Imaging spectrum of immunomodulating, chemotherapeutic and radiation therapy-related intracranial effects

Abstract

Objective:

A wide range of treatment-related side effects result in specific neurologic symptoms and signs and neuroimaging features. Even to the most seasoned neuroradiologist, elucidating therapy-related side effects from other common mimics can be challenging. We provide a pictorial survey of some common and uncommon medication-induced and therapy-related neuroimaging manifestations, discuss pathophysiology and common pitfalls in imaging and diagnosis.

Methods:

A case-based review is utilized to depict scenarios on a routine basis in a general radiology or neuroradiology practice such as medication-induced posterior reversible encephalopathy syndrome to the more challenging cases of pseudoprogression and pseudoregression in temozolmide and bevacizumab therapy in gliobastoma treatment protocols.

Conclusion:

Knowledge of the treatment-induced imaging abnormalities is essential in the accurate interpretation and diagnosis from the most routine to most challenging of clinical situations. We provide a pictorial review for the radiologist to employ in order to be an invaluable provider to our clinical colleagues and patients.

Introduction

There are a wide range of treatment-related effects, which result in specific neurologic signs and symptoms and in some cases, specific neuroimaging features. Even to the most seasoned radiologist and neuroradiologist, distinguishing therapy-related side effects from other disease processes and common mimics can be challenging. In this review, we discuss the effects of immunomodulation, chemotherapeutic agents and radiation therapy (RT) on the brain.

PRES

Posterior reversible encephalopathy syndrome (PRES) was first described in 1996 by Hinchley et al, as a clinical-radiologic cerebral process stemming from autoregulatory dysfunction.¹ PRES affects both children and adults with a wide variety of presentations including seizures, cognitive and memory impairment, headache, and/or cortical visual disturbances.² Commonly seen in hypertensive patients, it is also seen in eclampsia/pre-eclampsia, sepsis, autoimmune disease and chemotherapy and calcineurin inhibitor medications.^3,4

For purposes of this review, medication-related PRES includes the use of chemotherapy agents and calcineurin inhibitors, which are utilized in solid organ as well as bone marrow transplant patients. Several commonly used chemotherapeutic agents have been implicated in development of PRES, which include methotrexate, cytarabine, cisplatin, gemcitabine, tiazofurin and bevacizumab. Chemotherapy-induced PRES has a much higher mortality rate and incidence of residual deficit in cancer patients compared with their non-cancer population counterpart with PRES.⁵ Therefore, early diagnosis is very important in these cases in order to initiate appropriate and timely management.

Methotrexate, a cell cycle specific folate analog, is commonly used in treatment of acute lymphoblastic leukaemia, lymphoma, osteosarcoma and medulloblastoma. There is a linear dose relationship of methotrexate administration, which results in acute to chronic neurotoxicity, which is worse with intrathecal administration. It is most commonly seen in paediatric population, but can manifest in the adult population.^4,6–8

Calcineurin inhibitors, such as cyclosporine and tacrolimus, are used to prevent rejection in solid organ and bone marrow transplantation and can also account for PRES.

There are several mechanisms by which these medications can result in PRES, but the general premise is through direct action on cerebral vasculature inducing vasospasm, which culminates in hypoperfusion, brain ischaemia, and vasogenic oedema from endothelial cell injury and swelling.^1,6 Regardless of the theorized cause, typical cases of PRES are reversible following prompt symptomatic treatment, blood pressure control or reduction or cessation of medication, although the appearance of lesions on imaging tends to lag behind clinical improvement.^4,9

Imaging findings related to PRES can be divided into typical or atypical manifestations as described by Bartynski and Boardman.¹⁰ More commonly, there are regions of hemispheric subcortical oedema in occipital and parietal lobes (Figure 1a) followed by frontal lobes, inferior temporal-occipital lobe junction and cerebellum.⁶ The less common, often difficult to diagnose, imaging spectrum includes patchy or focal areas of oedema isolated to the basal ganglia, thalami, corpus callosum, brainstem and periventricular white matter (Figure 1b) without the hemispheric cortical oedema.¹ Restricted diffusion is not a common association; however, when present, it suggests irreversible injury and can lead to incomplete clinical recovery. Although of unknown significance, there can be associated haemorrhage and variability in enhancement pattern. In a small percentage, between 15 and 17%,⁴ there is associated petechial or microhaemorrhage and subarachnoid or intraparenchymal haemorrhage. Although this number could potentially be more where susceptibility-weighted imaging is utilized, as it is more sensitive than GRE T₂* in evaluating for haemorrhage.^1,11 Other imaging findings seen in PRES include diffuse and segmental narrowing of medium to large arteries, mimicking vasculitis or reversible cerebral vasoconstriction syndrome, on computed tomography angiography, magnetic resonance angiography and conventional angiography.^2,6

Figure 1.

