Pseudophenomena, that is, imaging alterations due to therapy rather than tumor evolution, have an important impact on the management of glioma patients and the results of clinical trials. RANO (response assessment in neurooncology) criteria, including conventional MRI (cMRI), addressed the issues of pseudoprogression after radiotherapy and concomitant chemotherapy and pseudoresponse during antiangiogenic therapy of glioblastomas (GBM) and other gliomas. The development of cancer immunotherapy forced the identification of further relevant response criteria, summarized by the iRANO working group in 2015. In spite of this, the unequivocal definition of glioma progression by cMRI remains difficult particularly in the setting of immunotherapy approaches provided by checkpoint inhibitors and dendritic cells. Advanced MRI (aMRI) may in principle address this unmet clinical need. Here, we discuss the potential contribution of different aMRI techniques and their indications and pitfalls in relation to biological and imaging features of glioma and immune system interactions.
Glioblastoma multiforme (GBM) is the most common primary brain tumor in adults [
Infiltrative nature of diffuse gliomas makes it difficult to eliminate microscopic disease despite macroscopic gross total resection. Recurrence of GBM is inevitable, and the median overall survival (OS) time of GBM patients receiving the standard treatment, which consists of maximal safe resection followed by radiation and adjuvant temozolomide, is about 14–16 months [
The central nervous system (CNS) has been traditionally considered an immune-privileged system; however, it has been proved that immune cells can cross the blood-brain barrier (BBB) to gain access to the brain parenchyma and can leave the CNS to reach the cervical lymph nodes. Considering that the immune system has access to the brain and that GBM expresses multiple tumor antigens that can be targeted by immunotherapeutic approaches, the development of immunotherapy has gained considerable interest over the last decade [
Converging data indicate that cancer epitopes can be recognized by the immune system and therefore an immune reaction can be mounted to erase or block tumor growth. Resistant tumor clones, grown under immune pressure, create an immune suppressive environment that leads to the formation of relevant tumor. These general observations also apply to brain tumors. Cancer immunotherapy strategies are aimed at reverting such immune suppression [
Novel immunotherapeutic strategies being investigated to treat glioblastoma can be broadly divided into three major classes: active immunotherapy, adoptive immunotherapy, and immunomodulatory strategies [
Initial data show prolonged OS (23 to 38 months) in GBM patients treated by vaccines [
Efficacy of therapy is assessed by clinical examination and magnetic resonance imaging (MRI). Pseudoprogression, that is, imaging features suggesting tumor progression that is not confirmed subsequently, occurs in up to 30% of patients within three months after radiochemotherapy [
To address these issues, the iRANO (immunotherapy response assessment in neurooncology) committee redefined the response assessment criteria for patients with neuro-oncological malignancies undergoing immunotherapy: the “limbo” window when radiologic worsening does not suggest immunotherapy suspension has been widened to six months, after which true progression, if detected, should be backdated [
Conventional MRI (cMRI) has limitations in differentiating tumor progression/recurrence and immunotherapy responses [
Amino acid PET (mainly with methionine and fluoroethyltyrosine) has been used to enlighten the greater metabolic activity of malignant tumoral tissue compared to radionecrosis and might also help in differentiating progression from treatment-related alterations during immunotherapy [
Evidence that aMRI techniques can differentiate pseudoprogression and tumor recurrence has been reported in radiotherapy and chemotherapy, and promising data suggest they may differentiate at early-stage responder and nonresponder patients to immunotherapy. The purpose of this review is to summarize current research on MRI assessment for patients undergoing immunotherapy with a major focus on aMRI parameters.
