Recent advances in the biology and treatment of brain metastases of non-small cell lung cancer: summary of a multidisciplinary roundtable discussion.

This article is the result of a round table discussion held at the European Lung Cancer Conference (ELCC) in Geneva in May 2017. Its purpose is to explore and discuss the advances in the knowledge about the biology and treatment of brain metastases originating from non-small cell lung cancer. The authors propose a series of recommendations for research and treatment within the discussed context.

Introduction

Recent insights into the biology of non-small cell lung cancer (NSCLC) have led to a wealth of novel therapies, including targeted agents and immune checkpoint inhibitors with significant clinical activity. So far, there are limited data on the efficacy of these drugs in patients with brain metastases (BMs) but intracranial responses have been documented in emerging studies.

At the meeting, a multidisciplinary group of experts discussed the biology of BMs as well as the anatomy of the blood–brain barrier (BBB). The group considered treatment options for NSCLC and their effect on BMs, focusing on targeted treatment and combination treatment for epidermal growth factor receptor (EGFR) mutated NSCLC and those with anaplastic lymphoma kinase (ALK) rearrangement.

Incidence of BMs in NSCLC

BMs are the most common intracranial neoplasms with significant morbidity and mortality.1 2 In lung cancer, 30–50% of patients will be diagnosed with BMs during their disease, with rising frequency because of the availability of novel imaging techniques and improved survival rates. Fifty per cent of lung cancer BMs occur at disease presentation and 50–60% as the only site of distant disease. BMs often present as multiple lesions, although in one third of patients BMs are singular.

BMs occur initially in 20% of patients with NSCLC,1 in 10–20% with advanced NSCLC,3 with numbers as high as 40–50% in those with stage III lung adenocarcinoma,4 20–40% in those with ALK-rearranged tumours,5 and 45–70% in those who have ALK-rearranged NSCLCs and have been pretreated with an appropriate tyrosine kinase inhibitor (TKI).3

EGFR-activating mutations are present in about 10–20% of white patients with NSCLC.6 7 In patients with EGFR mutation, the incidence of BMs at the time of diagnosis is 25%, which is slightly higher than in unselected patients, suggesting that EGFR mutations might be associated with a metastatic tropism to the brain and then with an increased risk of BMs.8 9 Furthermore, the brain is a common site for relapse of disease in patients previously treated with TKIs in about 30–60% of EGFR-mutated NSCLCs.10

Prognosis of BMs

Overall survival (OS) of patients after the diagnosis of BMs remains poor with significant clinical problems. The prognosis depends on the patient’s age and performance status, the type of the primary tumour, the time from diagnosis of the primary, the overall disease activity, and the location and extension of extracranial and intracranial disease.11–14

Biology and molecular alterations of NSCLC BMs

There are many biological aspects of growth of metastases in the brain for which scientific progress has been made and where further progress in our understanding will be helpful in developing new treatments. This extends from the biology of brain colonisation by metastatic cells from the initial stages (asymptomatic metastases) to advanced stages when the disease is clinically diagnosed.

The use of experimental models has allowed construction of a sequential map showing key mechanisms of a metastatic cell developing in the brain (see figure 1).

Figure 1

Key mechanisms of a metastatic cell developing in the brain (see reference22).

Crossing the BBB

The ability of cancer cells to cross the BBB involves general and specific mediators.15–18

The naive brain microenvironment

After extravasation most metastatic cells die, partly due to reactions of the brain microenvironment responding to their presence.19–21 This suggests that a naive brain microenvironment initially repels many potential cancer cells. Some molecular mediators of natural defences have been reported.20 21 It is also true, unfortunately, that some metastatic cancer cells avoid this initial bottleneck by blocking anti-tumour components of the reactive microenvironment.20 21 The surviving cells closely interact with pre-existing blood vessels in the brain by vascular co-option and colonisation of the crucial perivascular niche.19 21 22 This is all mandatory to the final progression to brain macrometastases.

