How Keytruda Is Used to Treat Brain Metastases

Keytruda (pembrolizumab) treats brain metastases by blocking a protein called PD-1 on immune cells, essentially removing a brake that tumors exploit to hide from the immune system. For years, oncologists assumed immunotherapy could not work in the brain, and patients with brain metastases were routinely excluded from clinical trials. That assumption turned out to be wrong. Phase II trials have shown that pembrolizumab can shrink brain metastases in a meaningful fraction of patients with melanoma and non-small cell lung cancer, and the drug is now used in clinical practice both alone and alongside radiation or chemotherapy for selected patients with brain involvement.

Why the Brain Was Considered Off-Limits

The brain has its own immune environment, separated from the rest of the body by the blood-brain barrier. For decades, this barrier led oncologists to assume that large antibody molecules like pembrolizumab simply could not reach brain tumors in therapeutic concentrations, and that the brain’s immune landscape was too suppressed for checkpoint inhibitors to do much even if they got there. Patients with active brain metastases were excluded from nearly all major immunotherapy trials as a result.

That dogma began to crack in the mid-2010s when early case reports and small trials showed that some patients with brain metastases responded to PD-1 inhibitors despite the theoretical barriers.¹ The blood-brain barrier is not perfectly intact in most brain metastases. Tumor growth disrupts it, which likely allows at least some drug penetration. That said, how well pembrolizumab crosses into the brain remains variable from patient to patient. Case series have documented patients who achieved strong systemic responses but developed new brain lesions, while others showed remarkable shrinkage in both the body and the brain.² The inconsistency has practical implications: a good response outside the brain does not guarantee the same inside the skull.

Evidence in Melanoma Brain Metastases

Melanoma was one of the first cancers where pembrolizumab was tested against brain metastases, partly because melanoma responds well to checkpoint inhibitors in general and has a high rate of spreading to the brain. In a phase II trial of patients with melanoma and active brain metastases, about a quarter had their brain tumors shrink, and responses that did occur tended to be durable. Roughly half the patients in that trial were still alive at two years, with a median overall survival of 17 months.³ Those numbers sound modest until you consider that melanoma brain metastases historically carried a prognosis measured in weeks to a few months.

A separate retrospective study reinforced the signal. Among patients with melanoma brain metastases treated with anti-PD-1 therapy (which includes pembrolizumab), median overall survival reached nearly 38 months, compared with about 19 months for those treated with the older checkpoint inhibitor ipilimumab and roughly 13 months for those on BRAF-targeted drugs.⁴ Treatment with anti-PD-1 therapy was independently associated with longer survival, alongside surgical removal of brain tumors and stereotactic radiosurgery.

Evidence in Non-Small Cell Lung Cancer

Brain metastases are common in lung cancer, affecting a large proportion of patients over the course of their disease. In an open-label phase II trial, pembrolizumab produced brain metastasis responses in about 30% of patients with non-small cell lung cancer (NSCLC) whose tumors expressed PD-L1 at 1% or higher. The responses were durable in many cases and comparable to the systemic response rate the drug achieves outside the brain.⁵ Patients whose tumors did not express PD-L1, however, had no brain responses at all in that trial, underscoring how much the biomarker matters for lung cancer specifically.

An earlier report from the same trial program found brain responses in about a third of NSCLC patients and about a fifth of melanoma patients, with most lung cancer responders maintaining their response at the time of analysis.⁶ What makes these results striking is that they come from patients with untreated brain metastases who had not received prior brain-directed radiation, meaning pembrolizumab alone was doing the work.

Combining Pembrolizumab With Radiation

In practice, many patients with brain metastases receive stereotactic radiosurgery (SRS), a focused form of radiation that targets individual tumors. A growing body of evidence suggests that combining SRS with checkpoint inhibitors like pembrolizumab may be more effective than either treatment alone. The rationale is that radiation can cause tumor cells to release signals that make the immune system more aware of them, while the checkpoint inhibitor keeps the immune response from being shut down.

In a study of melanoma patients with brain metastases, those who received checkpoint immunotherapy around the same time as SRS had significantly greater tumor shrinkage than those who got SRS alone. At five months after treatment, tumors shrank by roughly 52% in the combination group compared with about 15% with SRS alone. Receiving immunotherapy concurrently with SRS also predicted better freedom from new brain lesions.⁷ Timing appears to matter: the benefit was strongest when the two treatments overlapped closely rather than being separated by weeks.

