There are many ways by which malignant cells can spread to the leptomeninges: (1) hematogenous spread , mostly by arterial contamination or via the venous plexus of Batson in pelvic cancers; (2) direct dissemination from adjacent BM or primary brain tumors; (3) through centripetal migration, along perineural or perivascular spaces. The latter mechanism is mostly observed in head and neck cancers, and in breast and lung cancers, when there are known vertebral and paravertebral metastases [42–44].

Once circulating tumor cells (CTC) have breached the subarachnoid space (SA) , CTC spread through CSF, resulting in diffuse and multifocal seeding of the leptomeninges. The base of brain (especially the basilar cisterns) and dorsal surface of the spinal cord (especially the cauda equina) are predominantly affected [45, 46]. The CSF outflow may be obstructed by ependymal nodules or tumor deposits leading to a communicating hydrocephalus. Obstruction of CSF corridors affects predominantly the IVth ventricle, basal cisterns, cerebral convexity, and spinal SA.

It is important to diagnose LM in its initial stage, when neurological impairment is minimal, the patient performance status (PS) is maintained, and the tumor burden is low as these factors are correlated with response to treatment and OS.

In summary, patients with suspected LM should undergo, prior to any anticancer treatment, one or two lumbar punctures to establish the presence of CTC, cranial MR-Gd, spinal MR-Gd and a radioisotope CSF flow study (the later only in instances where intra-CSF chemotherapy is to be administered). If cytology and MR are not contributive, a ventricular or lateral cervical CSF analysis may be required. If CSF cytology remains persistently negative despite compelling clinical and radiological manifestations consistent with LM, it is appropriate to commence LM-based therapy.

In patients with cancer, the diagnosis of LM is often straightforward. The situation is more challenging in patients without any known cancer, or if causality is uncertain (for example, the primary cancer is either in remission or presumed cured) and with a negative CF cytology. Other causes of subacute and chronic meningitis must then be excluded as discussed above [99]. Eventually, if a systemic evaluation does not identify an extraneural cancer, a primary brain tumor should be considered [100–104].

New neurological manifestations must be distinguished from those secondary to parenchymal disease, from complications of intra-CSF treatment, systemic therapies or RT, from concomitant medications, from neurological or extraneurological concurrent disease, and more rarely from paraneoplastic syndromes [82]. Transient manifestations should not be assumed to be LM-related neurological progression. As mentioned earlier, CSF cytological analysis remains the gold standard for the diagnosis of cancer cells in the CSF despite a low sensitivity of a first CSF examination (45–55%), and the usual need of successive CSF samples to adequately assess cytology. The one-dimensional response evaluation criteria in solid tumors (RECIST) criteria are not adapted for the assessment of LM as the imaging characteristics of LM in general are not measurable at least as defined by current brain tumor response criteria [105].

In order to improve the methodology of future clinical trials, a RANO working group critically reviewed published the literature regarding randomized clinical trials (RCT) and trial design in patients with LM [106]. Six RCT have been published regarding the treatment of LM secondary to solid tumors. As seen in Table 5.4, there were significant differences in the response assessment, such that interpretation of the trial results is challenging. Among these RCT, only a single trial attempted to determine whether intra-CSF chemotherapy was superior to systemic therapy only. Otherwise, this fundamental question has not been prospectively answered in patients with solid cancers and LM. The methodology of the 6 RCT varied considerably regarding pretreatment evaluation, type of treatment, and response to treatment. The RCT all share several common initial assessment features including evaluation of CSF cytology. Positive CSF cytology has been required for participation in all published RCT. Additionally, a neurologic clinical assessment and a predefined performance status were required as inclusion criteria in all RCT. Nevertheless, currently in neuro-oncology there is no standard instrument to assess the neurologic examination and consequently RCT in LM have lacked a rigorous method to determine clinical disease progression [106]. Also poorly defined in the published RCT is the utility of neuroimaging assessment (e.g., how radiographic findings alter treatment and how neuroimaging is used in the response assessment independent of CSF cytology and clinical examination) [106]. The RANO LM working group has developed a series of instruments to address these deficiencies [97] that includes a standardized neurological examination, definitions of CSF cytology and a new radiographic evaluative instrument. The working group recommends that “all patients enrolling in clinical trials undergo CSF analysis (cytology in all cancers; flow cytometry in hematologic cancers), complete contrast-enhanced neuraxis MRI and radioisotope CSF flow studies (in patients treated with intra-CSF therapy only)” [97]. “Response based on CSF cytology is considered when CSF converts from positive to negative and with a second confirmatory determination. CSF cytology is not to be considered in isolation in evaluation of response of patients with solid tumors as patients with persistence of positive cytology may continue on treatment if clinically and radiographically stable or improved” [97]. The committee created a composite score to quantify MRI abnormalities recognizing that most LM lesions are non-measurable. They concluded that “radiographic assessment of LM is subjective, qualitative and graded as stable, progressive or improved. Similar to CSF cytology, radiographic disease progression in isolation, i.e., stable CSF cytology and neurological assessment, would be defined as LM disease progression” [97]. It is hoped that ultimately this work will provide a standardized approach for LM assessment with validated criteria for response to treatment, and defined endpoints for clinical trials.

