Radiotherapy for Leptomeningeal Carcinomatosis in Breast Cancer Patients: A Narrative Review

Simple Summary Leptomeningeal carcinomatosis (LC) is a rare event in breast cancer (BC) patients that carries an abysmal prognosis. Little progress has been made in this field in the last few decades. Despite innovations in radiotherapy (RT), there is no univocal evidence of its impact on survival. Due to the rarity of the diagnosis, only a few prospective trials have evaluated the role of RT for LC in BC. Nonetheless, most BC patients with LC currently receive RT, depending on local protocols and individual convictions. This review presents the current knowledge on the indications and feasibility of RT for LC in BC, focusing on new technologies and perspectives. Abstract Leptomeningeal carcinomatosis (LC), defined as the infiltration of the leptomeninges by cancer cells, is a rare oncological event with the most common etiology being breast cancer (BC), lung cancer, and melanoma. Despite innovations in radiotherapy (RT), firm evidence of its impact on survival is lacking, and concerns are related to its possible neurotoxicity. Owing to a paucity of data, the optimal treatment strategy for LC remains unknown. This review discusses current approaches, indications, and contraindications for various forms of RT for LC in BC. A separate section is dedicated to new RT techniques, such as proton therapy. We also summarize ongoing clinical trials evaluating the role of RT in patients with LC.


Introduction
Leptomeningeal carcinomatosis (LC), defined as an infiltration of the leptomeninges by cancer cells, is a rare event in solid tumors, with the most common etiology being breast cancer (BC), lung cancer, and melanoma [1]. The reported incidence of LC in BC patients ranges widely from 0.8% to 6.6% in clinical reports and from 2.6% to 16% in autopsy series [2][3][4][5][6][7][8][9]. However, these statistics can be biased by analyzing cohorts of high-risk patients. Indeed, in an unselected cohort of 1915 BC patients, the 5-year incidence of LC was merely 0.3% [10]. The coexistence of LC with brain metastases (BM) has been reported in 4-14% of cases, but the actual rates remain unknown [11][12][13]. The incidence of LC in BC is higher in younger patients and those with a larger primary tumor, advanced nodal disease, histological grade 3, negative estrogen receptor (ER), positive human epidermal growth factor receptor 2 (HER2), triple-negative type, and a high proliferative index [14]. Lobular carcinoma is particularly associated with LC and accounts for about 35% of BC cases [14]. This is in contrast with only about a 7% incidence of lobular carcinoma among parenchymal BM, suggesting an affinity of this subtype for leptomeningeal dissemination.
Previous surgery for BM, especially for supratentorial tumors, also increases the risk of LC [15,16]. A meta-analysis showed that BM resection followed by stereotactic radiosurgery (SRS) carries a higher risk of developing LC than resection followed by whole-brain radiotherapy (WBRT) [17]. Another suggested risk factor for developing LC is SRS used alone [18][19][20]. However, in a large retrospective study including nearly Two prospective trials evaluating the role of combined intrathecal chemotherapy (ITC) and RT showed the OS benefit of RT [36,37]. However, both included patients with LC from various solid tumors, with only a small subset of BC patients. In the phase 2 study, patients with at least one adverse prognostic factor (Karnofsky performance status [KPS] of < 60%, severe and multiple neurological deficits, encephalopathy, extensive systemic disease with few treatment options, and bulky BM) received MTX ITC concurrently with RT [37]. The concomitant treatment was well tolerated, with no major toxicities or side effects related to RT. Mild or moderate skin reactions and hair loss occurred in all patients undergoing brain RT, and 22% experienced mild and moderate otitis media. Moderate and severe toxicity occurred in 20% of cases, which seems acceptable, considering the expected OS benefit. In the second trial, comparing single and combination ITC, RT to the brain, spine, or whole craniospinal axis was administered in 50% of cases [36]. Concurrent RT significantly improved the response rate and OS; however, allocation to RT was not randomized.
As several studies have shown that CSF flow interruption is associated with decreased survival, RT remains the treatment of choice to remove the flow obstruction, reduce toxicity, and enhance the efficacy of ITC [38][39][40][41].

