Meningeal carcinomatosis has a higher incidence in patients with oncogene-addicted NSCLCs such as ROS1-positive disease, mainly due to the availability of effective therapies associated with unprecedented overall survival [1, 2]. Accordingly, our patient developed meningeal carcinomatosis following a long disease history, during which she received two ROS1-TKIs in sequence, namely crizotinib and lorlatinib. At the time of diagnosis of meningeal carcinomatosis, the SFM G2032R was detected in both plasma and CSF, which prompted us to employ repotrectinib, a drug with known activity against this resistant variant of ROS1-positive lung cancer [8]. Whole brain radiotherapy (WBRT) was not considered based on the fact that we expected a CNS response given the preliminary activity reported with repotrectinib in the brain [3, 8]. Another reason for avoiding WBRT was its uncertain activity in leptomeningeal disease, especially in oncogene-addicted NSCLC, and our intention to spare the toxic effect of brain radiation and preserve quality of life [10]. Of note, after roughly 1 week of treatment, symptoms of increased intracranial pressure rapidly improved, until complete resolution at approximately week 2 of therapy. In addition, steroids were safely tapered and completely discontinued within 1 month of treatment. Importantly, steroids could not be the reason for symptom improvement because: (1) steroids were administered at the time of admission to the hospital 2 weeks before repotrectinib, and no improvement was noted thereafter; (2) neurologic manifestations of disease did not reappear for a relatively long period of time despite steroid discontinuation ever since. Unfortunately, despite a remarkable partial metabolic extracranial response, neurologic symptoms re-emerged after approximately 3 months, with pathological and radiological confirmation of progression of meningeal carcinomatosis.

To test the ability of repotrectinib to cross the BBB, we measured its concentrations in plasma and CSF. Of note, the CSF/plasma albumin quotient of our patient (= 33.8) was compatible with marked disruption of the BBB, which led us to hypothesize that the drug could effectively penetrate into the CSF (Table 1). We found a plasma and CSF concentrations of 1317.0 ng/mL and 14.9 ng/mL, respectively, for a CSF/plasma ratio of 0.0113, which suggests that approximately 1% of the drug is able to penetrate through the BBB into the CSF (Table 1). Therefore, the presence of repotrectinib in the CSF could justify the exceptional symptomatic response observed in our case. Importantly, the intracranial activity of repotrectinib against brain metastases from ROS1-positive NSCLC has been recently reported. In the phase 1/2 TRIDENT-1 trial, intracranial response of CNS metastases was 33% (8 out of 24 patients with measurable and not measurable lesions) in patients pretreated with one ROS1-TKI [3]. In this sense, repotrectinib appears to have some degree of anti-tumor activity against ROS1-positive CNS metastases in ROS1-TKI refractory patients. Our report extends the potential CNS activity of repotrectinib also to meningeal carcinomatosis, which represents a devastating complication from NSCLC, generally associated with an overall survival of less than 1 year [11].

Nevertheless, we must acknowledge that other factors beyond mere drug penetration through the BBB may contribute to the antitumor effectiveness of a drug against CNS metastases, which may explain the short duration of CNS-response observed in our patient. For instance, repotrectinib is found to be highly bound (95.4%) to plasma proteins, thus possibly limiting the effectiveness of target inhibition by the low fraction of unbound repotrectinib [repotrectinib investigator’s brochure, data on file]. Furthermore, repotrectinib has been shown to act as substrate for the efflux transport P-glycoprotein, which may further limit repotrectinib’s activity in the CNS by keeping the drug out of the brain [12].

Interestingly, the present report suggests that the CSF penetration of repotrectinib is intermediate between crizotinib (median = 0.2%) and lorlatinib (median for unbound drug = 67.9%) (Table 2) [13,14,15,16]. In line with this, lorlatinib has demonstrated an exceptional antitumor activity in ROS1-positive NSCLCs with CNS-only disease progression who were refractory to the ROS1-TKI crizotinib [4]. Consistently, a recently published case report has described a “Lazarus” response to lorlatinib in a patient with meningeal carcinomatosis without additional mutation(s) potentially implicated in resistance across the ROS1 kinase domain (e.g., G2032R) [17]. By contrast, in our case, in which meningeal carcinomatosis was detected after crizotinib and lorlatinib along with documentation of the SFM G2032R mutation, repotrectinib appeared to be the optimal treatment choice.

In conclusion, we report for the first time a clinically significant, though short-lived, response of meningeal carcinomatosis to repotrectinib in a heavily ROS1-TKI-pretreated patient with a G2032R mutation, which is in line with its 1% CSF/plasma penetration that was found in this specific case. Evidently, the penetration of repotrectinib may depend on several factors such as type of CNS disease (parenchymal and/or meningeal involvement) and type/number of prior local treatments received (stereotactic and/or WBRT) that may influence BBB permeability and that cannot be captured by a single case report. Given the common involvement of CNS as metastatic site following treatment with a ROS1-TKI, novel active agents should be specifically developed to both address ROS1-dependent resistance mechanisms and actively penetrate the BBB.