Several criteria have been proposed in the last two decades to assess response to therapy in gliomas: the standard method is based on contrast-enhancing images in T1 and on hyperintensity in T2 or FLAIR (fluid-attenuated inversion recovery) sequences. Nevertheless, enhancement on T1 reflects nonspecific impairment of the BBB, a reduction or lack of enhancement can be due to tumor shrinkage but also to antiangiogenic therapy, due to vascular normalization besides tumor infiltration (pseudoresponse). On the other hand, in pseudoprogression, an early, subacute reaction to treatment (e.g., radiotherapy) is associated with contrast enhancement, edema, and possible mass effect, and sometimes, associated clinical symptoms initially suggest tumor progression but subsequently resolve without any further treatment and can be associated to longer survival [
Pseudoprogression is generally not associated with clinical deterioration in radiochemotherapy [
Volume of enhancement lacks to differentiate between progressive disease and pseudoprogression. Moreover, the pattern of enhancement in pseudoprogression is not specific and can be nonhomogeneous, mimicking GBM, nodular, “cottony,” and sometimes quite intense as in “flare” inflammatory phenomena also observed after local intracerebral/intratumoral immunotherapies [
RANO (response assessment in neurooncology) criteria, including cMRI were published in 2010 to address the issues of pseudoprogression after radiochemotherapy and/or pseudoresponse during antiangiogenic therapy [
Pseudoprogression can be more frequent after immunotherapy. The precise mechanism of pseudoprogression, occurring in up to 30% of patients with glioblastoma after radiochemotherapy, is poorly understood [
In 2009, the increased interest in evaluating immunotherapies led to the development of immune-related response criteria (irRC) [
The iRANO committee, integrating guidance for progressive imaging findings from the irRC with RANO criteria, redefined the response assessment criteria for patients with brain tumors undergoing immunotherapy providing novel iRANO criteria [
iRANO criteria (modified from [
RANO criteria for high-grade gliomas | ||
---|---|---|
Complete response (CR) | (i) Disappearance of all enhancing | |
|
||
Partial response (PR) | (i) ≥50% ↓ sum of biperpendicular | |
|
||
Stable disease (SD) | (i) Does not qualify for CR, PR, | |
|
||
Progressive disease (PD) | (i) ≥25% ↑ sum of biperpendicular | |
|
||
iRANO criteria | ||
|
||
If ≤6 months |
If >6 months | |
|
||
Is a repeat scan required |
Yes | No |
|
||
Minimal time interval for |
≥3 months | Not applicable |
|
||
Is further immunotherapy |
Yes | Not applicable |
|
||
Does a new lesion |
No | Yes |
FLAIR (a–e) and contrast-enhanced T1-weighted images (f–j): postsurgical (a, f), increasing edema (b, c), enhancement (g, h) and subsequent reduction (d, e, i) of both, and remission of the enhancing lesion (j) in the course of immunotherapy with dendritic cell vaccine.
Jun 2014
Aug 2014
Oct 2014
Jan 2015
Mar 2015
Jun 2014
Aug 2014
Oct 2014
Jan 2015
Mar 2015
In recent years, Zach et al. proposed a new method to distinguish active tumor and treatment-induced effects [
Different from other methodologies, TRAMs are not user-dependent, less acquisition-dependent (i.e., they only need good-quality 3D T1 sequences), and relatively simple to be acquired.
The inconveniences are that patient has to wait longer outside the scanner and that timing of postcontrast acquisitions is quite critical. The choice of the first time point is important because right after contrast injection, the gadolinium signal rises fast and the signal has to be high when the images are acquired in order to be sensitive to tumor regions (blue). On the other hand, this acquisition time point has to be early enough not to lose sensitivity to treatment effects (red). The closer to the maximal peak value, the larger is the difference between early and delayed signal and, consequently, sensitivity. For these reasons, 3–5 minutes should be used and, due to the fast changes in signal intensity, it is important to fix this time point for each patient follow-up. The choice of the second time point is mainly affected by the time the tumor takes to clear gadolinium from the tissue. Inter- and intratumor variability in clearance times exist, but after 1 hour, the signal changes slowly; therefore, the second time point can be flexible (to allow for a practical clinical application in a busy radiology department) between 1 and 1.45 hours postinjection.
TRAMs have been used in radiochemotherapy and antiangiogenic therapy allowing discrimination between tumor and treatment-related effects (Figure
Contrast-enhanced T1-weighted images (a–f) and the calculated TRAM postchemoradiation (g–l) (images were acquired 0.7, 2.5, 4, 6, 7, and 8 months postchemoradiation). Temporary enlargement of enhancing lesion (c-d) is shown; as it can be seen, the red volume growth rate was prevalent above the blue volume (i-j), favoring pseudoprogression over progression. Pseudoprogression was later confirmed by the decrease in all volumes 7 and 8 months postchemoradiation (e-f, k-l). Modified from [
The rationale for applying TRAM analysis to immunotherapy lies on the differentiation between tumor and immune cells: preliminary data showed different components in enhancing lesions during immunotherapy with dendritic cells, with prevalence of “blue” regions in early progressive cases. However, longer follow-up in responder versus nonresponder patients is needed to understand if TRAMs can define immune-mediated pseudoprogression as they do in postradiotherapy follow-up.