Whole-Brain Radiotherapy
WBRT is still the most widely used RT technique in LC treatment [40]. However, assessment of its impact on survival as a single modality is difficult, as in most cases, it is combined with systemic or intrathecal ChT (Table 2). In a retrospective analysis investigating the efficacy of WBRT as a single modality, conventionally fractionated RT was performed via parallel opposed fields [43]. The planning target volume encompassed the whole brain and the meningeal space (i.e., the lamina cribrosa and basal cisterns). The toxicity of RT was low, with alopecia, nausea, headache, and fatigue being the most common side effects. There was no grade 3 or 4 toxicity. The authors concluded that WBRT is an effective palliative treatment of LC for patients unfit for ChT and with low KPS. Nevertheless, improvement of neurological deficits was reported in only 11% of patients. The safety of WBRT was also confirmed in a prospective randomized study assessing the role of ITC in LC [44]. RT did not increase neurotoxicity, even if combined with ITC. However, in a historical series, the same author described disseminated necrotizing leukoencephalopathy (DNL) in four patients with BC LC treated with WBRT, followed by low-dose ITC MTX [45]. As DNL also developed in five non-irradiated patients, the results were inconclusive.
According to the German Society of Radiation Oncology (DEGRO) guidelines, the clinical target volume in WBRT should encompass the cerebrum plus cerebellum and the brainstem down to the caudal limit of the second vertebral body [32]. Importantly, the meningeal space with the lamina cribrosa and basal cisterns should be included. Preferred dose regimens are 30 Gy/10 fx (5 fx per week) and 20 Gy/5 fx in patients with an unfavorable prognosis or 20 Gy/10 fx in patients with a predicted survival exceeding 12 months.
In a single-center retrospective study, Okada et al. showed that the dose of 30 Gy given in ≥10 fx provided significantly better OS than 30 Gy in <10 fx (median OS of 2.6 and 0.6 months, respectively); however, the patient groups were small (24 and seven, respectively) [46].

Stereotactic Body Radiation Therapy
According to the EANO-ESMO and National Comprehensive Cancer Network (NCCN) guidelines, focal RT should be considered for well-circumscribed, symptomatic lesions. It can relieve cauda equina syndrome, cranial palsies, focal pain, or obstruction of the CSF flow in 30% and 50% of patients with spinal and intracranial blocks, respectively [24,47]. Thanks to the high precision of treatment delivery and higher biologically effective dose, SRS may be a preferred treatment option for central nervous system (CNS) lesions localized near critical structures [48]. DEGRO guidelines for palliative RT in metastatic BC propose SRS at a single dose of 15-25 Gy (specified for isodose 80-90%) for lesions smaller than 3.5 cm in diameter and fractionated stereotactic RT for bigger lesions. Depending on the treatment volume, recommended fractionation schedules are 4 × 8.7 Gy, 5 × 7 Gy, 6 × 5 Gy, or 10 × 4 Gy. In the case of additional WBRT, the single fraction of 15-18 Gy (depending on the tumor size) or fractionated regimen of 6 × 5 Gy are preferred. The gross tumor volume is delineated at the MRI, and the planning target volume is created by adding an isotropic margin of 1-2 mm [31,32].
We have not identified any phase 2 or 3 randomized studies of SRS for LC in BC patients. The recommended RT regimens are extrapolated mainly from BM treatment or based on retrospective reviews, case series, and expert opinions. Most of the studies included patients of different histologies or evaluated mixed SRS/WBRT cohorts. Nevertheless, considering the potential benefits, this option seems reasonable whenever focal irradiation is indicated.
In the series by Wolf et al., out of 16 patients with LC managed with cranial SRS, five were BC patients [49]. In the entire group, five patients had received WBRT earlier.
The median margin dose delivered was 16 Gy in a single fraction of the 50-80% isodose volumes. Subsequent MRI was available for 14 patients. Five achieved disease stabilization, eight partial remissions, and one progressed. The median OS from the end of SRS for LC was 10 months, and the one-year OS was 26%. Six more patients needed subsequent WBRT due to distal progression, with a median gap of six months since SRS. The authors concluded that focal LC could be successfully performed with SRS. In some cases, SRS can eliminate or postpone WBRT with its side effects, including neurocognitive dysfunctions, alopecia, and fatigue [49]. Lekovic et al. described a case of a BC patient treated with a combination of SRS, craniospinal irradiation (CSI), and ITC with trastuzumab [50]. During the course of the disease, she received 24 Gy/3 fx for Meckel's cavity and auditory canal tumors, CSI of 30 Gy with ITC, followed by focal RT to spinal metastases (25 Gy/5 fx) and the cerebellar hemisphere (18 Gy/1 fr.). This multimodal treatment allowed for an impressive 46-month good-quality survival.