Vascular co-option and dormancy

This process is mediated by cancer cell adhesion molecules and integrins,21 23 as well as by secreted molecules from the co-opted endothelial cells.24 25 Brain metastasis can manifest many years after the diagnosis of the primary tumour. Dormancy/quiescence is thought to play an important role in brain metastasis and recent data have started to show mediators of this biology,26 which is also linked to the perivascular location. Eventually these cells will re-awake and start to grow aggressively. The mechanism mediating this process has been linked to the ability of cancer cells to recognise components of the basal lamina.23 27

Angiogenesis

Additional ways to interact with the vasculature involve the formation of new blood vessels, called angiogenesis. This appears crucial for the formation of BMs from lung adenocarcinoma, in an entity-specific and organ-specific manner. Interestingly, inhibition of this early angiogenic switch prevented metastases from outgrowth and arrested them in a microscopic state, making angiogenesis inhibition an interesting approach for BM prevention in lung adenocarcinoma.19 28

Invasive fronts

Even though the mechanisms of brain colonisation are more linked to the initial stages of metastasis, some might also apply to advanced stages of the disease and thus offer opportunities for improved therapies. For instance, invasive fronts have been described in 50% of BMs, which in many cases correlates with the process of vascular co-option.29 The correlation of invasive fronts with poor prognosis in patients with BMs30 might offer a therapeutic window to target mediators of vascular co-option after surgery to reduce local relapse.

Genomic alterations

Therapies, including targeted ones that are effective outside the brain, can fail in this organ; reasons for this might include the genomic and other molecular divergence of BMs compared with primary tumours and other metastatic sites.31–37 Although still in need of more evidence for their contribution to BM progression, specific genomic alterations found in brain metastasis offer actionable mutations to be exploited. In addition, BM-specific mutations could also be found in liquid biopsies from blood or cerebrospinal fluid (CSF) and have been used to evaluate response to therapy and the presence of residual disease.33 36 This emerging field of BM would benefit from the use of recent advances in genetic engineering using CRISPR/Cas9 combined with experimental models.

Cancer cell–brain microenvironment interaction as target for therapy and prevention

The divergent evolution of metastatic cells in the brain might respond to the significant pressure generated by its environment that is made of different cellular components, and homeostatic regulation. The microenvironment thus might be a crucial pillar to explain the specificity that applies to brain metastasis. Close cancer cell–microenvironment interactions have created an interesting scenario, where experimental therapies have probed the critical support that cancer cells receive from altered components of the microenvironment. For instance, established BMs have been shown to assemble gap junctions with surrounding reactive astrocytes.38–42 This interaction can allow metastatic cells to detoxify themselves from the accumulation of potentially toxic metabolites generated by various sources of stress, including chemotherapy.42 A combination of BBB-permeable drugs targeting gap junctions has recently supported the potential of targeting interactions with the microenvironment. This might offer BM-specific therapies and prevention strategies in the future, possibly independent of the tumour entity.

The relevance of the BBB for medical treatment of NSCLC BMs

The BBB is the physical, chemical and metabolic barrier that segregates blood from the interstitial fluid of the central nervous system (CNS) for protection of the CNS against overexposure, pathogens and toxins. Molecules with a molecular weight over 500 Daltons (the molecular weight of 98% of drugs) are generally considered to not readily cross an intact BBB, but other physico-chemical properties also influence brain penetration. This is important for primary and secondary prophylaxis; however, mixed responses have been seen.

The rising incidence of BMs may be partly due to the fact that some therapeutic compounds can control tumour growth outside of the CNS but do not, or only partially, penetrate the BBB.43 Therefore, tumour cells that have successfully invaded the brain may not be affected by these agents, making the brain a potential ‘sanctuary site’ for cancers.44

There is an ongoing controversy about the role of the BBB breakdown in affecting the activity of systemic therapies for BMs.45 46 In health, but also many CNS diseases, the specific anatomical and molecular constitution of the BBB limits access of the vast majority of molecules to the brain: specialised endothelial cells connected by tight junctions, the vascular basement membrane, pericytes, astrocytic foot processes, and specialised transporter systems strictly regulate extravasation, while active exclusion mechanisms like glycoprotein P (P-gp), breast cancer resistance protein (BCRP) and the family of multidrug-resistant proteins exclude xenobiotics effectively.47 Thus, the BBB remains a complex obstacle for drug delivery to the CNS. Several techniques have been tested to direct therapeutics across the BBB, including disruption of the BBB, modification of drugs, inhibition of efflux transport, and Trojan horse approaches that use endogenous transporter properties of the BBB.48 One problem with those approaches is that even if an active compound can cross the endothelium, it is not guaranteed that it will reach the target cell. Recent evidence suggests that lowering the affinity of an antibody directed against the transferrin receptor allows for greater release of the antibody on the abluminal surface of the vessel, and entry into the brain parenchyma.47 49