A systematic review and meta-analysis has since confirmed that combining PD-1 or PD-L1 inhibitors with SRS can enhance tumor control through what appears to be a synergistic relationship.⁸ This combination is increasingly used in clinical practice, though the optimal timing and sequencing remain active areas of investigation.

Combining Pembrolizumab With Chemotherapy for Lung Cancer

For patients with NSCLC and stable brain metastases, adding pembrolizumab to platinum-based chemotherapy has shown substantial survival benefits. A pooled analysis of three major trials (KEYNOTE-021, KEYNOTE-189, and KEYNOTE-407) found that patients with brain metastases who received pembrolizumab plus chemotherapy had a median overall survival of about 19 months, compared with roughly 8 months for those who received chemotherapy alone. The survival advantage in the brain metastasis group was at least as large as the benefit seen in patients without brain involvement.⁹ This is noteworthy because brain metastases have traditionally been associated with worse outcomes across the board, and having brain involvement did not blunt the benefit of adding pembrolizumab.

It is worth noting that these trials enrolled patients whose brain metastases were treated and stable, not patients with large, symptomatic, or actively growing brain tumors. That distinction matters when interpreting who this combination is appropriate for.

The Corticosteroid Problem

Many patients with brain metastases need corticosteroids like dexamethasone to control swelling around their tumors. This creates a dilemma, because corticosteroids suppress the immune system, and immunotherapy depends on a functioning immune response. For tumors outside the brain, corticosteroid use has been linked to worse outcomes with PD-1 inhibitors.

The brain, however, may be a special case. In preclinical mouse models, corticosteroids reduced the effectiveness of anti-PD-1 therapy against tumors growing in the body but did not diminish its benefit against tumors growing inside the brain. The researchers proposed that the brain’s unique immune environment may offer some protection against the immunosuppressive effects of steroids, allowing checkpoint blockade to remain effective even when systemic immunity is dampened.¹⁰ These are animal findings, so they come with caveats. But they offer some reassurance that patients who need steroids for brain swelling are not necessarily losing the benefit of pembrolizumab for their intracranial disease. Most clinicians still try to minimize steroid doses where possible.

Why PD-L1 Testing Is Not Straightforward

Pembrolizumab works best in tumors that express PD-L1, the protein that binds to PD-1 and tells immune cells to stand down. In lung cancer, PD-L1 expression is practically a prerequisite for response in the brain. But testing for PD-L1 in brain metastases has several complications.

First, the PD-L1 levels in a brain metastasis do not always match the levels in the original tumor. One study found that PD-L1 was expressed in about 24% of brain metastases and 29% of the primary tumors they came from. The two were concordant roughly three-quarters of the time, meaning about a quarter of cases had a mismatch.¹¹ The immune cell infiltration in brain metastases also tends to be lower than in the primary tumor: significant CD8 T-cell presence (the cells that actually kill tumor cells) was found inside about 44% of brain metastases compared with about 62% of primary tumors.

Second, PD-L1 expression can change over time and varies depending on which antibody test is used. Different studies have found conflicting results about whether PD-L1 levels predict response, and some of the disagreement stems from unstandardized testing methods.¹² In practice, this means that a PD-L1 test from a biopsy taken months earlier or from the primary tumor may not accurately reflect what is happening in the brain metastasis at the time of treatment. When brain tissue is available from surgery, testing the brain metastasis directly gives a more reliable picture.

Telling Tumor Growth Apart From Immune Inflammation on MRI

One of the trickiest aspects of using pembrolizumab for brain metastases is interpreting follow-up brain scans. Immunotherapy can cause a phenomenon where tumors appear to get bigger or new lesions appear on MRI before the treatment actually starts working. This is sometimes called pseudoprogression: immune cells flooding into the tumor create swelling and inflammation that looks like tumor growth on imaging.

The Response Assessment in Neuro-Oncology (RANO) working group has recommended that when patients show imaging findings that look like progression within six months of starting immunotherapy, including what appear to be new lesions, doctors should confirm the progression on a follow-up scan before changing treatment, provided the patient is not clinically deteriorating.¹³ Pulling a patient off an effective treatment too early because of a misleading scan is a real risk, but so is staying on an ineffective treatment while brain tumors grow. Neuro-oncologists and radiologists have to weigh these possibilities for each individual patient.