The median OS of untreated patients with LM is 4–6 weeks [3, 14, 48, 94, 107–113]. Despite aggressive treatment, LM has a poor prognosis. The survival of patients with combined treatment is usually less than 6 months with a median OS of 2–3 months [3, 10, 13, 14, 22, 48, 54, 56, 89, 90, 109–113]. Table 5.5 illustrates reported survival in patients with LM from the recent literature.

The goal of LM-directed treatment is to improve or stabilize neurologic status, maintain neurological quality of life, and prolong survival. Nevertheless, deciding which patients to treat remains a challenge. The NCCN CNS guidelines (version 1.2012) have attempted to distinguish between patients reasonably considered for treatment versus those patients in whom supportive care is most appropriate [49, 50, 94, 107, 114, 115] (Table 5.6).

Based on the literature, the histology of the primary cancer is known to be the prominent prognostic factor with regard to OS in LM [50, 116]. Multivariate analysis has confirmed the association between survival and primary tumor type and the better prognosis of BC compared with lung cancer or melanoma-related LM [10, 68, 117]. Breast cancer LM has a relatively good prognosis among all solid tumor-related LM, with a median OS of 3.3–5 months [22, 54–56, 109]. Modest improvement in lung cancer-related LM may in part reflect increasing use of targeted agents such as tyrosine kinase inhibitors (TKI) based on recent reports of patients with lung cancer and LM with reported median survival of 3–6 months [13, 89, 118–120]. LM secondary to melanoma continues to demonstrate the worst prognosis of all solid tumor-related LM with a median OS estimated between 10 weeks and 4 months before the era of the immunotherapy and targeted therapies [107–110].

In addition to tumor histology, multivariate analysis in series including all types of solid tumors confirms the associations between OS and the performance status (PS), the age at LM diagnosis, the presence of a LM-related encephalopathy, time from diagnosis of primary tumor to diagnosis of LC (>67 weeks), the CSF protein level, the treatment modality (administration of systemic therapy), and the presence of a short-term response to treatment [10, 68, 117, 121–124]. In one report, gender was significant in multivariate analysis likely due to an interaction between gender and tumor type [10]. Recently, the quantification of CSF EpCAM+ cells obtained with flow cytometry immunophenotyping (FCI) before any treatment has been reported to be an independent prognostic factor for OS in LM patients with a cutoff value of 8% [74].

In patients with BC and LM, multivariate analysis demonstrates an association between OS and PS as well as PS and treatment (number of prior chemotherapy regimens, receipt of combined treatment modality, coadministration of systemic chemotherapy or intra-CSF chemotherapy) [11, 22, 54, 55, 109, 122, 125]. Histological characteristics (histological grade, hormone receptor status, triple-negative status), the initial CSF protein level, the number of prior chemotherapy regimens, status of systemic disease (i.e., isolated CNS metastases or bone metastases), the presence of an early and/or combined cytological and neurological response to treatment and the initial CSF cyfra 21-1 and protein levels were also identified as significantly associated with prognosis in BC-related LM [11, 22, 54–56, 109, 122–125].