The majority of brain macrometastases, that is, metastases of more than 1 mm diameter which are detectable with common imaging techniques, do show signs of disturbance of the BBB, although to a varying extent.50–53 Therefore, the challenges of crossing the normal BBB do not fully apply to BMs, even though some aspects of the BBB are preserved in BMs. It is a matter of debate whether a BBB breakdown in brain tumours allows penetration of systemic chemotherapies to the single cancer cells of the brain tumour in sufficient concentrations. In clinical specimens, highly variable tumour levels have been reported for different agents.54 55 Of note, lapatinib and trastuzumab, two agents with no significant activity against breast cancer BMs, can be found in relevant concentrations in BMs in clinical and preclinical specimens, which makes it highly likely that the BBB is only partially relevant for the lack of CNS activity of some drugs.55 56 In accordance with this, it has been demonstrated that trastuzumab-emtansin (T-DM1), a derivate of trastuzumab, is able to show signs of clinical effectivity in HER2-overexpressing BMs, further supporting the notion that it is not the BBB penetration but other microenvironmental mechanisms in the brain that make certain drugs ineffective.57

It has been shown that increased BBB permeability is associated with accelerated metastasis growth.58 Two closely related mTOR/PI3K inhibitors, one of them with a minor chemical modification that allows the two main exclusion transporters constituting the BBB (P-gp and BCRP)59 to be bypassed, had different effects on these metastases: while the BBB non-permeable inhibitor only affected permeable metastases, the BBB permeable one had strong anti-tumour effects on non-permeable micrometastases, and even dormant cancer cells in the brain.58 Furthermore, nuclear morphology changes and single cell regression patterns implied that both inhibitors target cancer cells independently of their relative position to the blood vessel, making BBB permeability the limiting step for drug diffusion to cancer cells in the brain.58 Another preclinical study found a highly variable uptake of doxorubicin and paclitaxel of different metastases from the same breast cancer cell line, so that cytotoxic concentrations were reached in only 10% of the most permeable metastases.52 It is widely assumed that classical chemotherapies with proven activity on systemic metastases of many cancers have limited, if any, activity on BMs,60 probably with the exception of primary chemotherapy of lung cancer BM.61 This can be due to a lack of sufficient BBB breakdown to allow primary extravasation of the drug and rapid secondary exclusion by P-gp, but also specific resistance mechanisms that are different in the brain, such as protection of extravasated cancer cells by astrocytes38 or other brain resident cells.

Treatment of NSCLC BMs

Overview

For a long time, BMs in lung cancer have been considered a final event and were treated either by whole brain radiation therapy (WBRT) or palliative care. However, since the arrival of new systemic and targeted therapies, more effective treatments for BMs are available with the aim to increase local control, and if possible survival, without affecting neurocognition.

Current treatment algorithms of NSCLC BMs offer symptom control measures and therapeutic measures. Modern disease-directed management includes:62

surgical resection,

radiotherapy,

chemotherapy,

targeted drugs,

multi-modality approaches.

Limited metastatic lesions

For limited metastatic lesions (one to three metastases) neurosurgical resection is one of the main therapeutic options, with stereotactic radiosurgery (SRS) being the main alternative, known to be equivalent to surgery in term of local control.

Resection can also be combined with radiotherapy, such as SRS or WBRT. These combinations have been addressed in two recent randomised trials, which showed that postoperative SRS was associated with a significant increase in local control compared with observation63 and that postoperative WBRT was associated with an increase in neurocognitive deterioration compared with postoperative SRS without any difference in OS.64

Another approach is the combination of SRS and WBRT. One recent randomised trial has shown that even if WBRT combined with SRS is associated with better brain control, WBRT induces significant higher neurocognitive deterioration compared with SRS alone, without any difference in OS.65 However, another randomised trial comparing WBRT plus SRS with SRS alone showed that for a subgroup of patients with good graded prognostic assessment, a benefit of adding WBRT to SRS was obtained in OS.66

While treating patients with WBRT, neurocognition can be preserved by performing hippocampal sparing67 or adding mementine.68 A randomised phase III clinical trial aiming to compare time to neurocognitive failure between WBRT plus mementine to WBRT with hippocampal preservation and mementine is currently being performed (ClinicalTrials.gov identifier: NCT02360215).