Axial FLAIR demonstrates multifocal areas of signal abnormality involving the parieto-occipital cortical (a) and periventricular white matter (b) with associated enhancement (not shown) and complete resolution of bilateral parieto-oociptal and periventricular white matter signal abnormality on repeat imaging 2 months later (not shown).

When typical and/or atypical imaging findings are present, a pertinent differential diagnosis while taking clinical picture into account needs to be considered, which include ischaemic stroke, venous thrombosis, progressive multifocal leukoencephalopathy (PML), vasculitis, encephalitis, reversible cerebral vasoconstriction syndrome and postictal reversible oedema. Among the differential diagnoses, bilateral posterior cerebral artery infarction is important and is readily distinguishable, as PRES typically spares the calcarine and paramedian cortex of the occipital lobe, unlike infarction.

IRIS

Immune reconstitution inflammatory syndrome (IRIS), first described in 1992, is an inflammatory condition that presents after the initiation of highly active antiretroviral therapy (HAART) in patients with HIV.¹² It begins from weeks to years after the start of treatment and represents a heightened reaction to an antigen from a prior or persistent opportunistic infection (OI) or, in rare cases, a self-antigen.^12,13 When occurring in response to an OI, IRIS is conventionally described as either paradoxical IRIS or unmasking IRIS. Paradoxical IRIS results from an inflammatory reaction in response to a treated OI from which an antigen remains.¹⁴ Unmasking IRIS occurs when a subclinical OI is suddenly combatted by a reconstituted immune system. IRIS can occur in any organ and as a result of any regimen of HAART, with an overall incidence of 25–35% in patients with HIV.¹²

Central nervous system IRIS (CNS-IRIS) is much rarer, with an estimated incidence of 0.9–1.5% and mortality ranging from 5 to 15%. Long-term outcome is generally favourable, and the rates of CNS-IRIS among patients receiving HAART is lower than that of OIs among patients not receiving HAART.¹⁵ CNS-IRIS occurs in both children and adults with the following risk factors: a very low CD4 + count prior to starting HAART, an untreated OI, a rapid response to HAART, and possibly an underlying genetic predisposition.¹⁴

The pathogenesis of IRIS is poorly understood and may vary depending on the offending pathogen.¹² The list of infectious agents associated with CNS-IRIS is extensive and includes Toxoplasma, Cryptococcus, Varicella Zoster virus, Cytomegalovirus, JC virus, HIV, Mycobacterium tuberculosis (TB) and Mycobacterium avium complex, among others.^14,16–19 Of these, PML caused by JC virus and Cryptococcal meningitis are two of the most common OIs associated with the development of CNS-IRIS.¹²

A diagnosis of CNS-IRIS should be considered in the setting of deteriorating neurological status and new or worsening neuroimaging abnormalities following the initiation of HAART in a patient with HIV. Clinically, CNS-IRIS can range from mild to fulminant and include symptoms related to an existing or previous OI or new symptoms, including fever, lymphadenopathy and headache.¹⁶ Cultures may be negative, as antigen from a past OI may be sufficient to produce CNS-IRIS. Laboratory results suggestive of CNS-IRIS include an order of magnitude or greater decrease in viral load and an increasing CD4+ T-cell count.¹⁴ While CNS-IRIS is often made as a diagnosis of exclusion, T-cell infiltration on pathology confirms the diagnosis. The differential diagnosis for a patient with suspected CNS-IRIS includes OIs, cerebrovascular accidents, CNS primary malignancy, autoimmune disease, and drug toxicities.¹⁷