In recent retrospective series of patients with NSCLC and LM, multivariate analysis confirmed that PS, low initial CSF protein level, low initial CSF (white blood cells) WBC count were all significantly associated with a better OS [13, 89, 118]. The treatment modality (systemic therapy and intrathecal chemotherapy) and clinical improvement after intra-CSF chemotherapy were all significant predictors of increased OS [13, 89, 118, 120]. The impact of whole brain radiotherapy (WBRT) in the treatment of LM is unclear [13, 89, 118–120]. In EGFR-mutated NSCLC, EGFR inhibitors have produced durable responses in patients with LM [13, 89, 113, 118, 119, 126–130]. In patients with melanoma, multivariate analysis suggests that a history of a primary melanoma lesion originating on the trunk predicted shorter OS and that intra-CSF chemotherapy predicted longer OS [110]. Nevertheless, unlike prognostic factors present at LM diagnosis and before any treatment, the interpretation of prognostic factors related to treatment and mostly derived from retrospective series remains uncertain. All such series are single-institution retrospective studies in which treated patients are heterogeneous with respect to recognized LM-prognostic features. Because of the retrospective nature of these studies, there were no a priori determinants of treatment such that treatment was defined individually, without apparent standardization making cross-comparisons between treatment groups difficult [131]. Moreover, two series, one in NSCLC and the other in all types of solid tumor LM, report that the placement of a VPS was significantly predictive of improved OS in LM [118, 132]. These observations seem counter-intuitive and are difficult to reconcile.

The goals of treatment include palliating neurologic symptoms and whenever possible stabilizing or improving patient neurologic function as well as prolonging survival. Since the prognosis of LM varies noticeably depending upon the primary tumor type and extent of both neurologic and systemic disease, parameters separating poor-risk from good-risk patients are helpful to determine the appropriate therapeutic approach for an individual patient. The poor-risk and good-risk patients categories are illustrated in Table 5.6. LM ideally should be diagnosed early in the disease course before the appearance of fixed and disabling neurological deficits. Early LM-directed treatment may allow maintenance of quality of life and potentially improve survival. A combined treatment approach (i.e., systemic and intra-CSF chemotherapy and site specific radiotherapy) may provide better palliation in patients with LM [133].

The main surgical interventions in LM are ventriculoperitoneal shunting (VPS) for symptomatic hydrocephalus, placement of a ventricular (rarely lumbar) access device (e.g., an Ommaya or Rickham reservoir) to facilitate administration of intra-CSF chemotherapy, and meningeal biopsy. When both a VPS and Ommaya ventricular access device are needed, an on–off valve may be an option but the patient should be able to tolerate having the VPS placed in the off position so as to permit drug installation into the ventricles and time for ventricular transit and distribution into the nonventricular CSF compartments [134–136]. Complications of VPS rarely exceed 10% in recent series [135, 136] but include the potential for peritoneal tumor dissemination, device failure, device malpositioning, device exposure, infection, hemorrhage, and leukoencephalopathy [45, 135, 136].

When a ventricular access device is placed, confirmation post-implantation of correct intraventricular (IVent) placement requires a brain CT or alternatively a radioisotope CSF flow study before intra-CSF drug administration [137–139]. Hemorrhage at the time of device placement occurs in less than 1% of patients. Device infection (4–10% incidence) is due mainly to Staphylococcus epidermidis and is either a complication that occurs in relationship to surgery or results from contamination at the time of device access [133, 134, 138–140]. In instances where the ventricular device becomes infected occasionally, the device may be left in situ and treated with both intravenous and IVent antibiotics [93, 140–144]. Most often, however, device infections requires removal and if indicated, replacement of the reservoir [93, 135, 142]. A rare complication in patients with increased intracranial pressure is retrograde tracking of instilled chemotherapy along the catheter, resulting in subgaleal or intraparenchymal collections of CSF that may become symptomatic and require revision or replacement with a ventriculoperitoneal shunt [5].

