Leptomeningeal Metastasis from Non–Small Cell Lung Cancer and Current Landscape of Treatments

Leptomeningeal metastasis (LM) is a devastating complication of metastatic cancer, which occurs when neoplastic cells invade the meningeal space. LM is found in approximately 5% of patients with malignant tumors and most commonly occurs in patients with lung cancer, breast carcinoma, and melanoma (1). The postmortem series showed an incidence of LM of approximately 20% or more with many solid tumors (2–5), suggesting that LM might often be underdiagnosed. Diagnosis of LM remains challenging given the heterogeneous presenting signs and symptoms, and requires a thorough neurological examination, cerebrospinal fluid (CSF) analysis, and MRI of the brain and spine with gadolinium.

In the past, the presence of LM often represented a failure of treatments for the metastatic disease, which sometimes represents the final event for the patient and the almost total absence of effective treatments before molecular-targeted therapies and immunotherapy. The management of LM in patients with non—small cell lung cancer (NSCLC) requires a multidisciplinary approach. Treatment options include surgery for ventricular shunt placement, radiation therapy to bulky or symptomatic disease sites, systemic or intrathecal chemotherapy, molecular-targeted therapy, and immunotherapy. The recent development of molecular-targeted therapy and immunotherapy with better CNS penetration is changing the landscape of LM management, but the prognosis of LM remains dismal. This review focuses on the epidemiology, pathophysiology, clinical presentation, diagnosis, and current treatment strategies, with a particular emphasis on novel agents, including targeted therapies, for leptomeningeal disease in patients with NSCLC.

LM is found in approximately 5% of patients with malignant tumors and most commonly occurs in patients with lung cancer, breast carcinoma, and melanoma (1). The incidence of LM in patients with NSCLC was 3.4% in molecularly unselected patients and higher in ALK-rearranged (10.3%) and EGFR-mutant subgroups (9.4%; refs. 6, 7). NSCLC with mutant EGFR and altered anaplastic lymphoma kinase (ALK) genes is more likely to relapse with LM (8). Concomitant brain metastases affect one-third of the patients. The median overall survival time of patients with NSCLC with LM remains grim and ranges from 3.6 to 11 months (9, 10), mostly related to the use of novel therapies (11). The European Association of Neuro-Oncology (EANO) class IIA patients (LM with a typical linear enhancement of meninges in MRI and typical neurological symptoms, negative/inconclusive CSF cytology) had the longest overall survival (OS), whereas type I LM (LM with positive CSF cytology) patients had the shortest OS (12). A retrospective study including 149 patients with NSCLC with LM reported that treatment with EGFR-TKIs, normal CSF flow, lack of fixed neurological impairments, and good performance status were all associated with a better outcome, whereas poor performance status, high CSF protein level, and high initial CSF WBC count were associated with poor prognosis (13). Yin and colleagues developed a molecular graded prognostic assessment (molGPA) model specific for estimating survival in lung cancer patients with leptomeningeal metastases (14). In this model, positive EGFR/ALK, Karnofsky performance score (KPS) of 60 and more, and the lack of extracranial metastasis (ECM) were associated with better overall survival rate. The molGPA model stratified the patients in three groups: high, moderate, and low risk. In the training set, the median OS for high, moderate, and low risk LM patients was 0.3, 3.5, and 15.9 months, respectively (P < 0.001; ref. 14).

The brain and spinal cord are covered by meninges consisting of three layers known as the dura mater, arachnoid, and pia mater, from superficial to deep, respectively. The meningeal system can also be divided into pachymeninges, which only contains the dura mater, and leptomeninges, which contains the arachnoid and pia mater. The space between the arachnoid and the pia mater is called the subarachnoid space and is filled with CSF. Tumoral involvement of the dura mater (pachymeninges) should be differentiated from the tumoral involvement of the leptomeninges as the tumor spreads to the leptomeninges, which enables tumor cells to spread through the CSF.