The particular neuroimaging findings associated with CNS-IRIS depend on the OI associated with the inflammation. As opposed to non-IRIS PML, IRIS PML typically exhibits white matter lesions that are hypodense on CT and hyperintense on T₂ and fluid-attenuated inversion recovery (FLAIR) MRI sequences (Figure 2a,b). These lesions may enhance, have peripheral restricted diffusion and/or exert mass effect.^12,19 In Cryptococcal meningitis IRIS, leptomeningeal enhancement is common, often associated with communicating hydrocephalus.¹² Such enhancement is uncommon in Cryptococcal meningitis prior to beginning HAART. Manifestations of TB IRIS most commonly include meningitis and space-occupying lesions suspected to be tuberculomas, which are seen as low-density lesions on non-contrast-enhancing CT with perilesional oedema.^16,18 Meningeal enhancement and communicating hydrocephalus suggest TB IRIS meningitis rather than worsening TB meningitis, although both presentations can be similar.¹⁸ Toxoplasma encephalitis IRIS may also be difficult to distinguish from its non-IRIS counterpart; therefore, clinical presentation, increased T₂ FLAIR signal abnormalities and enhancement of the leptomeninges should be used to diagnose IRIS.¹⁶ CMV IRIS can present as worsening uveitis or encephalitis, with typical MR findings including solitary focal mass lesions and ventriculitis.¹⁶Although the imaging presentations of CNS-IRIS may differ by pathogen and location, when imaging findings are corroborated with an increased inflammatory response in conjunction with laboratory results indicates a reconstituting immune system. Most importantly, resolution of the imaging findings on follow up examination almost immediately after stopping HAART, clinches the diagnosis of CNS IRIS (Figure 2c).

Figure 2.

Baseline axial FLAIR (a) prior to the start of antiretroviral medication reveals no cortical or subcortical FLAIR signal. While on antiretroviral therapy, the patient developed seizures and new FLAIR imaging (b) reveals new nonspecific subcortical signal abnormalities. Antiretroviral treatment was held immediately with repeat brain MRI 9 days later (c) reveals significant improvement in signal abnormalities. FLAIR, fluid-attenuatedinversion recovery.

Anti-tumour necrosis factor therapy-induced demyelinating disease

Inflammatory bowel disease (IBD) may involve multiple organ systems with varying levels of clinical severity and differing treatment strategies. The treatment of IBD primarily relies on medical therapies, including sulfasalazine, mesalamine, glucocorticoid, 6-mercaptopurine, azathioprine, methotrexate, cyclosporine and TNF-alpha inhibiting agents, such as infliximab, natalizumab, adalimumab or certolizumab pegol. Severity of disease frequently dictates the preferential selection of these medications.²⁰

Although TNF-alpha inhibitors may provide significant benefit for IBD, a history of demyelinating disease may be considered as a contraindication to therapy due to known and unwarranted demyelinating side effect associated with drugs like adalimumab. Trials with adalimumab reported demyelinating process in 0.2% of 3160 patients.²¹ Most commonly, TNF-alpha inhibitor-associated demyelination involves the development or exacerbation of multiple sclerosis (MS) or other CNS demyelinating process. Demyelinating peripheral neuropathies may also be seen with TNF-alpha inhibitors, and have been reported in association with Guillain-Barre syndrome, Miller Fisher syndrome, chronic inflammatory demyelinating polyradiculoneuritis-like neuropathy, multifocal motor neuropathy, Lewis-Sumner syndrome and small fiber sensory neuropathy. In addition, axonal sensory and motor polyneuropathy with or without encephalopathy has been reported.^22–29

MR imaging features of TNF-alpha inhibitor-associated demyelinating disease is similar to the imaging features seen in MS and includes T2 and FLAIR hyperintense signal within the white matter, although gray matter can also be involved (Figure 3a).³⁰ In the active phase, homogenous or ring contrast enhancement of the lesion is expected (Figure 3b).^31,32 Restricted diffusion should not be present in the lesion and excludes acute infarct as an alternative diagnosis.

Figure 3.

Axial FLAIR (a) and axial T₁ post-contrast (b) MR images demonstrate an enhancing left thalamic lesion without restricted diffusion (not shown). Repeat axial FLAIR and axial T₁ post-contrast MR imaging obtained 3 months (not shown) following the cessation of adalimumab demonstrates resolution of enhancement and slightly decreased left thalamic signal abnormality. FLAIR, fluid-attenuatedinversion recovery.