Proton Therapy
Proton therapy (PT), with its unique physics resulting in a steep dose decrease, is a very tempting option for CNS treatment, particularly for CSI. A classic photon RT results in a substantial dose delivered to the whole spinal column and anteriorly located organs, mainly the intestines and kidneys, and is rarely used [31,[51][52][53]. Proton CSI allows for less gastrointestinal and hematological exposure [54]. PT of CSI was investigated in a prospective dose-escalation phase I trial including 21 patients with LC from solid tumors, seven of whom with BC [55]. The clinical target volume included the entire brain, with proper coverage of the meninges, thecal sac, and proximal sacral nerve roots. As there was no dose-limiting toxicity (DLT) in the first six patients, all subjects received a 30 Gy relative biological effectiveness (RBE) dose in 3 Gy RBE/fx. This hypofractionated regimen replicated a popular palliative photon beam RT schedule. DLT, including grade 4 lymphopenia, grade 4 thrombocytopenia, and grade 3 fatigue, occurred in two patients from the expansion cohort and resolved without any specific treatment. The median OS was 8 months, and four patients achieved CNS disease control for longer than 12 months. The authors concluded that hypofractionated proton CSI is safe and feasible in patients with LC.
Most recently, a randomized Phase II study compared PT of CSI with standard involved-field photon RT in 63 patients with LC (36 patients with non-small cell lung cancer and 27 with breast cancer) [56]. The CNS-PFS favored PT (median 7.5 months vs. 2.3 months with photon-beam RT; p < 0.001). OS was also superior with PT (median 9.9 months vs 6.0 months for photons; p = 0.029). There were no significant differences between both therapies in the frequency of grade 3 and 4 toxicities.

Craniospinal Irradiation
Due to the presence of cancer cells in the CSF, the neuroaxis seems to be a reasonable target in LC. However, the use of photon CSI is discouraged by international guidelines due to its significant toxicity, the difficulty of RT planning, and its unconfirmed survival benefit [23,24,31,47]. So far, no trials have evaluated the feasibility of CSI exclusively in the BC population, and all knowledge is based on small case series and reviews (Table 3). Two studies reported the toxicity of CSI with 2D planning [51,52]. In the study by Hermann et al., patients were treated with CSI with (n = 10) or without (n = 9) ITC MTX [52]. Early adverse events included myelosuppression (G3 in four patients and G4 in one), dysphagia, mucositis, and nausea. There was no late toxicity. In another study, 17 symptomatic patients (six with BC) were irradiated for the entire neuraxis, with an additional WBRT dose of up to 50.4 Gy in nine patients and concomitant ITC MTX in five [52]. There was one toxic death due to an intracranial hemorrhage. Late toxicities included grade 3 infection in one patient and grade 1 myelitis in three (18%). Eleven patients received further therapies after CSI.
The excessive toxicity of CSI described in the aforementioned studies may likely be limited with modern RT techniques, such as volumetric modulated arc therapy (VMAT), helical tomotherapy (HTT), or PT. In a case report of a BC patient with LC treated with VMAT CSI, the mean bone marrow dose was 15.3 Gy, and bone marrow V20 was only 36% [59]. HTT, evaluated in three studies, was also found to be a useful therapeutic modality with acceptable toxicity [53,57,58]. However, one study reported a worrisome incidence of serious adverse events, including three toxic deaths [58].
Some authors have attempted to develop prognostic scores for decision-making. In one study, age below 55 years, KPS > 70%, and neurological response to treatment were identified as favorable prognostic factors for OS [31]. In another study, risk factors included KPS < 70% and the coexistence of an extracranial disease [57]. The median OS for patients with no, one, and two risk factors was 7.3 mo., 3.3 mo., and 1.5 mo., respectively.
A recently published review summarized 13 studies, including a total of 275 patients treated with CSI for LC of different histologies (the most common being leukemia and BC) [60]. The median CSI dose was 30 Gy, and 18% of patients received PT. Fifty-two percent of patients had improvement or stabilization of neurological symptoms. The median OS for all patients and for those managed with marrow-sparing PT was 5.3 months and 8 months, respectively. The most common treatment-related toxicities were fatigue and hematologic and gastrointestinal events. The authors concluded that CSI is a viable, yet relatively toxic option for LC. Proton CSI was discussed in an earlier section.