Invasion of malignant cells into the leptomeningeal system is thought to occur through a variety of routes, including direct seeding from the brain parenchyma, hematogenous seeding (especially in hematologic cancers), and dura, bone, and endoneurial/perineural invasion. Cancer cells in CSF encounter physiologic challenges such as inflammation and limited micronutrient. Chi and colleagues designed a study investigating the cancer cell mechanism to overcome this effect (15). They reported that cancer cells appear to outcompete macrophages for iron, allowing them to thrive in the CSF. Chi and colleagues discovered that the iron-binding protein lipocalin-2 (LCN2) and its receptor SCL22A17 were by cancer cells but not macrophages in the CSF. These macrophages have been shown to produce inflammatory cytokines that stimulate LCN2 expression in cancer cells, but they do not produce LCN2. The LCN2/SLC22A17 system promotes cancer cell proliferation in LM animal models, whereas iron chelation treatment inhibits it (15). Boire and colleagues discovered that component 3 (C3) was upregulated in the leptomeningeal cancer cells and shown to be necessary for cancer cell growth in CSF (16). Boire and colleagues reported that the C3a receptor in the choroid plexus epithelium is activated by cancer cell-derived C3, which disrupts the blood–CSF barrier, which enables plasma components such as amphiregulin and other mitogens to enter the CSF and support cancer cell overcome mitogen-poor microenvironment of CSF (16). There is also literature suggesting an increased risk of developing LM in patients who have undergone neurosurgic metastatic resection compared with patients who have not undergone neurosurgery, suggesting iatrogenic spread (17–19).

The initial clinical presentation of LM is often subtle. Multifocal neurologic involvement in a patient with known cancer should raise clinical suspicion of LM. The initial manifestations may include cranial nerve palsies, headaches, back pain, visual disturbances, diplopia, hearing deficits, changes in cognition, radiculopathies, myelopathies, or spinal cord syndromes such as cauda equina syndrome (3). In a series of 150 patients with solid tumor LM from Memorial Sloan Kettering Cancer Center (MSKCC) who were followed up between 2002 and 2004, the most common presenting signs and symptoms were headache (39%), nausea and vomiting (25%), leg weakness (21%), cerebellar signs (17%), altered mental status (16%), diplopia (14%), facial weakness (13%), back pain (12%), leg numbness (12%), facial weakness (8%), and facial numbness (6%; refs. 1, 20). In the same cohort (20), dizziness, fatigue, gait difficulty, aphasia, vision loss, hearing loss, dysarthria, meningeal irritation, arm pain, leg pain, and bowel/bladder dysfunction were reported at presentation in ≤5% of the patients.

The differential diagnoses include subacute to chronic meningitis, skull base/dural/parenchymal metastases, primary leptomeningeal melanomatosis, and metabolic and toxic encephalopathies (21).

The diagnosis of LM is based on neurological examination, CSF analysis, and radiographic findings. Positive CSF cytology is considered the gold standard for the diagnosis of LM, although multiple CSF samplings may be necessary. The sensitivity can be increased up to 75% and 85% with a second CSF analysis (22) compared with the sensitivity of the initial CSF analysis, which is reported to be as low as 50%. For the diagnosis, treatment, and follow-up of patients with LM from solid tumors, the European Association of Neuro-Oncology-European Society of Medical Oncology (EANO-ESMO) group has developed a diagnostic flowchart that contains neurologic symptoms, imaging, and CSF cytology (23). On the basis of the combination of these three factors, LM can be classified as type I (positive CSF cytology) or type II (probable/possible), with typical MRI features and neurological symptoms. MRI findings had been classified as linear enhancement of meninges (subtype A), nodular enhancement of meninges (subtype B), both (subtype C), or hydrocephalus (subtype D). A retrospectively involving 254 patients with LM from solid tumors using the EANO-ESMO guidelines showed shorter OS in patients with type I compared with type II LM, whereas systemic or intrathecal treatment is linked to better OS in type I LM, but not in type II LM (24), albeit this has to be investigated further in larger datasets and prospective studies. Diagnostic tests and imaging techniques are discussed below.

The definitive diagnosis of LM is based on direct visualization of tumor cells in the CSF by cytology. However, positive CSF cytology has high specificity and sensitivity, between 80% and 95%, which might require repeated LP (25). Obtaining large CSF volumes (>10 mL), processing CSF specimens promptly, and obtaining CSF from the closest site for symptoms or radiologic involvement have been shown to increase the sensitivity of CSF cytology results (25, 26). Notably, 20% of individuals with clinically or radiographically unambiguous LM were reported to have a negative CSF evaluation (1, 25), in which cases the diagnosis can be made in the clinical context supported by neuroimaging findings alone (27). High protein content, low glucose concentration, lymphocytic pleocytosis, and positive cytology for malignant cells are all characteristic CSF findings of LM. Although most individuals do not have all of these characteristics, it is unusual to have a completely normal CSF examination (3, 25, 26). LP is a relatively safe procedure, but adverse events related to LP include hemorrhage, post-LP headache, bleeding, cerebral herniation, minor neurologic symptoms such as radicular pain or numbness, and back pain (28). EANO-ESMO clinical practice guidelines recommend through physical exam, CSF cytology, and cerebrospinal MRI as part of diagnostic workup for any patients with cancer who presented with concerning symptoms for LM (23). CSF cytology is also used classification of LM (Type 1: confirmed with CSF cytology, Type 2: unequivocal/negative CSF cytology), which can be used in prognostication (29). EANO-ESMO clinical practice guidelines also recommend repeating LP, if initial CSF analysis unequivocal/negative CSF in patients with suspected LM (23).