The mainstay of treatment of TNF-alpha inhibitor-associated demyelination is simply cessation of the medication, which results in resolution of symptoms and signs and findings on follow up imaging. The long-term prognosis of demyelinating disorders is not well established and can be unpredictable, even after discontinuation of the medication. Chronic neuropathy or spontaneous relapse after medication discontinuation has been reported, while other patients stabilize after drug dosage reduction.²⁵ The limited data available suggest that the subsequent management of these patients is likely similar to that of idiopathic demyelinating diseases.²⁵

The new onset of demyelinating lesions in patients without prior history of MS after anti-TNF-alpha therapy has been reported and provides support for the argument that post-treatment demyelinating lesions arise as a direct result of the therapy. However, it is also possible that these demyelinating lesions may represent previously undiagnosed, subclinical MS or that some patients requiring anti-TNF-therapy for diseases, such as rheumatoid arthritis, may be at higher risk for demyelinating disorders.³³ Despite the debated theories, there is ongoing research into the causation and association.

Progressive due to natalizumab in multiple sclerosis

Current therapies in MS are focused on modulating and suppressing the immune system. Natalizumab is a monoclonal antibody against alpha-4-integrin used in relapsing MS. Although significant benefit has been demonstrated in its efficacy to reduce relapse rate, disability progression, and MRI-detectable activity, the use of natalizumab is limited by the risk of PML, which is also described as an IRIS.³⁴ Recent estimate of the overall PML incidence in natalizumab-treated patients was 3.78 cases per 1,000 patients [95% CI (3.46–4.12) per 1,000 patients].³⁴ Three significant risk factors for natalizumab-associated PML are John Cunningham virus (JCV) seropositivity, prior use of immunosuppressants, and treatment duration ≥2 years.³⁴ Due to the established risk of PML while on natalizumbab, a risk-benefit stratification is typically used prior to initiating treatment. Although PML is usually diagnosed during natalizumab therapy, it has been reported to occur up to 109 days after natalizumab cessation.³⁵ A patient’s serum anti-JCV antibody titers are followed closely before initiation and during the treatment period. Detection of JCV DNA in CSF or brain biopsy is required for definitive PML diagnosis. During the early stages of the disease process, JCV DNA may not be detectable in the CSF, confounding the diagnosis.

MRI is the most sensitive imaging tool in the detection of PML lesions, and imaging findings can be present up to months before clinical symptoms occur.³⁶ The typical MR imaging characteristics in natalizumab-induced PML include large areas of T₁ hypointense signal and subcortical and juxtacortical white matter T₂/FLAIR signal abnormalities without mass effect in the early phase of disease. Over time, there is progressive involvement of the cortical gray matter (Figure 4a), most commonly involving frontal and parieto-occipital lobes. Gray matter structures such as basal ganglia can also be involved. Post-contrast images demonstrate areas of punctate contrast enhancement (Figure 4b).³⁷ Diffusion-weighted imaging may either show true restriction or T₂ shine-through.^36,38

Figure 4.

Axial FLAIR (a) and T₁ post-contrast (b) images through the temporal lobes reveals abnormal FLAIR signal in the juxtacortical and subcortical white matter with possible subtle punctate focus of enhancement (arrows). DWI (not shown) demonstrated corresponding increased signal related to T₂ shine through and not restricted diffusion. DWI, diffusion-weightedimaging; FLAIR, fluid-attenuatedinversion recovery.

Natalizumab-associated PML carries high morbidity and mortality rates with one estimated mortality rate of 29% (n = 35).³⁹ Once diagnosed, treatment is usually plasma exchange.^40,41 Early detection of PML or detection of PML in clinically asymptomatic patients may lead to improved clinical outcomes and lower mortality rates.^{34,37,39,40,42} In addition to natalizumab, several other medications have been associated with PML, including efalizumab, mycophenolate mofetil, rituximab, and alemtuzumab, fingolimod, and dimethyl fumarate.^43,44