Multiple metastatic lesions

In multiple metastatic lesions (more than three BMs) WBRT is still an option for most patients, alone or in combination with SRS, a radio-sensitiser or chemotherapy. However, SRS on more than four and up to 10 BMs is feasible,69 with no more late toxicity in neurocognition compared with patients with one to four brain metastases.70 Some patients, especially those with a poor performance status, receive chemotherapy or steroids alone.71 The addition of targeted drugs such as erlotinib as radio-sensitisers to WBRT has failed to show benefit in local controls or OS but has increased toxicity.72–77 The addition of chemotherapeutic agents such as temozolomide to radiation has also failed to improve survival but increases toxicity.78–82

Systemic chemotherapy

Chemotherapy plays a limited role in the treatment of BMs because of its inability to cross the BBB. However, response rates as high as 30–40% have been reported in the brain with platinum-based chemotherapy, similar to rates observed extracranially.61 83

Targeted drugs

EGFR TKI therapy

Among patients with NSCLC with EGFR mutations, TKIs seem more effective than chemotherapy in controlling intracranial disease. EGFR TKIs are low molecular weight organic compounds with low to moderate CSF penetration rates differing between first-generation to third-generation drugs.84 85

EGFR TKIs of the first generation, such as gefitinib and erlotinib, and of the second generation, such as afatinib, have recently been integrated in the treatment algorithm of advanced metastatic mutated NSCLC as first-line therapy, replacing conventional chemotherapy because of improved response and survival rates.86–88

Retrospective data and phase II study experiences have indicated that gefitinib and erlotinib have significant intracranial activity.89 90

For afatinib, phase II data, results from a compassionate use programme as well as pre-specified subgroup analyses suggest significant intracranial efficacy. This substantiates preclinical and clinical observations that afatinib can penetrate the BBB at concentrations sufficient for initiating anti-tumour activity.91–93

EGFR TKIs of the third generation, such as AZD 3759 and osimertinib, have recently accelerated the debate over the role of modern targeted therapy for the treatment of BMs in mutated NSCLC as a potential substitute for brain radiation. This is because of preclinical and clinical evidence proving them to be more effective, showing a promising blood brain penetration and the potential to overcome EGFR TKI resistance. They also challenge the concept of upfront WBRT by being potentially more effective but less neurotoxic.94–97

Preclinical studies have shown that osimertinib induces sustained tumour regression in an EGFR-mutated PC9 mouse brain metastasis model,36 and exhibits a greater distribution into mouse brain tissue than gefitinib, rociletinib or afatinib. Clinically, osimertinib has greater efficacy than platinum/pemetrexed in patients with T790M-positive NSCLC, including those with CNS metastases in a second-line setting.86 95

AZD 3759 has primarily been designed for crossing the BBB. Clinical experience for AZD 3759 exists from a phase I study in pretreated patients. By dosing up to 300 mg twice a day, there has been a significantly higher tumour shrinkage intracranially than extracranially. Grade 4 toxicity of less than 10% was reported for rash, diarrhoea and pruritus.96 97

TKIs for patients with ALK-rearranged NSCLC

Currently, there are five compounds registered for patients with NSCLC and ALK rearrangement: these are the TKIs crizotinib, ceritinib, alectinib, lorlatinib and brigatinib. All these compounds have been registered with the US Food and Drug Administration (FDA); crizotinib, certinib and alectinib have also been approved by the European Medicines Agency (EMA) with an approval of brigatinib currently pending.