Three different MRI techniques have been developed to study brain microvascular hemodynamics. Two are based on the injection of gadolinium-based contrast medium, dynamic susceptibility contrast (DSC) and dynamic contrast enhanced (DCE), and the other, arterial spin labeling (ASL), uses blood as an internal contrast medium.
DSC [
After the acquisition, in the postprocessing step, a signal-time curve is extracted from every voxel of the brain volume. The curve is composed of a first baseline, prior to contrast arrival, a sharp peak corresponding to contrast bolus arrival and a final portion that represents contrast recirculation. Starting from the signal, the concentration of the contrast medium during time is mathematically obtained in every voxel. From every concentration-time curve obtained, four semiquantitative parameters can be derived: (a) CBV, defined as the ratio between the blood volume passed in a region and that entering that region [
CBV has found applications in the study of brain gliomas for tumor grading [
BBB disruption is a frequent condition in brain gliomas. In this condition, the contrast molecules can leak from vascular space and reach the parenchyma. This results in an unwanted T1-weighted effect leading to under- or overestimation of CBV that can be overcome by the injection of a prebolus of contrast to saturate T1.
The principal limitations of DSC are related to the sensitivity of the sequence to BBB disruption or susceptibility artifacts at natural interfaces (e.g., trabecular bone, paranasal sinuses, skull base, and sella), generally heavier at high field strengths.
This technique uses gadolinium to characterize the BBB and estimate its damage [
The first step is the conversion of enhancing signal-time curves into concentration-time curves. From the concentration-time curves, a first not specific index can be derived: the initial area under the curve (iAUC). Comparable to CBV, this index could give indication about contrast leakage. Higher iAUC values generally correspond to more malignant conditions, where vessel permeability is high. This index however includes multiple information such as flow and permeability.
In order to have more detailed and specific information about BBB damage, the tissue in each voxel is represented by a multicompartment model composed of vascular, extravascular, and intracellular space. Using pharmacokinetic models [
With DCE, it is possible to extract quantitative values on BBB damage and intra- and extravascular volumes. Their estimation however is limited by water exchange phenomena, by the choice of the AIF, and by the robustness of the fitting procedure. Moreover, DCE uses a T1-weighted sequence that, different from DSC, is not affected by susceptibility artifacts. A 3T scanner or higher is preferable.
This technique does not use any external contrast medium: the contrast is the blood entering the brain that, magnetically labeled, can be used to estimate the CBF [
The main limitations are related to the intrinsically poor signal-to-noise ratio, due to the low fraction of free-water spins and to their relaxation. For this reason, the sequence generally used is echo-planar, allowing the acquisition and the averaging of multiple volumes.
DSC has been used to evaluate early responses [
DCE has been used for differential diagnosis [
ASL has recently found application in the study of brain tumors [
DSC perfusion has been studied during immunotherapy confirming that elevation of CBV in a region with contrast enhancement supports the diagnosis of malignant tumor [
In eight patients treated with dendritic cell immunotherapy [
Interestingly, a mismatch between enhancing volumes and high CBV volumes has been described in 11/79 examinations in three of six immune-treated patients: the region with elevated CBV was never larger than the region with contrast enhancement. Histopathological evaluation in two cases showed malignant cells with numerous proliferating vessels with thrombosis or ruptures [
DCE at 7T field strength has been used in immunotherapy studies in rat models of GBM: increased Ve was found in tumors responding to treatment due to tumor cell death and reduced proliferation, as indicated by the decreased growth index on histology. On the contrary, progressive lesions exhibited the greatest growth index and Ve was decreased with a tendency to reduce transvascular transport (
As vessel permeability can be affected by inflammation, endothelial junctions become less tight,
ASL has not yet been used in immunotherapy follow-up and would be an option in patients with renal failure and severe allergy to contrast agents, also limiting the potential risk of chronic contrast accumulation [
Brownian motion of water molecules inside the tissues brings water to diffuse in different brain compartments. This motion can be detected using a modified version of a spin-echo sequence that includes two strong gradients (diffusion gradients) [
Different from DWI that only gives information about the amount of water displacement, diffusion tensor imaging allows to infer water-motion directionality [
The role of DTI metrics in tumoral characterization is debated. Some studies found that the tumoral core is characterized by low MD and high FA values [