Measuring serum and CSF tumor markers in a clinically appropriate context can aid in the diagnosis of LM, as several studies have shown elevated tumor markers (e.g., CEA, PSA, CA-15-3, CA-125, MART-1, and MAGE-3 in melanoma) in CSF compared with serum, suggesting LM even with negative CSF cytology (30–34).

Emerging diagnostic techniques such as liquid biopsy can provide helpful information for LM diagnosis and monitoring. CTCs and circulating cell-free tumor DNA (ct-DNA) in the CSF are the two most well-developed biomarkers that can be detected by liquid biopsy. Several studies have found that the CellSearch approach, which uses immunomagnetic selection, identification, and quantification of CTCs in the CSF, is more sensitive than traditional cytology and MRI in detecting leptomeningeal metastases (35–37). A few studies reported that CTC in CSF could be predictive of survival in LM (38, 39); however, given the small sample size of studies, larger and prospective cohorts will be useful in determining the cut-off value (40). CTCs liquid biopsy can also isolate single CTCs to screen for genetic abnormalities common to solid tumors. One study reported a highly concordant molecular profile (89.5%) among CTCs from the CSF of patients with LM from EGFR-mutated or ALK-rearranged NSCLC and their primary tumors (39). Liquid biopsy for detecting CTCs in the CSF has been shown to increase the sensitivity and specificity of diagnosing LM secondary to epithelial tumors in the right clinical context (36, 41). One study included 81 patients with a clinical suspicion of LM but an unequivocal MRI to compare the performance of CTC in CSF and cytology at diagnosing LM; the sensitivity of CTC assay was 94% and specificity was 100%, whereas the sensitivity of cytology was reported to be 76% (42). Another study showed that a cutoff of ≥1 CSF-CTC/mL had superior sensitivity and specificity compared with integrated clinical diagnosis (clinical suspicion with positive CSF cytology or unequivocal neuroimaging findings; ref. 41). Furthermore, CSF ct-DNA analysis has been demonstrated to be effective in enhancing LM diagnosis (43). Using ct-DNA from CSF for detecting genetic alterations in brain tumors has substantially better sensitivity than using plasma (43). One study involving 26 patients with LM secondary to EGFR showed that driver genes were found in 100% (26/26) of CSF cell-free DNA (cfDNA), 84.6% (22/26), and 73.1% (19/26) of CSF precipitates and plasma samples, respectively (44). The study also reported that CSF ct-DNA was superior to plasma in detecting loss of heterozygosity of TP53 (73% vs. 7.7%, P < 0.001; ref. 44). Next-generation sequencing of CSF to risk-stratify patients with brain metastasis secondary to lung adenocarcinoma was demonstrated in research comprising 94 patients with brain metastasis secondary to lung adenocarcinoma (45). The researchers identified five molecular subtypes associated with different overall survival. They also reported that EGFR mutations in conjunction with CDK4, CDK6, MYC, and MET were related to poor outcomes (45). Zheng and colleagues compared the CSF gene sequencing of two cohorts of people. In the first cohort, patients with LM secondary to EGFR mutated NSCLC who underwent CSF and plasma genotyping before initiation of the first dose of Osimertinib, patients with an EGFR exon 19 deletions exhibited a longer median intracranial progression-free survival (iPFS) than those with an EGFR exon 21 L858R mutation (11.9% vs. 2.8%; ref. 46). The second cohort involved patients with EGFR-mutated advanced NSCLC who developed LM while on Osimertinib therapy. In cohort 2, patients with T790M loss in the CSF showed a lower median iPFS than those with T790M reserved (7.4 months vs. 13.6 months, P = 0.01; ref. 46).

Overall, liquid biopsy from CSF can be useful for diagnosis, detection of genetic mutations, and monitoring therapy responses in LM, but further studies are needed to establish clear cut-offs and standardize different techniques. There is a necessity to update the guidelines in the light of new advancements.