The development of the ALK-directed TKI crizotinib took a rather short time between the discovery of the importance of ALK rearrangement and the introduction of the drug. In the PROFILE 1014 trial, crizotinib showed a significant improvement in progression-free survival (PFS) compared with chemotherapy in patients with ALK rearrangement.98

The incidence of BMs constitutes a major problem in patients with NSCLC and ALK rearrangement. About one third of TKI-resistant tumours harbour ALK mutations, including an amplification which occurs in 10% of mutations of the remaining 25%.99–103

Therefore, the question arises whether the CNS acts as a ‘sanctuary’ for the development of metastases, as up to 70% of recurrences occur within this anatomic area.

Drugs developed after crizotinib, targeting ALK rearrangement, such as ceritinib, alectinib and brigatinib, have the ability to induce a remarkable CNS response in patients who have been pretreated with crizotinib. They have quite a different side-effect profile however.

When considering alectinib, responses in the CNS were complete in 20% in patients with measurable CNS metastases,104 whereas the use of brigatinib in the identical setting produced intracranial overall response rates in 42%–67% of patients.105

Very recent data have shown a significant superiority of alectinib over crizotinib in untreated ALK-positive NSCLC regarding the duration of PFS and the time until CNS progression.106

Nevertheless, the question of the best treatment sequence – if any - emerges and will have to be the topic of further clinical investigations.

Radiotherapy

Apart from surgery and targeted drugs, radiotherapy (especially radiosurgery or hypo-fractionated stereotactic radiotherapy (HFSRT)) is one of the main weapons to increase local control, and if possible survival, without affecting neurocognition.

Because the combination of WBRT with SRS or surgery does not increase OS but neurocognitive deficit65 107 108 and because SRS alone compared with the combination of SRS with WBRT has been shown to lead to the same OS and less neurocognitive deficit, with a shorter time to intracranial failure, SRS is now considered a standard treatment for patients with BMs.

A study from 2016 showed the relevance of postoperative SRS compared with observation, bringing better local control without toxicity and no difference in OS.109 More recently this has been confirmed by a randomised clinical trial showing that postoperative SRS led to significantly higher local control than observation, with the same OS.63 The trial showed a higher benefit of postoperative SRS for small cavities (0–2.5 cm) compared with large ones.

Another randomised trial comparing postoperative radiation with WBRT plus SRS on non-resected BMs versus SRS on the cavity of resected metastases plus SRS on non-resected BMs showed that postoperative WBRT led to a higher neurocognitive deficit and the same OS compared with SRS alone. However, this trial showed poorer local control and worse brain control for patients treated with postoperative SRS compared with those treated with WBRT.64These conflicting results could be due, at least in part, to the presence of microscopic tumour infiltration not targeted by postoperative SRS. If local control is important, we have to address and aim to obtain better brain control without neurocognitive deficit, as well as a better OS. This is when the combination of radiotherapy, especially SRS or HFSRT, with targeted drugs or immunotherapy comes in to optimise BM treatment.

Combination and sequencing of medical therapies with radiotherapy for NSCLC BMs

The irradiation anti-tumour effect is driven by direct and indirect effects. Irradiation can induce tumour cell death as mitotic cell death, apoptosis, but also autophagy and senescence.110

SRS or HFSRT acts through the induction of apoptosis of endothelial cells, thus leading to tumour radio-sensitisation.111 More recently it has been shown that irradiation can induce an immune cell death through CD8 T-cell infiltration and by the stimulation of tumour antigen presentation.112 113 It has been shown that SRS could induce the expression of programmed death ligand 1 (PDL1) in tumours and that association of SRS and programmed cell death protein 1 (PD1) treatment led to radio-sensitisation in preclinical models.114

In addition to the local effect of radiotherapy in combination with immunotherapy, radiotherapy is also able to induce an abscopal effect, that is, an anti-tumour effect outside the irradiation field. This could be of great interest in tumours with high metastasis potential, such as lung cancer.