Ipilimumab and lymphocytic hypophysitis

Ipilimumab, a monoclonal antibody approved by the FDA for treatment of advanced melanoma in 2011, represents a new class of “elegant” chemotherapeutic drugs, which augment immune-mediated responses against malignant cells rather than direct cytotoxic effects against rapidly developing cells. This has been shown to be effective in improving both objective response rates and stable disease rates when compared with other established protocols for the treatment of advanced melanoma.⁴⁵ Ipilimumab, a monoclonal antibody, specifically targets and inhibits cytotoxic T-lymphocyte antigen 4 (CTLA4). CTLA4 is an immunomodulator, which promotes T-cell anergy by blocking the positive feedback interaction between T-lymphocytes and antigen presenting cells, with the net effect of anti-CTLA4 therapies inducing T-cell proliferation and resulting in an unchecked immune system.⁴⁶ These “immune-related adverse events” are related to patient dosing and cumulative lifetime dose.^47,48

Anti-CTLA4-induced hypophysitis was first identified in phase I trials with a 5% incidence,⁴⁹ with additional studies showing rates between 1 and 6% in treated patients. ⁴⁹ The pathogenesis of anti-CTLA4 hypophysitis is thought to be similar to lymphocytic hypophysitis, a rare disorder which generally presents in peripartum females secondary to lymphocyte predominant infiltration of the pituitary gland and stalk.⁵⁰ Typical presenting symptoms of hypophysitis include those related to pituitary gland enlargement such as headaches with visual field impairment and symptoms related to gland dysfunction and panhypopituitarism.⁵⁰

The characteristic MR imaging feature, which typically precedes development of symptoms, is a heterogeneously enhancing, enlarged pituitary gland (Figure 5a). Blansfield et al showed 7 of 8 patients developed a 60% or greater increase in craniocaudal dimension of the pituitary gland.^49,51 The mainstay of treatment is administration of steroid therapy with or without ipilimumab cessation with improvement in pituitary gland findings occurring within days.^49,51 Although some return of pituitary function may be seen, clinical improvement is slow and most patients rely on long term exogenous hormonal replacement for symptomatic control.⁴⁹

Figure 5.

Sagittal (a) contrast-enhanced MRI obtained following onset of symptoms demonstrates an enlarged and heterogeneously enhancing pituitary gland and infundibulum after treatment with ipilimumab (an antiCTLA-4 antibody) in conjunction with leuprorelin. Repeat MRI sella reveals volume loss of pituitary gland without recovery of normal pituitary function (not shown).

Pseudoprogression and pseudoregression in glioblastoma treatment

Glioblastoma (GB) is the most common primary brain tumour in adults with poor prognosis. The current standard of treatment of newly diagnosed GB is surgical resection followed by an adjuvant course of RT and long-term administration of temozolomide (TMZ).⁵² In recurrent GB, bevacizumab, an anti-vascular endothelial growth factor (VEGF) agent, is also approved for treatment.⁵³ Post-operative treatment-related imaging changes are typically seen, but the imaging changes after treatment with RT, temozolamide and bevacizumab can each result in marked changes in imaging patterns not reflecting the true effects of treatment, referred to as pseudoprogression and pseudoresponse.⁵⁴

RT, both in isolation and in conjunction with TMZ, is known to result in an MR imaging pattern termed pseudoprogression, wherein patients with GB imaged shortly after the completion of RT are found to have enlargement of the contrast-enhancing part of the lesion mimicking tumour progression, and over time, these changes ultimately stabilize or regress (Figure 6a–e).⁵⁴ Although the underlying pathophysiology is complex and incompletely understood, the mechanism involves inflammatory changes and increased vascular permeability from blood-brain barrier breakdown, resulting in oedema and leakage of contrast.⁵⁴ In most cases, the onset of radiologic changes occurs within 3 months of RT completion, and in general, are clinically asymptomatic.⁵⁵ The histologic correlation with imaging findings suggests the incidence of pseudoprogression in patients with GB treated with TMZ and RT is approximately 20%.⁵⁶ Interestingly, radiologic evidence of pseudoprogression is actually associated with improved 1 and 2 year progression-free and overall survival compared with patients without evidence of pseudoprogression.⁵⁶ This is thought to be related to methylation and silencing of the DNA promoter for MGMT, a DNA repair enzyme, which is strongly associated with pseudoprogression, and is associated with significantly improved positive response to chemoradiotherapy in GB.^57,58

Figure 6.