Radiotherapy Guidelines
The current RT guidelines for LC are summarized in Table 4. The NCCN guidelines stratify patients with LC into two categories [47]. The good-prognosis group consists of those with KPS ≥ 60%, no major neurologic deficits, minimal systemic disease, and availability for reasonable systemic treatment options. Patients with KPS < 60%, multiple, serious, major neurologic deficits, extensive systemic disease, limited treatment options, bulky CNS disease, and encephalopathy are a poor prognostic group. For the group with a good prognosis, NCCN recommends systemic ChT, ITC, or RT. Patients in the poor prognosis group may receive palliative treatment or the best supportive care. NCCN guidelines do not specify the technical aspects of RT, such as the dose or irradiated volume, which should depend on the histology and sites requiring palliation.
The EANO-ESMO guidelines recommend, in general, treatment of LC with ChT or ITC, targeted therapies, and RT, or their combination [24]. Typical target volumes for RT in patients with cranial neuropathies include the skull base, the interpeduncular cistern, and the first two cervical vertebrae. In patients with cauda equina syndrome, the irradiated volume should include the lumbosacral vertebrae. Guidelines allow for focal irradiation for cauda equina syndrome or cranial nerve palsies after excluding other causes, even without corresponding MRI findings.
In 2010, DEGRO published practical guidelines for palliative RT of BM and LC in BC patients [31,32]. In LC with spinal manifestation, the clinical target volume should encompass the gross tumor volume with a safety margin matched to the individual clinical requirements. The guidelines also mention the RT technique and dose schemes cited in the appropriate sections. LC, leptomeningeal carcinomatosis; SRS, stereotactic radiosurgery; CSI, craniospinal irradiation; WBRT, whole-brain radiotherapy; IF-RT, involved-field radiotherapy; CSF, cerebrospinal fluid.