Radiotherapy and TKIs

To optimise the effect of SRS or HFSRT in NSCLC BMs, the combination of targeted drugs with such irradiation is a promising treatment. EGFR-mutated as well as ALK-positive NSCLCs have a higher risk of BMs, and EGFR as well as ALK pathways are known to control radio resistance. The association of EGFR inhibitors or ALK inhibitors with radiotherapy will lead to radio-sensitisation. Even if such inhibitors already penetrate the BBB, radiotherapy is known to disrupt the BBB and will help these inhibitors to penetrate. Several studies have shown the relevance of the combination of SRS or WBRT with TKIs with regards to intracranial progression but also, for some of them, in terms of OS. A pooled analysis showed that the combination of radiotherapy and TKIs had significant benefits in terms of objective response rate, time to intra-cranial progression and OS.115 A recent retrospective study showed that WBRT and TKI treatment led to longer time to intracranial progression compared with SRS and TKI, or TKI alone.116 Another retrospective study showed that patients with exon 21 mutation, when treated with WBRT and TKI, had a significantly higher OS and PFS compared with those treated with TKI alone.117 No difference was seen for patients with exon 19 deletions. Also, a recent study showed that performing radiotherapy (SRS or WBRT) before TKI treatment significantly increased the median OS compared with radiotherapy only in the case of failure,118 suggesting that SRS before EGFR TKI treatment is better than TKIs alone, at least for patients with exon 21 mutations. However, randomised trials need to be performed in this area.

A few trials are currently being undertaken associating SRS with ALK inhibitors and these should be developed.

Radiotherapy and immunotherapy

Another possibility is the combination of immunotherapy and radiotherapy, particularly SRS with checkpoint inhibitors. These combinations have been mostly studied in melanoma BMs, with SRS and ipilimumab treatment, but anti-PD1 and SRS combinations have also been reported.119 120 Some studies report encouraging results even for OS, while others do not.121 Some recent retrospective studies have shown that SRS performed before and concurrently to immunotherapy would have better results than SRS performed after.120 122 123 Because of the incidence of pseudo-progression with these combined treatments, the evaluation of their efficacy needs to be performed with multimodal imaging (see figure 2). Again, clinical trials for the evaluation of such combinations in NSCLC BMs in patients without mutation, but also in those with EGFR mutation, are needed.

Figure 2

Pseudo-progression after sterotactic radiosurgery (SRS) and anti-programmed cell death protein 1 (PD1) treatment for a lung brain metastasis. Before SRS treatment (A) and 6 months after SRS and anti-PD1 treatment (B); increase in the irradiated brain metastasis on T1 gadolinium MRI, without vascularisation on perfusion.

Neurosurgical resection of NSCLC BMs

Neurosurgical resection of BMs in patients with NSCLC is an indispensable treatment option in the multimodal management of such tumours. The subsequent initiation of postoperative radiotherapy in these tumours has demonstrated a positive impact on OS.124 125 Furthermore, the extent of resection in NSCLC BMs is an important factor for patient prognosis. Thus, a complete ‘macroscopic’ removal of surgically treated BMs results in a significantly better patient prognosis than an incomplete tumour resection.126 However, local recurrence of BMs after neurosurgical resection is not uncommon in clinical practice even after macroscopic complete resection and postoperative radiotherapy.

It was long assumed that BMs are well demarcated from the surrounding brain tissue. In 2013 Berghoff et al found in an autopsy study that only about half of the BMs show a well demarcated growth pattern. Tumour infiltration of the surrounding brain tissue of metastases was observed in the other half of cases. Perivascular growth into the brain parenchyma distant from the brain metastasis (‘vascular co-option’) was present in 18% of cases and diffuse infiltration of the surrounding brain tissue (‘diffuse infiltration’, like in a malignant glioma) was observed in 32% of cases.29

In a recent prospective study, Siam et al found that tumour infiltration of the surrounding brain tissue is a common finding especially in BMs from NSCLC, present in 75% of cases. In some of these NSCLC BMs the tumour infiltration was observed more than 2 mm away from the resection cavity.30 Thus, tumour cells might remain in the surrounding brain tissue despite a complete ‘macroscopic’ resection of BMs and result in local recurrence.

To overcome this limitation, Yoo et al proposed a ‘microscopic total resection’ of single BMs in non-eloquent areas with additional removal of at least 5 mm of surrounding brain tissue. Such a microscopic total resection resulted in a significantly better local tumour control rate in BMs than conventional complete resections. However, this 5 mm safety margin in microscopic total resections was arbitrarily selected126 and thus a more selective tool to visualise tumour infiltration of the surrounding brain tissue of BMs would be of interest.