MRI of the brain and full spine with gadolinium enhancement is the gold standard radiologic modality for LM imaging and has been shown to be superior to CT imaging (47, 48). Sing and colleagues (49) reported that standard contrast-enhanced T1-weighted MR sequences are more sensitive than contrast-enhanced fast fluid attenuation inversion recovery (FLAIR) sequences in detecting intracranial neoplastic leptomeningeal disease. Optimally, imaging studies should be completed before lumbar puncture (LP) to prevent false-positive leptomeningeal enhancement secondary to inflammation from the LP (50). Approximately 20% to 30% of patients with LM report having a normal MRI at the time of diagnosis (51). MRI findings consistent with or suggestive of LM include leptomeningeal, subependymal, dural, or cranial nerve enhancement; superficial cerebral lesions; and communicating hydrocephalus (1, 5, 42). Ko and colleagues conducted a retrospective study to investigate MRI findings in 283 patients with LM from NSCLC and reported that positive MRI findings were suggestive of a heavier disease burden than negative imaging findings in patients with LM who died from CNS causes (52). An example of MRI involvement in LM is shown in Figs. 1 and 2.

The sensitivity of CT scan is reported to be 23% and 38%; hence, it should only be used for patients who are unable to undergo MRI (53). Radionuclide ventriculography studies using technetium-99m-DTPA (Tc-99) or 111Indium-DTPA are not indicated for diagnosis but can be useful in evaluating patency of the ventricular system before intrathecal chemotherapy administration. Abnormalities in radionuclide CSF studies have been shown to be associated with poor outcomes and increased treatment-related toxicities (54–56).

Current treatment for LM requires a multidisciplinary approach. The goals of treatment are to stabilize neurologic symptoms, improve quality of life, and prolong survival with minimal toxicity. Patients with solid tumors and leptomeningeal metastases are divided into two categories according to the US National Comprehensive Cancer Network recommendations: good risk and poor risk. According to the NCCN guidelines (57), best supportive care is recommended for patients in the poor risk category, which includes limited performance status, significant and major neurologic impairments, broad systemic disease with few therapeutic choices, bulky CNS disease, and encephalopathy, and intensive systemic treatment is indicated for individuals in the good risk category (characterized by satisfactory performance status, no substantial neurological abnormalities, limited systemic disease, and reasonable systemic treatment alternatives if needed). This classification may not apply to individuals with NSCLC who have actionable mutations, as some of the most recent molecular treatments have demonstrated high blood–brain barrier penetration and promising antitumor efficacy. The formulation of new guidelines for this group of patients is warranted in light of these promising action (58, 59).

Information to guide treatment decisions is scarce because LM accounts for a small percentage of CNS metastases (11%–20%; ref. 60). RT is commonly used for symptom relief, CSF flow correction, and debulking during preparation for chemotherapy (61). Focal RT may be indicated for clinically symptomatic sites and bulky disease. Yan and colleagues analyzed 51 EGFR-mutated patients with NSCLC with LM and reported that WBRT did not improve the objective response rate (ORR) or intracranial disease control rates (DCR; ref. 62). Li and colleagues analyzed 184 patients with LM secondary to NSCLC and reported OS benefits in patients who received TKI therapy (10 months vs. 3.3 months, P < 0.001), but 42 patients who received WBRT did not have a longer OS than those who did not receive WBRT, and a combination of WBRT and TKIs had no additional survival benefit (7). A retrospective study reported that WBRT was associated with better prognosis in patients with NSCLC with LM (13). A systematic review of eight studies involving 389 patients with NSCLC and LM found no conclusive evidence that WBRT prolonged survival (63).

There is no universally accepted plan for palliative WBRT dose fractionation, although common variations include 20 Gy/5 or 30 Gy/10 administered daily, 5 days a week (64, 65). In certain institutions, lengthier schedules (such as 37.5 Gy/15 and 40 Gy/20) may occasionally be employed (64). Acute adverse effects related to WBRT include cerebral edema, nausea, vomiting, headache, anorexia, seizures, dermatitis, and malaise, whereas late AEs include chronic fatigue, neurocognitive impairment, cerebrovascular damage, and pituitary malfunction. These side effects might be limiting factors to the utility of the treatment in practice. Ongoing clinical trials are investigating the role of WBRT, including a phase one trial for WBRT combined with avelumab in patients with solid tumors (NCT03719768), and a phase III trial for WBRT combined with veliparib among NSCLC (NCT01657799).

There are also limited data available on the role of craniospinal irradiation (CSI) in LM. A retrospective study involving 25 patients with LM who received craniospinal radiation reported stabilization of neurological symptoms in 40% of the cohort, but the grade 3 myelosuppression rate was 32% (66). A recent phase-1 clinical trial involving 24 patients treated with proton CSI in LM secondary to solid tumors reported a median CNS PFS of 7 months and OS of 8 months (67). In this study, 2 of 24 patients developed Grade 4 thrombocytopenia and lymphopenia and/or grade 3 fatigue (67). An ongoing phase I trial (NCT03520504) is investigating proton radiation to the brain and spinal cord among patients with LM from solid tumors.