Starting from the DTI metrics and from the directional information extracted in each voxel, it is possible to reconstruct three-dimensionally the pathways of white matter fibers. This technique, called tractography, is widely used in association with functional MRI before tumor resection [
DWI has been widely performed to characterize brain gliomas and monitor radiochemotherapy or antiangiogenic therapy [
Progressive disease in high-grade gliomas, different from pseudoprogression, seems characterized by low ADC values due to hypercellularity [
DTI has increasingly been performed to study high-grade gliomas; histogram analysis and fDMs can provide early evidence of low-grade glioma modifications during chemotherapy with respect to cMRI (i.e., RANO criteria [
During immunotherapy, an inflammatory reaction is expected, carrying edema and reduced tumor cell density on one side but immune cell accumulation and hypercellularity on the other side as in other brain inflammatory diseases such as encephalitis or abscesses that exhibit lower ADC values than normal brain [
In a pilot study on eight patients treated with dendritic cell immunotherapy, minimum ADC levels were lower in enhancing lesions at progression [
Serial parametric response mapping of ADC in 21 children carrying pontine glioma treated by peptide-based vaccination following radiation therapy showed fractional decreased ADC in the four patients experiencing pseudoprogression [
Very recently, in a retrospective analysis of 10 recurrent GBM patients [
DTI has been proposed in the follow-up of gliomas with treatments other than immunotherapy [
MRS aims to study brain metabolism identifying and quantifying some relevant metabolites in a specific region. The water proton is most commonly used, as it is easy to implement in most of the scanners. Unlike MRI, which uses the two-dimensional signals to derive images of the brain, MRS uses a monodimensional 1H signal to estimate relative metabolite concentrations. Two are the principal MRS modalities: (a) chemical shift imaging (CSI) that gives a spatial distribution of the metabolites taking at the same time spectra deriving from multiple brain voxels (a grid) and (b) single voxel spectroscopy (SVS) that only acquires spectra from a little portion (VOI) of the brain. Both SVS and CSI do not cover the entire brain volume.
The most common metabolites investigated by MRS are (a) N-acetyl aspartate (NAA), a neuronal marker decreasing when neuronal integrity is affected; (b) choline (Cho), a marker of increased cellular turnover usually elevated in tumors and inflammatory processes; and (c) creatine (Cr), which gives a measure of energy storage. In brain tumors, NAA results generally decreased and Cho increased, probably due to the membrane turnover. Other metabolites whose concentration generally changes in brain tumors are lactate, due to the anaerobic glycolysis; lipids, probably because of membrane disruption and necrosis; and myoinositol, associated to “crowding” of glial cells. Most recently, tumor characterization and therapeutic monitoring benefited from the possibility to study 2-hydroxyglutarate (2HG), an oncometabolite accumulating in tumors carrying isocitrate dehydrogenase (IDH) gene mutations [
Absolute quantitative MRS gives the concentration of the metabolites in a given voxel. Ratios of metabolite concentrations and metabolic maps (i.e., colorimetric maps reporting the single metabolite or ratio values in every voxel of the CSI grid), can be obtained.
MRS has been used for glioma diagnosis, grading, and response monitoring [
GBM is usually characterized by high concentration of Cho, decreased Cr and NAA, and presence of lipids in necrotic areas. Because of the inter- and intraindividual heterogeneity of high-grade gliomas, metabolite concentrations can vary considerably. MRS can detect the presence of high Cho levels (and Cho/Cr or Cho/NAA ratios) within enhancing and perifocal tissue thus enlightening the presence of glioma: specificity is high but mixed scenarios with coexistence of glioma and treatment alterations are frequent (Figure
MRS during immunotherapy, after surgery, and radiotherapy plus temozolomide. Left, spectra; right, voxel positioned within enhancing lesions. (a) High Cho and low NAA with minimum lipids in recurrent glioma. (b) Preserved Cho and NAA levels with evident though not prevalent lipid peak in pseudoprogression (the same case is shown in Figure
MRS findings were reported in two GBM patients after multimodal treatment with surgery, radiation, intralesional immunotherapy (IL-4 toxin), and chemotherapy: pseudoprogression was observed with extensive and increasing enhancement which nearly completely regressed after four to six months. In both patients, MRS did not show increased Cho within the enhancing areas [
A “harmonic” reduction of Cho, Cr, and NAA in the presence of lipids is usually associated with radionecrosis while lipids in the presence of elevated Cho/Cr ratio and low NAA suggest the presence of high-grade glioma. In immunotherapy, lipids have been described as substrate of NK T-cells [
Susceptibility-weighted imaging (SWI) is a tool for high-resolution imaging of the vasculature. The technique relies on the phase signal of a T2
SWI has a role in the evaluation of the vascular organization of brain gliomas and of neoangiogenesis that rapidly produces small, tortuous, and immature vessels leading to microbleedings and surrounding edema and also in the identification of tumor calcifications.