5-Aminolevulinic acid

One innovative approach might be the selective visualisation of brain metastasis tissue with the intraoperative fluorescence marker 5-aminolevulinic acid (5-ALA). In a recent study, Kamp et al found that BMs can be visualised during surgery with the assistance of 5-ALA in about two thirds of cases127 (see figure 3). It is not clear so far if the 5-ALA fluorescence technique is also able to visualise tumour infiltration of the surrounding tissue of BMs. This should be investigated in multi-centre studies.

Figure 3

Application of 5-aminolevulinic acid (5-ALA) induced fluorescence during resection of a brain metastasis. Parts of the brain metastasis specimen derived from resection (A) can be visualised by 5-ALA-induced fluorescence (B). After surgical resection of the brain metastasis (C), the surrounding, potentially still tumour-infiltrated, brain tissue demonstrates 5-ALA-induced fluorescence (D).

Neurosurgical interventions for the analysis of drug concentrations and biomarkers

Concept of ‘window of opportunity’ studies: measurement of tissue concentrations of antineoplastic agents in BMs

BMs have been widely considered ‘extra-axial’ lesions, thus not being restricted by the BBB. In contrast to gliomas, penetration of antineoplastic drugs from the intravascular space into the tumour tissue of BMs is less a matter of debate.62 However, solid data are scarce. Mostly, there has been indirect evidence for drug tissue penetration into BMs due to observation of any response in MRI scans after systemic chemotherapy.58 128 Recently, scores derived from blood values have been described to estimate survival of patients with BMs.129 Only a few studies are available dealing with the measurement of tissue concentrations of antineoplastic agents in BMs. In a meta-analysis of 1441 potentially relevant publications, only 12 turned out to provide solid data on tissue concentrations of chemotherapeutic drugs in BMs.54 The tissue-to-blood ratio showed huge variations between different drugs which had also been used for solid tumours with subsequent BMs. As microsurgical resection offers direct access to the tissue, exposure of the patient to systemic therapy prior to surgery would allow the tissue concentration within the specimen to be quantified. Since the intravascular space contributes hereto, pharmacodynamics and pharmacokinetics of the compound must be considered to correct the values measured accordingly. Prerequisites of such a study would be:

the drug should be in use for cancer therapy,

phase I studies are already completed,

the toxicity is known and reasonably low,

there are no side effects which could be relevant for surgery (significant immediate or early bone marrow toxicity, embolic or bleeding disorders),

the serum half-life (tissue half-life) is known to find the most appropriate timing of drug delivery in relation to tissue sampling,

there is a calculation of serum level to estimate the influence of the intravasal drug,

intra-operative pharmacokinetics if applicable.

Morikawa et al conducted such a study for capecitabine and lapatinib in BMs of breast cancer.55 Capecitabine and lapatinib were shown to penetrate to a significant but variable degree into BMs of breast cancer. However, drug delivery to the BM tissue was variable and appeared in some cases too low to be effective. Overall, the tissue concentration varied considerably between the few cases under investigation, especially according to different preoperative dosages and timing of drug administration in relation to the surgical procedure. This highlights the importance of standardising such protocols to generate meaningful data. Thus, it could be crucial to elucidate mechanisms which limit drug concentration.

Such window-of-opportunity studies could be a promising tool to obtain information about inter-individual variation of drug concentrations and optimal drug dosage of new compounds for systemic treatment of BMs to reduce the risk of using ineffective drugs (waste of opportunity) and missing effective drugs (loss of opportunity) for efficient and evidence-based planning of early phase II trials (see figure 4).

Figure 4

Concept of a window-of-opportunity study to obtain information about inter-individual variation of drug concentrations and optimal drug dosage of new compounds for systemic treatment of brain metastases.

Identification of genetic/molecular signature

Several genetic signatures have been characterised in solid tumours, which are not only relevant prognostically but determine the oncological management.62 However, little is known about whether BMs share the signature of the primary tumour and whether multiple BMs are of clonal origin with the same molecular pattern. Especially in cases where the molecular profile determines the therapeutic management, it would be mandatory to know the profile of the BMs as well. Assuming the BMs and primary lesion would not necessarily match in terms of their molecular signature, at least a stereotactic biopsy of the BMs would be necessary to tailor the treatment. The strive to personalise cancer treatment, also for BMs, might increase the demand for tissue analysis of BMs even in the case of a known primary lesion. Molecular analysis of CSF samples has been shown to be a useful tool to identify clinically relevant genomic alterations, including aberrations not found in primary tumours in patients with BMs.130 Thus, liquid biopsies from CSF may improve personalised therapy of patients with BMs.