For patients with LM from NSCLC with systemic metastasis who do not have actionable mutations, systemic chemotherapy is the treatment of choice because it has been shown to be an independent predictor of survival (9, 51). There is no consensus regarding the standard of chemotherapy, and promising results with agents such as bevacizumab and pemetrexed have been reported (68, 69). Targeted therapies are discussed in a separate section.

A pooled analysis showed that intrathecal chemotherapy is an effective treatment for individuals with LM from NSCLC (69); however, the best drug, dose, and regimen have yet to be determined. Methotrexate, cytarabine, and thiotepa are the most regularly utilized intrathecal chemotherapeutic drugs (5, 51), but none of the intrathecal chemotherapy (IT) regimens have been shown to be superior (5). In a phase I study of 13 patients with LM and NSCLC who received weekly intrathecal pemetrexed in addition to twice-weekly systemic pemetrexed, the ORR was 31% and 54%, respectively (70). A phase II clinical trial involving 30 patients with LM secondary to EGFR-mutated NSCLC who progressed on TKI therapy treated with IT pemetrexed combined with dexamethasone showed an 84.6% of clinical response rate (22/26), and median OS was 9 months (71, 72). Most frequent reported adverse effect was myelosuppression (30%) and recommended pemetrexed dose was 50 mg (71). A randomized clinical trial involving 34 patients with LM secondary to solid tumors (21/34 NSCLC) investigated the role of IT pemetrexed (10 mg) followed by involved-field RT for 3 days and showed a clinical response rate of 68%, median OS of 5.5 months; however, 53% of patients developed adverse events (myelosuppression, liver injury, radiculitis) including 6 patients with Grade 3, and 1 patient with Grade 4 adverse events (AE; ref. 73).

IT chemotherapy can be administered via injection during a LP or by placing an Ommaya catheter, which is a small implantable device placed under the scalp that drains into the lateral ventricle through a tube. Compared with LP, Ommaya catheter offers multiple benefits, including less painful administration of IT and more homogenous drug delivery; however, Ommaya catheter has its own drawbacks, including the necessity of surgery for placement of the catheter, complications such as infections, intracranial bleeding, and symptomatic leukoencephalopathy (74). One study showed an OS benefit with intraventricular chemotherapy via the Ommaya catheter compared with LP (9.2 months vs. 4 months, P = 0.0006; ref. 75). Glantz and colleagues (76) conducted a randomized controlled trial and found that intraventricular methotrexate was associated with a higher rate of progression-free survival than intrathecal methotrexate (43 days vs. 19 days, P = 0.048).

In select patients with NSCLC with CNS involvement, including LM, systemic application of molecularly targeted treatments has shown clinical benefits (77). Multiple successful studies have led to the approval of molecularly targeted medicines with increased CNS penetration (61). Current strategies for targeted therapies are as follows:

LM occurs in approximately 9% of individuals with NSCLC with EFGR mutations (78). A single-center retrospective analysis including 136 patients with LM from EGFR-mutated NSCLC reported longer OS in patients treated with TKIs compared with those who did not receive TKIs (10.0 m vs. 3.3 m, P < 0.001). The analysis also demonstrated that WBRT did not provide further OS benefit (79). Selected studies regarding EGFR TKI treatment for patients with EGFR mutated NSCLC listed in Table 1.

Multiple case reports have described clinical and radiologic improvement with gefitinib therapy, a first-generation EGFR inhibitor, at standard doses (80, 81) or high doses (82) among patients with LM from lung adenocarcinoma. A retrospective study including 35 patients with LM from EGFR-mutated NSCLC who experienced disease progression after failure of standard-dose EGFR-TKIs showed that high-dose erlotinib (various dosages and regimens of high-dose erlotinib were used: 200 mg on alternate days, 300 mg on alternate days, 300 mg every 3 days, 450 mg every 3 days, and 600 mg every 4 days) showed a radiologic response in 30% of patients, and symptomatic improvement in neurologic symptoms in 50% of patients. The median survival time from the diagnosis of LM in patients treated with high-dose erlotinib and those not treated with erlotinib was not statistically different (6.2. months in the erlotinib arm vs. 5.9 in the control arm, P = 0.94; ref. 80). A retrospective analysis involving 22 patients with LM from EGFR-mutant NSCLC showed that OS was longer in erlotinib-treated patients than in gefitinib-treated patients (6.6 months vs. 2.1 months, P = 0.07; ref. 83). Nanjo and colleagues reported that gefitinib resistance of PC-9/LMC-GR cells was associated with MET copy-number growth with MET activation, despite EGFR-T790M being negative in the acquired gefitinib-resistant mouse model (84). Furthermore, combined use of EGFR TKI with crizotinib, which has MET activity, led to a regression in LM with an acquired EGFR-TKI–resistant mouse model (84). In certain patients with LM related to NSCLC and prior TKI failure and overamplification of MEK, Cheng and colleagues proposed using a MEK inhibitor with a MEK inhibitor in addition to an EGFR-TKI (58).