Future Perspectives
Since LC is a manifestation of disease spread, it is unlikely that any radiotherapy developments will substantially improve the survival of this miserable entity. However, newer RT techniques, such as PT, SRS, HTT, or heavy ion irradiation, can decrease treatment toxicity. The main issue of LC systemic treatment is the poor blood-brain barrier permeability of most medications. The activity of a paclitaxel trevatide, a new experimental drug designed to have greater potential to cross this barrier, will be assessed in a phase 3 trial (NCT03613181). Recently, an increasing number of BC patients with targetable molecular alterations have been managed with targeted therapies. In advanced HER2-positive BC, the combination of trastuzumab, pertuzumab, and docetaxel is considered the standard first-line treatment [61]. A recent meta-analysis showed that IT trastuzumab is a reasonable and safe treatment for BC LC. This method resulted in CNS-PFS of 5.2 months, a median OS of 13.2 months, and a significant clinical improvement in 55% of cases [62]. In the NCT04588545 trial, patients with LC will receive this regimen intrathecally, concurrently with WBRT or focal RT. Low molecular weight HER2-tyrosine kinase inhibitors, such as lapatinib, neratinib, and tucatinib, have been shown to be effective in BC patients with BM, but there are scarce data on their activity in LC [63,64]. A phase 1 study NCT03661424 will evaluate the role of a bi-specific antibody (HER2Bi)-armed activated T-cells (HER2 BATs) in HER2 positive patients with meningeal spread. BC patients harboring hereditary BRCA1/2 mutations respond to treatment with polyadenosine diphosphate ribose polymerase inhibitors, including olaparib, veliparib, talazoparib, and iniparib, and these compounds are other potential options in LC treatment [65]. Inhibition of the cyclin D1 pathway (CDK4/6 inhibitors) is an effective strategy for ER-positive BC, but its clinical efficacy in LC from BC is disappointing [66,67]. For tumors not harboring drug-targetable mutations, another option is systemic immunotherapy, especially checkpoint blockade with antibodies against the programmed cell protein-1 (PD-1) or its ligand (PD-L1). In a phase 2 study using pembrolizumab (PD1-antibody), 60% of 20 patients (17 with BC) met the primary endpoint of three-month OS [68]. Toxicities of grade 3 and higher, most frequently hyperglycemia, nausea, and vomiting occurred in 40% of the patients. A recruiting NCT03719768 trial will evaluate the role of concurrent RT and avelumab, a PD-L1 antibody, on BC LC.

Discussion
LC is a rare and devastating event in the course of BC. The survival of patients treated for LC from BC has not significantly improved within the past few decades. Our PubMed. gov search identified more than 280 articles, with a growing number of publications in recent years ( Figure 1). However, only 39 (~14%) are original papers, with single publications annually ( Figure 2).
The optimal management of LC remains undefined, as there is no level I evidence from randomized clinical trials. RT, used alone or combined with systemic or intrathecal therapies, remains the main treatment modality for LC. However, its use is based on standard practices, local protocols, or individual presumptions, and not on robust evidence. The only recent recommendations for the treatment of LC come from NCCN and are dedicated to LC in general, and not specifically to BC. The DEGRO and EANO-ESMO guidelines were published 12 and 5 years ago, respectively. Due to the rarity of LC and its dismal prognosis, only a few prospective trials have been conducted. All were phase 1 or 2 (Table 5). Symptomatically, only one out of four completed trials published its final results. As of April 2022, there are six ongoing trials. Of those, four are evaluating the role of RT combined with intrathecal or systemic therapy, one is using proton RT alone for CSI irradiation, and one is comparing proton CSI versus photon IF-RT, including WBRT, focal spine RT, or their combinations. Only one of these trials is dedicated to BC patients; the remaining include mixed populations.       Another issue is the objective assessment of the response to RT in LC. The response evaluation criteria in solid tumors (RECIST) are not useful, as the infiltration of the meninges is often not measurable with this instrument. Numerous studies have shown that CSF cytology does not correlate with survival and clinical response, likely due to falsenegative testing of CSF [37]. Consequently, most studies used clinical evaluation based on neurological examination. However, the methodology of clinical assessment is subjective, may not be reproducible, and does not apply to all patients with LC (e.g., to patients with cognitive disorders). The Leptomeningeal Assessment in Neuro-Oncology (LANO) group developed a dedicated tool for evaluating the treatment response in LC [69]. However, due to its complexity and problems with validation, it has not been routinely implemented. An updated, simplified version of the LANO scorecard is under evaluation [70]. According to the EANO-ESMO recommendations, the diagnosis, response assessment, and follow-up of LC in BC patients should be based on a complete neurological examination, neuroimaging evaluation, and CSF cytology [24]. This classification seemed to be highly prognostic and was recommended for the stratification and design of clinical trials [71].

Conclusions
LC is a rare event in breast cancer patients and carries a bleak prognosis. Despite innovations in RT, little progress has been made on the use of this method in LC. Due to the rarity of the diagnosis, only single prospective trials have evaluated the role of RT for LC from BC. Faced with the difficulties in conducting prospective clinical trials, a registry of BC patients with LC might shed more light on this disastrous entity.
Funding: This research received no external funding.