 2 ; Table 5 ).



Table 4

Apparent diffusion coefficient threshold value parameters and reported results













































Region/Lesion Evaluation MR Field Strength B Values Threshold Sensitivity/Specificity/Accuracy
PN characterization (ADC) 1.5T and 3T 0, 500, b1000 s/mm 2 1.1–1.4 × 10 −3 mm 2 /s Pooled 83%/81%/91% Sensitivity (70%–90%)/specificity (74%–100%)
LN characterization (ADC) 1.5T 0, 300, 600 s/mm 2 1.85 × 10 −3 mm 2 /s 96.4%/71.4%/83.9% Pooled sensitivity, specificity, and accuracy: 77.4%–80%, 89%–95%, and 84.4%–97%, respectively
Described PPV and NPV: 95.2% and 77.1%, respectively
Pleural lesion (ADC) 3T 0, 50, 100, 500, 750, 1000 s/mm 2 1.52 × 10 −3 mm 2 /s 71.4%/100%/87.1% Differentiation of benign versus malignant
Mediastinal mass (ADC) 1.5T 0, 300, 600 s/mm 2 1.56 × 10 −3 mm 2 /s 95%/96%/94% Exclusion of cystic mediastinal masses
PPV and NPV: 94% and 96%, respectively
Cystic mediastinal mass (ADC) 3T 0, 100, 900 s/mm 2 2.5 × 10 −3 mm 2 /s 100%/100%/84% With this cut-off, the ADCs of all cystic benign lesions were higher and the ADCs of malignant ones lower.

Abbreviations: NPV, negative predictive value; PPV, positive predictive value.



Fig. 2


( A ) Multidetector CT with focused zoomed image ( inset ) reveals a partially solid millimetric nodule in keeping with minimal invasive adenocarcinoma. ( B ) Proton-density turbo spin-echo (TSE) axial image and ( C ) low b value DWI (b = 0 s/mm 2 ) confirm the solid part of the nodule ( white arrows ). ( D ) DWI confirms absence of high SI on high b value image (b = 1000 s/mm 2 ). ( E ) DCE–MR imaging maximum RSE and ( F ) area under the curve parametric maps show a low enhancement lesion ( gray arrows ). In this case, functional MR imaging rules out biological aggressiveness of this nodule.


Table 5

False-positive results and false-negative results of diffusion-weighted imaging and fludeoxyglucose F 18–PET/CT in chest lesions


































Diagnostic Modality False-Positive Results False-Negative Results
18 FDG-PET/CT PNs Inflammatory nodules Well-differentiated adenocarcinoma
Small PNs and metastasis
LNs Inflammatory LNs Small intranodal tumor deposits
Pleural lesions Inflammatory pleuritis
Talc pleurodesis
Tuberculous plaques
Parapneumonic effusion
Well-differentiated tumors
DWI PNs Intranodular inflammation Mucinous content (T2 shine through)
Intranodular necrosis
Nonsolid; part-solid lesions (high b value)
Low-grade adenocarcinoma
Small PNs and metastasis
LNs Tuberculous and nontuberculous granulomatous inflammation Small intranodal tumor deposits
Pleural lesions Fibrous plaques (T2 dark-through) Intratumor necrosis and inflammation


Varying results among quantitative characterization of PNs with DWI between series may be explained by the lack of standardization of acquisition and postprocess evaluation ( Figs. 3 and 4 ). Koyama and colleagues reported significant differences in minimum and mean ADC in the evaluation of lung tumors, reflecting the inherent heterogeneity of these lesions. Usuda and colleagues applied a DWI signal intensity (SI)-based analysis revealing that DWI SI was an indicator of the amount of tumor cells and its distribution. Tumor SI of the lesion-to–spinal cord ratio (LSR) of malignant lesions was found higher compared with that of benign ones. Contrarily, Liu and colleagues did not obtain any significant differences of the LSR between malignant and benign nodules. Koyama and coworkers, however, found significantly higher specificity and accuracy of LSR compared with ADC and IVIM-derived parameters in the differentiation of benign and malignant pulmonary nodules. They did not find significant differences in ADC, diffusion coefficient (D), and perfusion fraction (f) between malignant and benign PNs (see Figs. 3 and 4 ).




Fig. 3


Indeterminate PN characterization (17 × 19 mm) in a 70-year-old man with functional MR imaging. ( A C ) DWI with b values of 0 and 900 s/mm 2 reveal a nonrestrictive lesion, as confirmed in ADC map (ADC mean = 2.23 × 10 −3 mm 2 /s; white arrows ) ( C ). ( D ) IVIM-derived diffusion coefficient map confirming the absence of restriction (1.92 × 10 −3 mm 2 /s; white arrow ). ( E ) DCE–MR imaging maximum RSE map shows a low enhancing lesion ( white arrow ). ( F ) Postgadolinium 3-D–GRE sequence shows rim enhancement of this lesion corresponding to an old healed tuberculoma ( white arrow ).

Sep 18, 2017 | Posted by in MAGNETIC RESONANCE IMAGING | Comments Off on Functional MR Imaging in Chest Malignancies

Full access? Get Clinical Tree

Get Clinical Tree app for offline access