Afatinib is a second-generation EGFR TKI, has been shown to be effective in patients with LM caused by EGFR-mutant NSCLC (85, 86), including at least one report of a patient with LM responding to afatinib following disease progression with osimertinib (87).

Osimertinib is a third-generation EGFR TKI with considerable intracranial action. A multicenter phase I trial study (BLOOM; NCT02228369) involving 41 patients with LM from EGFR-mutated NSCLC who had disease progression on prior EGFR-TKI therapy showed an ORR of 62%, PFS of 8.6 months, and median OS of 11.0 months, and reported CSF clearance in 11/40 patients (88) with osimertinib 160 mg daily. The AURA-LM analysis examined the clinical efficacy of osimertinib 80 mg daily as a second-line treatment for EGFR T790M-NSCLC patients and demonstrated an ORR of 55%, CR of 27%, median PFS of 11.1, and median OS of 18.8 months (89). The phase III FLAURA trial showed overall prolonged OS with osimertinib compared with first-generation TKI in patients with previously untreated advanced EGFR-mutated NSCLC (38.6 m vs. 31.8 m, P = 0.0462; ref. 89). Four of the five patients in the osimertinib arm who had baseline radiologic evidence of LM obtained a full radiologic response, compared with one of the 2 patients with suspected LM in the comparator arm who exhibited a full radiologic response (89). Zheng and colleagues reported that median iPFS was significantly longer in patients with T790M-positive CSF genotyping (15.6 months) than in T790M-negative CSF (7.0 months; P = 0.04; refs. 46, 90).

AZD3759 is a novel, potent, oral EGFR TKI with an impressive CNS penetration (91). Yang and colleagues reported that free concentration of drug is equal in blood, CSF, or brain tissue and showed antitumor activity (92). A phase one clinical trial exhibited that AZD3759 at 200 mg twice daily showed a tolerable safety profile and good CNS penetration in EGFR-mutated NSCLC with brain or LM without any EGFR-TKI exposure or patients with LM who are pretreated with EGFR-TKIs (93). Grade 3 AE at the dose of 200 mg twice daily were skin and gastrointestinal (17%), hepatobiliary and renal (13%), and nutrition disorders (4%). No Grade 4 AE was reported.

In individuals with NSCLC with ALK arrangements, LM develops in approximately 10% of cases (6). At the time of diagnosis, up to 40% of individuals with ALK-positive lung cancer were found to have CNS metastasis (94). The ALK inhibitors crizotinib, ceritinib, alectinib, brigatinib, and lorlatinib have been approved by the FDA for the treatment of patients with NSCLC with ALK rearrangement. Crizotinib is a first-generation ALK inhibitor with limited CNS penetration (95–97).

Second-generation ALK inhibitors include alectinib, brigatinib, and ceritinib, all of which have important clinical activities in patients with brain metastases, including those with LM. In the phase III ALEX trial (98), alectinib was compared with crizotinib among patients with treatment-naive ALK-positive NSCLC and showed prolonged investigator-assessed median PFS with alectinib (34.8 months vs. 10.9 months). Although OS data remain immature, the alectinib arm showed a higher 5-year OS rate compared with crizotinib (62.5% vs. 45.5%), and an OS benefit was observed in patients with or without brain metastasis. A case report demonstrated a sustained clinical and radiologic response to alectinib (>15 months) in a patient who developed symptomatic leptomeningeal disease while on crizotinib (99). A case series of four patients with ALK-positive NSCLC and LM who had progressed on crizotinib and ceritinib demonstrated radiographic and clinical improvement with alectinib (100). A few LM cases have demonstrated that switching from one ALK inhibitor to another or dose escalation can be effective in overcoming crizotinib resistance (101–103). Zou and colleagues conducted a retrospective study involving 65 patients with ALK-rearrangement NSCLC with brain metastasis or LM reported that need of mannitol or steroid decreased significantly in patients treated with alectinib, and 8/9 patients who had LM with or without BM reported improvement in CNS-related symptoms (104).