Conclusion

The roundtable discussion highlighted the urgent need to define better treatments for prophylaxis and treatment of BMs in patients with NSCLC. The growing insights into the pathobiology of BMs and molecular treatment targets lead to novel therapy approaches that need to be tested in clinical trials enrolling patients with BMs.

Detailed studies of the biology have provided molecular mediators of critical processes in the metastatic colonisation of the brain that could offer novel avenues for translational efforts.

The naive brain microenvironment can eliminate multiple brain metastasis (BM) cells by activating innate immunity mechanisms. However, successful BM cells have developed ways to avoid them and progress in the metastatic cascade by colonising the crucial perivascular niche in the brain.

A critical process to initiate BMs involves the ability to co-opt pre-existing blood vessels, or grow by angiogenesis. Both mechanisms imply opportunities to target the growth of non-clinically detectable BMs in a preventive scenario.

Exploiting the functional contribution of genomic alterations identified in human next-generation studies on BMs will be critical to evaluate the potential of targeting actionable mutations. In addition, these alterations could be used to track BMs using liquid biopsies.

Given the specificity of the biology of BMs, the importance of the microenvironment in the progression of the disease and its potential to become a novel therapeutic target has started to be evaluated and pioneering clinical trials are ongoing.

Improvement of experimental models that include clinically relevant situations (ie, radiation, neurosurgery, targeted therapies, specific molecular alterations, spontaneous models of metastases, studies in animals with intact immune system) is critical to learn more about the biology of BMs, and to provide a foundation for successful translation into clinical practice.

Many open questions about the BBB and BMs remain, but it is not justified always to expect a limited BBB penetration and a lack of meaningful central nervous system (CNS) activity for any given drug without testing it. Recent data suggest that even large molecules like therapeutic antibodies are able to penetrate the BBB in patients with BMs, at least to some extent.

CNS metastases constitute a major problem in NSCLC, particularly in patients with ALK rearrangements and even more so in those with ALK rearrangements pretreated with crizotinib. The CNS seems to act as a sanctuary for the emergence of metastases. In such situations, treatments with ALK-rearrangement-directed tyrosine kinase inhibitors (TKIs) like ceritinib, alectinib or brigatinib (although each with different toxicity profiles) represent a viable and effective treatment option. However, the question of an appropriate treatment sequence emerges and this will have to be the topic of future clinical investigations.

The development of epidermal growth factor receptor (EGFR) TKIs has now reached third-generation agents with promising early results. Preclinical and clinical evidence shows them to be more effective, having a better blood brain penetration and the potential to overcome EGFR TKI resistance. Clinically, osimertinib has greater efficacy than platinum/pemetrexed in patients with T790M-positive NSCLC, including those with CNS metastases in a second-line setting. AZD 3759 has primarily been designed for crossing the BBB.

The combination of radiotherapy and TKIs or immunotherapy is a promising treatment for NSCLC BMs but needs to be developed and validated in biologically driven clinical trials in terms of dose and timing. The trials should be associated with neurocognitive studies as well as metabolic imaging to distinguish progression from pseudo-progression.

The surrounding tissue of NSCLC BMs represents an important target in the resection as well as in the medical treatment and radiosurgery/radiotherapy of such tumours. In surgery, a ‘microscopic total resection’ with additional safe removal of the surrounding brain tissue of about 5 mm in non-eloquent BMs showed improvement of the local control rate. Intraoperative markers for selective visualisation of tumour infiltration into the surrounding brain tissue have to be further investigated.

Window-of-opportunity studies could be a promising tool to obtain information about inter-individual variation of drug concentrations and optimal drug dosage of new compounds for systemic treatment of BMs. They could reduce the risk of using ineffective drugs (waste of opportunity) and missing effective drugs (loss of opportunity) for an efficient and evidence-based planning of early phase II trials.

The strive to personalise cancer treatment, also for BMs, might increase the demand for tissue analysis of BMs even in the case of a known primary lesion.