Multiparametric MRI: practical approach and pictorial review of a useful tool in the evaluation of brain tumours and tumour-like lesions

MRI plays a major role in the diagnosis, grading, treatment and treatment response assessment of brain tumours and other intracranial lesions. Conventional MRI provides the anatomical and structural details of lesions in the neuraxis; however, its specificity is limited. Even with recent improvements in contrast resolution, higher magnetic field strengths and improved contrast agents, tissue characterisation remains limited using conventional imaging acquisitions. As a result of diagnostic uncertainties, patients will undergo invasive biopsy of brain lesions, which is not without risk [1]. Several adjunct MR imaging techniques have been developed to quantitatively measure a number of biophysical properties of brain tissue in vivo, allowing regional changes in the tissue microstructural environment to be better characterised. These techniques include diffusion-weighted imaging (DWI), perfusion-weighted imaging (PWI) and magnetic resonance spectroscopy (MRS). DWI provides information about cellularity and water movement, PWI provides information about angiogenesis and vascularity and MRS provides information about the composition of various metabolites within the tissue. These quantitative methods provide information about tumour cellularity, proliferation, vascularity, vessel permeability and cell membrane turnover. Changes in physiological processes due to the nature of the underlying lesion are reflected in the information obtained. There have been a number of studies demonstrating that these techniques in combination can help improve differentiation of neoplastic from non-neoplastic lesions (for example, tumefactive demyelination, tumefactive vasculitis and other inflammatory disorders) [2, 3], grading of brain tumours [4], differentiation of glioblastoma pseudoprogression from true progression [5] and response of brain metastases to stereotactic radiosurgery (SRS) treatment [6]. Over time, there has been development of these adjunct advanced MRI techniques in isolation, beginning with MRS, DWI and then PWI. In clinical practice and throughout the literature, usually these techniques were compared with each other; however, recent studies show that the information gained from each of these techniques are complementary. In this pictorial review, we illustrate the use of a multiparametric MRI approach consisting of DWI, PWI and MRS in clinical neuro-oncology practice to help with the diagnosis of intracranial lesions, treatment planning and assessing response to treatment.

Our multiparametric studies are performed on a 3 T scanner (Magnetom Verio; Siemens, Erlangen, Germany) with a 32-channel phased-array head coil, although such studies can also be performed on other similar scanners and coils. Acquisition parameters are summarised in Fig. 1. Axial T2-weighted (T2W) images, T2W FLAIR and DWI (b value 1000) of the whole brain are generally obtained first. This is followed by dynamic susceptibility contrast-enhanced (DSC) perfusion imaging using gradient-echo echo-planar imaging (GE-EPI) during the first pass of a standard dose (7.5 mmol) bolus of gadolinium-based contrast agent (Gadovist, Bayer Schering Pharma, Berlin, Germany) administered intravenously at a flow rate of 6 ml/s. A total of 80 imaging volumes are acquired at a temporal resolution of 2.1 s with the bolus typically arriving between the 10th and 15th volume. This is followed by post-contrast 3D T1-weighted (T1W) magnetisation-prepared rapid acquisition with gradient echo (MPRAGE) sequence acquired in the axial plane with sagittal and coronal reformats.

MRS is performed using a combination of multi-voxel (for tumoural and peri-tumoural regions) and single-voxel point resolved spectroscopy PRESS sequences with short echo (TE = 30 ms) and intermediate echo (TE = 135 ms). TE 135 ms is usually performed to show lactate inversion at 1.3 ppm (J-coupling effect). Typically, 2D or 3D MR spectroscopic imaging (MRSI) is first performed in the axial plane choosing a slice or slab with the largest contrast-enhancing lesion area (or FLAIR if non-enhancing), area with restricted diffusion, or high perfusion. This is followed by single-voxel MRS with placement of the volume-of-interest further guided by the metabolic profiles estimated by MRSI. The single voxel method is used to maximise diagnostic yield by combining information from contrast-enhancement, DWI, DSC and MRSI to sample the most relevant part of the lesion likely to provide the highest quality spectra.