In early trials, brigatinib, a newer second-generation ALK inhibitor, demonstrated a reduction in brain metastases in patients previously treated with crizotinib and ALK inhibitor-naïve patients (101, 105). CNS PFS exceeded 14 months with modestly improved outcomes at 180 mg/day over 90 mg/day.

Lorlatinib is a third-generation ALK inhibitor with high CNS penetration. In an ongoing phase II study (NCT01970865) involving 197 patients with ALK-positive lung cancer whose disease progressed on prior ALK-TKI therapy, lorlatinib showed substantial intracranial activity in patients with ALK-positive NSCLC whose disease progressed on first- or second-generation ALK TKIs, regardless of baseline CNS involvement (106). Lorlatinib was active in two individuals with LM in this study, one of whom had a full and continued response at 22 months (106). In an interim analysis of the CROWN clinical trial (NCT03052608) involving 296 patients with previously untreated advanced ALK-positive lung cancer, patients who received lorlatinib showed longer PFS at 12 months (78% vs. 39%) and a higher frequency of intracranial response (83% vs. 23%) than those who received crizotinib (101, 107). A recent phase II clinical trial investigating the efficacy of Lorlatinib in patients with LM secondary to ALK-rearrangement NSCLC presented with CNS-only progression on the second-generation reported control of intracranial disease control in 12 weeks in 95% of patients, the intracranial objective response rate of 59%, and median PFS of 24.6 months (108). Patients who progressed only in CNC on the prior first- and second-generation of ALK inhibitor have a better response to Lorlatinib compared with patients with CNS and concomitant extracranial progression (108). Selected studies regarding ALK inhibitor treatment for patients with ALK rearranged NSCLC listed in Table 2.

Other oncogenes, such as ErbB2, KRAS, BRAF, PI3K, RET, PDGFR, ROS, MET exon 14 (METex14) skipping mutations, and MEK1 and HER2 have been found to be mutated, translocated, or amplified in NSCLCs (109), and might respond to targeted therapy. A case report showed that LOXO-292 demonstrated efficacy in a patient with LM from NSCLC with a RET mutation (110). Vemurafenib is presently being used in patients with lung cancer with BRAFV600E mutations, with one case report indicating that the drug provided 6 months of clinical and radiologic control in a patient with LM (111). Another case report of a patient with a V600E mutation who had progressed on dabrafenib/trametinib reported a quick response in the brain and leptomeninges to encorafenib and binimetinib (112). However, randomized trials focusing on LM in NSCLC with these rare mutations are needed.

Immune checkpoint inhibitors (ICI) have changed the current landscape of NSCLC management. There are several studies showing the promising effect of ICI in LM from NSCLC (113, 114). Hendriks and colleagues conducted a prospective study including 19 patients with LM from NSCLC who were treated with ICIs (13 with nivolumab and 6 with pembrolizumab), and showed a median PFS of 3.7 months, and a 6- and 12-months OS of 36.8% and 21.1%, respectively (113). A single-arm phase II clinical trial investigating the efficacy of pembrolizumab for LM in solid tumors (NCT02886585) reached its primary endpoint, and 60% of patients were alive 3 months after enrollment (115). A phase II clinical trial (116) that investigated the activity of pembrolizumab among patients with LM from solid tumors, including 23% of participants with NSCLC, showed an overall CNS response rate of 38% in 12 weeks, with an acceptable safety profile. Another phase II clinical trial involving 18 patients with LM secondary to solid tumors receiving combined ipilimumab and nivolumab showed promising activity and reported that 8 of 18 patients reached the primary endpoint of 3 months survival whereas one third of patients experienced grade 3 or 4 AE (117). Zheng and colleagues conducted a study including 32 patients with LM secondary to NSCLC who received ICI (nivolumab 21/32, pembrolizumab 9/32, atezolizumab 2/32) and reported that 62.5% of patients had neurologic symptom improvement, median PFS was 2 months, and OS was 4 months in patients who received single-agent ICI; median PFS and OS in patients who received other therapies with ICI was 3 and 5.4 months, respectively. In this study, all patients who received ICI as a single agent had cranial radiotherapy prior and received ICI as second-line therapy (118). Zheng and colleagues reported that patients with better Eastern Cooperative Oncology Group Performance Status (ECOG-PS) score had significantly longer PFS (P = 0.04), but there was no significant OS difference among patients who received single-agent ICI versus combined therapy (118). A retrospective study reported that incidence of LMD among patients who received postoperative stereotactic radiosurgery (SRS) with immunotherapy (either nivolumab or pembrolizumab) was less than among patients who received SRS alone (6% vs. 22%, P = 0.007; ref. 119). Prakadan and colleagues compiled the patients from two different clinical trials for patients with LM who received ICU and showed that ICI treatment alters the tumor microenvironment in patients with LM from any histology by conducting single-cell RNA and cell-free DNA profiling from CSF. Prakadan and colleagues reported that active immune response in CSF after intravenous ICI was correlated with OS (120). In both trials, there were increased overall levels of IFN-signaling, and cytotoxicity in CD8⁺ T cells posttreatment. These findings imply that intravenous ICI treatment modifies the immunologic milieu in the CSF of a subgroup of patients participating in these clinical trials and that this may have one of the mechanisms of the therapeutic benefit of ICI treatment (120). Another recent study used immune cell profiling of CSF using single-cell RNA sequencing in patients with brain metastasis showed that matching T-cell receptor clonotypes of CD8 T cells in CSF and brain tumor, which is very promising to use immune profiling of CSF developing cell-therapies for brain metastasis or exploring tumor microenvironment (121). The studies using immunotherapy in LM secondary to NSCLC demonstrated in Table 3.

