November 12, 2020 — Pembrolizumab led to a 38% central nervous system response rate and was well tolerated in patients with leptomeningeal metastasis from solid tumors. However, the study was closed early due to poor accrual.
Jarushka Naidoo, MD
November 12, 2020 — Pembrolizumab (Keytruda) led to a 38% central nervous system (CNS) response rate and was well tolerated in patients with leptomeningeal metastasis (LMM) from solid tumors, according to phase 2 results presented during the virtual 2020 SITC Annual Meeting.1 However, the study was closed early due to poor accrual.
Additionally, as part of an exploratory analysis, findings showed that CSF aneuploidy of t-DNA may be sensitive to detecting LMM, and immunologic subsets of response to PD-1 inhibitors were identified by multiplex cytokine profiles, which warrants further study, explained lead study author Jarushka Naidoo, MD, in a virtual presentation during the meeting.
“This investigator-initiated biomarker-rich study yielded several interesting findings. From an exploratory perspective, we identified that CSF aneuploidy of t-DNA may be sensitive for the detection of leptomeningeal metastasis, and we are looking into this in larger datasets of patients,” said Naidoo, a thoracic oncologist with Beaumont Hospital Dublin, and an adjunct assistant professor at Johns Hopkins University.
“From a clinical perspective, unfortunately, our study closed early due to poor accrual, and this highlights that this is a challenging population in which to do prospective trials, but it is possible. Reassuringly, pembrolizumab was well tolerated, and this is extremely important in a patient population that is traditionally quite frail, and in a population in which other standard therapies … are associated with far higher rates of toxicity,” Naidoo added. “We were reassured to see that pembrolizumab conferred an impressive 38% CNS response rate … and therefore, this needs to be studied in larger populations of patients to confirm this result, but could be used as a potential treatment option for patients with leptomeningeal disease from solid tumors.”
Patients with brain metastases from select tumors may benefit from immune checkpoint inhibitors, as seen in melanoma and non–small cell lung cancer (NSCLC), Naidoo explained. However, the benefit of this class of agents in those with other types of CNS metastasis is unknown.
LMM is a rare form of CNS metastasis and is most commonly seen in patients with breast carcinoma (18%), lung cancer (17%-26%), and melanoma (9%)2, and is associated with a median survival of 6 to 24 weeks.3 Currently, no universally standard therapies for these patients with LMM from solid tumors exist.
In this phase 2 trial, 16 patients in any line of therapy with advanced solid tumors who had LMM, detected via magnetic-resonance imaging or cerebrospinal fluid (CSF), were treated with single-agent pembrolizumab at 200 mg every 3 weeks for 2 cycles before being assessed for CNS response.
Investigators hypothesized that 35% of patients would achieve a response in the CNS, with a null hypothesis of 10%. Patients who derived benefit from pembrolizumab could continue treatment until disease progression or unacceptable toxicity.
To be eligible for enrollment, patients must have had cytologically confirmed LMM or radiologically detectable leptomeningeal disease, an ECOG performance status of 0 to 1, and a non-escalating steroid requirement. Patients could continue on targeted therapy if their CNS disease developed while receiving the agent and was found to be safe when combined with a PD-1 inhibitor.
Those who had prior whole-brain radiation within 3 months of enrollment, underwent prior neurosurgery within 3 weeks of study, had prior PD-1/PD-L1–directed therapy, had autoimmune disease that required systemic steroids or immunosuppression, and were contraindicated to lumbar puncture or MRI were excluded from enrolling on the study.
Twenty-six patients were screened for the study; 13 patients were excluded due to screening failure, in most cases due to a decline in performance status between the time of screening and proposed start of study treatment.
While the study closed early due to poor accrual, investigators were still able to analyze the data from the 13 patients who received treatment. Here, the median age was 57 years (range, 22-79), and 92.3% of patients (n = 12) had an ECOG performance status of 0. Patients had breast carcinoma (38.4%; n = 5), high-grade glioma (23.1%; n = 3), NSCLC (7.6%; n = 1), squamous cell carcinoma of the head and neck (7.6%; n = 1), and cutaneous squamous carcinoma of the skin (38.4%; n = 5).
Most patients (69.3%; n = 9) did not have extracranial disease, and more than half (53.8%; n = 7) had CSF-negative cytology. A total 76.9% (n = 10) received prior chemotherapy and also prior brain/spine radiation. Nine patients (69.2%) had 2 more lines of prior therapy, and 61.5% (n = 8) had received corticosteroids at study enrollment.
The primary end point was overall CNS response. Patients who sustained a complete or partial response (PR), or stable disease (SD), or on an overall CNS response determination were defined as satisfying the primary outcome measure of the study. Secondary end points were CNS progression-free survival (PFS), CNS overall survival (OS), and safety.
Exploratory end points were novel CSF biomarkers of LMM or response to immunotherapy; investigators assessed for the presence of tumor-derived DNA (t-DNA) in the CSF. Investigators also looked to evaluate immunologic features of the CSF through flow cytometry and multiplex cytokine panels.
Results showed that 5 patients had an overall CNS response to treatment with pembrolizumab, which comprised 2 complete responses (CRs), 1 PR, and 2 patients with SD. One CR was achieved in a patient with squamous cell carcinoma of the skin, with an ongoing CR at 3 years.
The second CR was in a patient with MET exon 14 skipping–mutated NSCLC which lasted for 9 months, followed by disease progression that metastasized elsewhere; the patient later died.
Additional data showed that the median CNS PFS was 2.9 months (95% CI, 1.2–not reached [NR]) and the median CNS OS was 4.9 months (95% CI, 3.7–NR).
“Notably, even though numbers are small, we do see a tail on the curve phenomenon in both of these survival curves, which is consistent with checkpoint blockade–prospective studies,” Naidoo said.
Regarding safety, 5 patients (38.5%) developed treatment-related adverse events (TRAEs), the most common all-grade events being pain in extremity, joint reduced range of motion, fatigue, and pruritus (n = 2 each). Grade 3 or higher TRAEs included fatigue and urinary tract infection (n = 1 each).
Immune-related adverse events were experienced by 3 patients (23.1%), none of which were high-grade in nature: pain in extremity (n = 2), joint reduced range of motion (n = 2), pruritus (n = 2), maculopapular rash (n = 1), limb edema (n = 1), and pneumonitis (n = 1).
For the exploratory end point, data also showed that CSF t-DNA may be sensitive for the detection of LMM. Use of the RealSeq aneuploidy assay, led to a sensitivity rate of 84.6% (95% CI, 54.6%-98.1%) compared with the 53.8% (95% CI, 25.1-80.8%) sensitivity rate through CSF cytopathology.
“Therefore, CSF t-DNA is being studied in larger numbers of patients with both primary brain tumors and suspected leptomeningeal disease for its potential utility as a means by which we may detect the presence of leptomeningeal disease,” Naidoo explained. “Unfortunately, we did not identify that CSF t-DNA aneuploidy is associated with response or survival in this particular data set. However, we did identify that CSF cytokine profiles may identify immunologic subsets of response.”