Craniospinal axis irradiation (CSI) is the only method of radiotherapy that treats the entire neuraxis and that may be reasonably considered as a single modality of treatment for LM. However, in the majority of adults CSI is rarely considered as most patients have previously had some region of the neuraxis irradiated, and as well have poor bone marrow reserve as a consequence of prior exposure to cytotoxic chemotherapy. Consequently, CSI and treatment-associated toxicities of myelosuppression and enteritis is deemed too toxic for routine use in adults with solid tumor-related LM. The role of alternative methods of CSI such as tomotherapy and proton radiotherapy , which could permit improved precision in radiation dosing and targeted volumes and consequently less hematological toxicity, has not been formally evaluated and may be an option in the future.

The majority of patients with LM receive involved-field radiotherapy to sites of symptomatic disease, bulky disease observed on MRI and to sites of CSF flow block defined by radioisotope ventriculography . Irradiation permits tumor masses not treated by intra-CSF chemotherapy (due to limited diffusion of intra-CSF chemotherapy) to receive palliative radiotherapy [145]. WBRT is most commonly administered at a dose of 30 Gy delivered in 10 fractions over 10 days. While pain relief and stabilization of neurological manifestations are regularly observed, unfortunately significant neurological improvement is exceptional. The failure of neurological improvement is due to demyelination, axonal and neuronal injury, and injury by infiltrating cancer cells, aspects that underscore the need for early treatment of LM [146]. Regardless of MRI findings (e.g., the absence of visible radiographic disease), lumbosacral irradiation is indicated in instances of cauda equine syndrome (low back pain, legs weakness, bladder or bowel dysfunction). Similarly, skull-base RT may be used for cranial neuropathies [8]. RT is also indicated to reestablish normal CSF circulation to improve therapeutic efficacy and reduce side effects of intra-CSF chemotherapy [1–6, 8, 94]. Communicating hydrocephalus is not infrequent in LM and is caused by malignant cells in the subarachnoid space that obstruct normal CSF resorption pathways. In these instances, WBRT or placement of a VPS is often required [147]. Shunting of CSF should be provided in patients with symptomatic or communicating hydrocephalus that does not rapidly respond to WBRT.

Unlike brain metastases, the impact of WBRT on OS is not clearly established in LM, even in radiosensitive cancers such as BC and NSCLC. Contradictory results have been reported that in part reflect the limited survival of patients with LM (<15% survive 1 year) [13, 89, 118–120]. In small case reports WBRT has been reported (as part of a combined modality treatment) to result in prolonged survival (from 13 to 19 months) in patients with melanoma and LM [148–152]. In these case reports, WBRT was either delivered concomitantly or prior to other treatment (intra-CSF or systemic). However, the retrospective nature of the data makes it challenging to determine the role of WBRT in the treatment of LM.

Some authors have posited that WBRT may enhance the efficacy of targeted drug therapy by either by radio sensitization or improving drug penetration into brain/CSF [148, 149]. Improved brain penetration following WBRT has been demonstrated with systemically administered trastuzumab, a monoclonal antibody a low rate of CNS penetration rate [153–155]. In addition to mechanically disrupting the BBB, radiation may hinder the efflux function of multidrug resistance proteins [149]. Indeed, the low CSF concentration of vemurafenib has been related at least partially to the active efflux by P-glycoprotein and breast cancer resistance protein [156].

Side effects of involved-field RT alone are uncommon and manifest mainly as radiation-associated fatigue. However, myelosuppression, mucositis, esophagitis, and leukoencephalopathy have been reported with more extensive radiation fields. Leukoencephalopathy (asymptomatic more often than symptomatic) may be a delayed consequence of WBRT and may be enhanced with coadministered methotrexate (MTX) (either systemic or intra-CSF). Ongoing clinical trials evaluating the safety of concomitant WBRT and intra-CSF liposomal cytarabine will determine if this is a common problem with chemoradiation or unique to MTX when combined with radiotherapy.