Normal brain ADC values for cortical grey and white matter are 833 × 10⁻⁶ mm² s⁻¹ and 701 × 10⁻⁶ mm² s⁻¹ respectively [7]. Mean ADC values in high-grade neoplastic lesions such as glioblastoma, anaplastic astrocytoma, and metastases have shown to be 700–780 × 10⁻⁶ mm² s⁻¹, lymphoma has shown to be 510 × 10⁻⁶ mm² s⁻¹ and low-grade tumours have shown to be 1090 × 10⁻⁶ mm² s⁻¹ [8]. To calculate the rCBV ratio, the ROI is generally compared with the normal-appearing contralateral white matter. The mean rCBV ratios in high-grade neoplastic lesions have shown to be 1.9, compared to 1.3 in low-grade neoplastic lesions [9]. Normative values for Cho/Cr at TE 135 ms range from 0.7–1.0 in grey matter and 1.2–1.4 in white matter, with slightly higher values seen in the brainstem and cerebellum [10]. Short TE (30 ms) shows more metabolites and is primarily used for assessing tumoural and non-tumoural lesions. Normal Cho/Cr ratios using short TE MRS are 0.6 in grey matter and 1.0 in white matter [11]. High-grade neoplastic lesions have shown to demonstrate a mean Cho/Cr ratio of 2.4 on short TE MRS, compared with a mean Cho/Cr ratio of 1.5 for low-grade neoplastic lesions [12]. As there is a wide variability of cut-off values for each parameter in the literature, based on the results of a number of studies, we defined high-grade neoplastic lesions to have cut-off values of ADC < 1000 × 10⁻⁶ mm² s⁻¹, rCBV ratio > 2.0 and Cho/Cr ratio > 1.8 [12‐15]. We utilised these parameters semi-quantitatively by defining the lowest ADC, highest rCBV and highest choline values within the lesion. This multiparametric information was read in combination with conventional imaging, clinical findings and other investigations.

There are some inherent challenges for adoption of multiparametric techniques in routine clinical practice, such as brain regions affected by susceptibility, small lesions and non-enhancing lesions. However, the adoption and widespread clinical use of multiparametric MRI protocols is improving with the use of higher magnetic field strength magnets, specialised coils and readily available vendor post-processing tools. We have incorporated a multiparametric MRI protocol consisting of DWI, PWI and MRS into our routine clinical practice for neuroimaging and our single-centre experience shows that these techniques clearly make a positive difference for individual patient management. It helps make more informed decisions at the tumour board (multi-disciplinary team) meetings, removing some uncertainty and leading to patients starting appropriate treatment earlier, which improves the overall survival rate and outcome. It is imperative that multiparametric information is read in combination with structural MR sequences, such as T1W, T2W, FLAIR, SWI, GRE to further characterise lesions. These semi-quantitative multiparametric parameters (ADC, rCBV, Cho) should be evaluated comprehensively and in conjunction with each other, rather than in isolation to narrow the differential diagnosis. With advances in these techniques, neuroradiology is in a unique position to evaluate the whole tumour and peri-tumoural environment, which could be a big limitation for histopathology, as commonly noticed in biopsy sampling error [26]. It is not uncommon for histopathology results to be re-reviewed following incorporation of these adjunct techniques in clinical practice, leading to a change in patient management.

There has been improvement in the standardisation of acquisition techniques over time, particularly with the publication of white papers on imaging [51, 52]. However, the cross-site and cross-vendor standardisation is still difficult to address, as there is some variability of threshold values and limited understanding of combining the parametric information. However, this will further improve with routine incorporation of these techniques in clinical practice with larger datasets and multi-centre studies.

Through this educational pictorial review, we have presented a variety of cases to demonstrate that multiparametric MRI using DWI, PWI and MRS in conjunction with conventional MRI is helpful for differentiating neoplastic from non-neoplastic lesions in the brain. It also helps in the grading of tumours, selecting biopsy targets particularly in non-enhancing lesions and assessing treatment response. We have also presented a practical approach to perform multiparametric MRI protocol in routine clinical practice.