Axial FLAIR (a) and axial post-contrast T1 (b) obtained 1 year after the cessation of RT and temazolomide revealed an area of FLAIR signal and enhancement in the right centrum semiovale and periventricular white matter, which was thought to represent treatment effects. The repeat MRI of the brain 3 months later reveals increased FLAIR abnormality and enhancement on axial FLAIR (c) and axial post-contrast T1 (d). Concurrent perfusion MR shows no increased perfusion (e). A definitive diagnosis of radiation necrosis was made with stereotactic biopsy. FLAIR,fluid-attenuated inversion recovery; RT, radiation therapy.

MR perfusion imaging has been shown to accurately discriminate between true progression and pseudoprogression. True tumour progression is seen with small percentage increases (greater than 5%) in relative cerebral blow volume and portend a poor prognosis, while in pseudoprogression there is decrease in relative cerebral blow volume, strongly suggestive of a better clinical prognosis (Figure 6g).⁵⁹

In contrast, pseudoregression/pseudoresponse is the term used to describe the imaging pattern associated with the use of bevacizumab, whereby the contrast enhancing portion of a lesion is less than the viable tumour.⁵⁴ As previously mentioned, bevacizumab, a monoclonal antibody, binds to VEGF, thereby inhibiting angiogenesis.⁶⁰ GBs are known to highly express VEGF, which results in increased tumour neovascularity, and bevacizumab is hypothesized to be potentially effective in the treatment of GB.⁵³ The imaging changes following initiation of bevacizumab in clinical trials were frequently dramatic, with marked decrease in size or complete resolution of areas of abnormal enhancement.^61,62

However, further research has shown that despite improved progression-free survival, there was no benefit to overall survival.^63–65 Specifically, recent Phase III clinical trials evaluating the benefit of adding bevacizumab to RT and TMZ as first line therapy demonstrate no improvement in overall survival by adding bevacizumab to the regimen.^63,64

Bevacizumab is hypothesized to cause rapid changes in vascular permeability, resulting in “normalization” of the blood–brain barrier⁶² This prevents gadolinium from effectively crossing the blood–brain barrier and accounts for the dramatic decrease in lesion enhancement. Additionally, the same mechanism results in corresponding improvement in abnormal T₂/FLAIR hyperintense signal.⁶² Although decreased contrast enhancement of a lesion can represent pseudoregression or true regression, correlation with abnormal T₂ and FLAIR and diffusion-weighted imaging signal abnormalities increase interpretation accuracy. A corresponding decrease in T₂/FLAIR hyperintensity is supportive of a true response, while an increase in T₂/FLAIR hyperintensity and increase in restricted diffusion is suggestive of progressed infiltrating malignancy despite the lack of worsening enhancement (Figure 7a–e).⁶¹ As a result, T₂/FLAIR signal changes are evaluated in conjunction with contrast enhancement in newer revised assessment neuro-oncology criteria used for the evaluation of GB therapeutic response.⁶¹

Figure 7.

Axial FLAIR and T1 post-contrast images (a, b) obtained 6 months after the initial diagnosis while on maintenance with Temodar demonstrates progression of confluent left frontal and basal ganglia signal abnormalities with new cortical and subcortical left occipital enhancing lesion when compared with prior examination (not shown). Repeat images (c–e) obtained 2 months following the above images and approximately 1 month after the patient started treatment with Avastin shows improved left frontal signal and enhancement, but increase in size of left occipital lobe signal abnormality with new restricted diffusion and resolution of enhancement. The left occipital lobe area represents an area of pseudoresponse/pseudoregression. FLAIR, fluid-attenuated inversion recovery.

Advanced MR approaches for the evaluation of tumour response are also under investigation; these include diffusion, spectroscopy, and perfusion imaging. PET has shown promising results in the assessing GB treatment response. Several studies have looked at the correlation of MRI and PET in diagnosis and in particular the correlation of decreased ADC map and increased PET uptake compared with gadolinium enhancement and increased PET uptake.⁶⁶ Although further studies are needed, PET in conjunction with MRI may provide more accurate assessment of tumour response in the future.

Conclusions

A working knowledge of both the clinical management of the patient and treatment-induced imaging abnormalities is essential to accurate interpretation and diagnosis. Through this review we provide a template for learners, radiologists and subspecialist to employ in image interpretation to provide value to our clinical colleagues and more importantly, our patients.