Supportive care is essential for LM management. Pain management, steroids, treatment of seizures, ventricular peritoneal (VP) shunt placement for relief of symptomatic hydrocephalus, and elevated ICP are all supportive therapies in patients with LM. In LM, there are no robust data to support the use of antiepileptics as seizure prophylaxis. There is also no agreement on systemic steroid treatment in LM other than at the lowest dose for the shortest time feasible (36, 41, 122). Peritoneal carcinomatosis reported as very rare complication of VP shunt placement in literature, although incidence remains unknown. A retrospective study of patients with LM from solid tumors (123) showed that patients who underwent LP or VP shunt placement reported a 50% improvement in symptoms, with 34% of total symptom relief after the procedure. Shunt-related complications included infection, shunt malfunction, and need for shunt revision. Another study involving 31 patients with hydrocephalus secondary to LM from lung adenocarcinoma who subsequently underwent palliative shunt placement reported symptom alleviation in 90.3% of patients, with a median survival of 7.7 months following LM diagnosis (124). Another retrospective study involving 190 patients with LM secondary to solid tumors reported that 83% of patients had symptomatic relief, and complications included infection (5%), shunt repair/externalization/removal (8%), and symptomatic subdural hygroma/hematoma (6.3%) whereas no abdominal seeding was seen (125). In day-to-day practice, the benefits of VP shunting in patients with increased intracranial pressure outweigh the theoretical risk of abdominal seeding (125).

As discussed above, the landscape of treatments for LM from NSCLC is quickly changing with the new advancement in diagnostic tools and treatment options. There are several prospective clinical studies currently underway to investigate new diagnostic procedures, monitoring measures, and therapy options for LM from NSCLC. Ongoing clinical trials of LM are presented in Table 4.

LM remains a fatal complication of advanced cancer, and its incidence is increasing, given the success of advanced cancer therapies. Drug penetration is limited to the leptomeningeal region, given the blood–brain barrier. Initial diagnostic workup should include CSF cytology, and neuraxial imaging with MRI. Given emerging novel diagnostics such as CSF liquid biopsy and the promise for early diagnosis and can offer valuable information to guide the therapy. The treatment of LM secondary to NSCLC needs to tailor according to an individual patient and requires multimodal treatment. The development of novel therapeutic agents with better CNS penetration, especially with. Patients with actionable mutations are candidates for targeted therapy. The third-generation EGFR inhibitor high-dose osimertinib (160 mg daily) recommended LM from EGFR-mutated NSCLC and has shown a survival advantage compared with standard therapies, such as RT, chemotherapy, and first- and second-generation TKIs. Patients with ALK-rearrangement should be treated with ALK inhibitors. Radiotherapy offers palliative relief in symptomatic patients with bulky disease. Although there is no consensus, WBRT did not show an OS benefit. Ongoing trials are investigating the role of WBRT in LM. Systemic or IT chemotherapy can be options for patients with no actionable mutation with systemic metastasis or progressive disease. There are several studies that showed promising results with a combined or single-agent immunotherapy. Clinical trials in LM secondary to NSCLC focus on monitoring clinical and radiologic responses, expanding data on other targetable mutations, and investigating response and resistance indicators are urgently needed.

No author disclosures were reported.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.