SWI has been used for tumor grading [
SWI data in immunotherapy have not yet been reported.
This technique, with contrast medium, might have a potential in immunotherapy response assessment as SWI features are a surrogate of vascularity and are more pronounced on lesional borders, where enhancement is frequently prominent in immune-treated tumors or inflammatory diseases. SWI might be considered to differentiate enhancing GBM from areas rich in immune cells, given that ITSS can be found in high-grade gliomas and the development of new ITTS suggests recurrent or progressive disease, features that are absent in lymphomas (i.e., hypercellular and lymphocytic tumors) [
In spite of the improvement determined by RANO criteria first and iRANO subsequently, the imaging definition of the actual dynamics of glioma and immune cell interactions and their impact on patient survival during checkpoint or DC immunotherapy remains unsatisfactory.
Mixed scenarios with coexistence of glioma and treatment alterations are often the rule; moreover, with regard to the diagnostic specificity of contrast enhancement and of aMRI features, the situation after multimodal treatment could become confusing. It seems plausible that aMRI may provide deeper insights than cMRI in the recognition of pseudo or real glioma progression. On PWI, contrast-enhancing areas secondary to immunotherapy inflammation should be less perfused than progressive/recurrent tumor, but also, vessel disruption and thrombosis due to high malignancy may inversely affect the perfusional pattern. Nevertheless, inflammation increases vessel permeability with effects on perfusional parameters. Likewise, low ADC can be associated to both tumoral and immune hypercellularity, and specific analysis has to be performed to discriminate. MRS is useful to obtain metabolic information within the enhanced areas, by determining high choline concentration and therefore identifying glioma within treatment alterations (Figure
Enhancing lesion (a) during immunotherapy with dendritic cell vaccine. Mismatch between T1-enhancing volume and CBV (b), the last being just slightly elevated; permeability (
In clinical practice, a combination of different techniques may be necessary to differentiate between pseudoprogression and tumor progression. Cut-offs in a single-shot examination cannot distinguish between progression and pseudoprogression, and the evaluation of longitudinal modifications of parameters in terms of intensity and pattern is recommended. MRI data need to be analyzed taking into account that gliomas are generally composed of different structural and functional regions and that multimodal treatments increase brain tissue heterogeneity. Two are the main approaches used: (a) the histogram approach [
Imaging approaches like these, evolving in-depth analyses of MRI data that take into account whole-lesion heterogeneity and parametric modifications in the course of treatment such as parametric maps, TRAMs, and histogram analyses, deserve further investigation as they may provide the robust tools that are presently missing for the definition of PFS (i.e., progression-free survival) and clinical benefit in glioma immunotherapy.
Axial diffusivity
Apparent diffusion coefficient
Advanced magnetic resonance imaging
Arterial spin labeling
Blood-brain barrier
Cerebral blood flow
Cerebral blood volume
Choline
Conventional magnetic resonance imaging
Central nervous system
Creatine
Dynamic contrast enhanced
Dynamic susceptibility contrast
Diffusion tensor imaging
Diffusion-weighted imaging
Fractional anisotropy
Functional diffusion maps
Glioblastoma multiforme
Initial area under the curve
Immunotherapy response assessment in neurooncology
Intratumoral susceptibility signals
Mean diffusivity
Magnetic resonance imaging
Magnetic resonance spectroscopy
N-acetyl aspartate
Overall survival
Progression-free survival
Parametric response maps
Response assessment in neurooncology
Radial diffusivity
Region of interest
Susceptibility-weighted imaging
Treatment response assessment maps
Extracellular extravascular volume fraction
Volume of interest.
Gaetano Finocchiaro received an honorarium from BMS for the participation to the SNO 2016 advisory board.
All authors have contributed to the work, concurred with the submission of the paper, and agreed with its content.