Radiation Necrosis Risk

When pembrolizumab is combined with radiosurgery, one concern is radiation necrosis, where normal brain tissue near the treated area dies due to a combination of radiation damage and inflammatory response. One study found that patients who received immunotherapy alone had a radiation necrosis or treatment-related imaging change rate of about 38%, compared with about 17% for those receiving chemotherapy only.¹⁴ The immunotherapy group had a statistically significantly higher risk. However, other studies have not found this same signal. A retrospective study of melanoma patients treated with Gamma Knife radiosurgery found no significant difference in hemorrhage or radiation necrosis rates related to immunotherapy use.¹⁵

The disagreement in the literature likely reflects differences in study populations, radiation doses, timing of immunotherapy relative to radiation, and how radiation necrosis was defined and detected. For now, most clinicians monitor patients closely with serial brain MRIs after combined treatment and manage radiation necrosis with steroids or bevacizumab if it becomes symptomatic.

Leptomeningeal Disease

Leptomeningeal metastasis, where cancer cells spread to the membranes surrounding the brain and spinal cord, is one of the most difficult complications to treat. It carries a very short life expectancy, typically measured in weeks. Pembrolizumab has been tested in this setting as well. In a phase II trial of patients with leptomeningeal metastasis from various solid tumors, about 38% showed a central nervous system response at 12 weeks, and two patients achieved complete responses. One of those complete responders survived more than three years. However, the median overall survival for the group was still only about five months, and median time before the disease progressed in the brain was roughly three months.¹⁶ Serious treatment-related side effects occurred in 15% of patients. These results are modest but represent progress in a disease where historically almost nothing worked.

Rare but Serious Neurological Side Effects

Pembrolizumab works by releasing the immune system’s brakes, which means the immune system can sometimes attack healthy tissue. Neurological side effects are uncommon but can be severe. A case report documented what was described as the first known instance of pembrolizumab-induced autoimmune encephalitis, where a patient with triple-negative breast cancer developed sudden difficulty speaking three days after starting the drug. After ruling out infection and metastatic spread, doctors concluded that the immune system had attacked the brain itself in response to the checkpoint inhibitor.¹⁷ Other neurological immune-related adverse events that have been reported with PD-1 inhibitors include myasthenia gravis, peripheral neuropathy, and meningitis, though all are rare.

The challenge with brain metastasis patients specifically is that new neurological symptoms could represent tumor progression, radiation effects, infection, or an immune-related adverse event. The diagnostic workup for any new symptom tends to be more extensive in this population.

Quality of Life After Treatment

Survival is not the only thing that matters. A longitudinal study of metastatic melanoma survivors treated with pembrolizumab found that even patients who survived long-term reported meaningfully lower quality of life compared with healthy populations across physical, emotional, cognitive, social, and role-functioning domains. About two-thirds experienced clinical levels of anxiety or depression at some point during follow-up. Nearly half met criteria for cancer-related post-traumatic stress, and about a third showed cognitive impairment on neuropsychological testing.¹⁸ The study was small and included patients with various disease sites, not only brain metastases, but these findings highlight that living longer with advanced cancer can come with a substantial emotional and cognitive burden. For patients with brain metastases specifically, the cognitive impact may be compounded by the tumors themselves and any brain-directed treatments.

Who Is Most Likely to Benefit

Not every patient with brain metastases is a good candidate for pembrolizumab. The strongest evidence exists for patients with melanoma or NSCLC whose brain metastases are small, asymptomatic or minimally symptomatic, and do not require high doses of corticosteroids. PD-L1 expression matters a great deal in lung cancer, where patients without PD-L1 expression on their tumors have had essentially no brain responses to pembrolizumab alone. In melanoma, responses have occurred regardless of PD-L1 status, though the numbers are better in PD-L1-positive tumors.

A study of NSCLC patients with high PD-L1 expression found that patients whose brain metastases were not specifically targeted with radiation still achieved complete brain responses in some cases, particularly when the brain lesions were small.¹⁹ Large, symptomatic brain metastases causing significant neurological problems generally need upfront treatment with surgery or radiation before immunotherapy can be considered. Patients with rapidly progressive brain disease or those who need high-dose steroids for symptom control are the hardest group to treat with checkpoint inhibitors.

The Cost Question

Pembrolizumab is expensive, and whether its cost is justified varies by tumor type and setting. A cost-effectiveness analysis that modeled pembrolizumab across multiple cancers found that for brain cancers, the drug was not cost-effective at a willingness-to-pay threshold of $150,000 per quality-adjusted life year, though it was cost-effective for certain other tumor types like colorectal and endometrial cancers.²⁰ That analysis looked at primary brain cancers rather than brain metastases from other tumor types, so the finding does not directly translate to the melanoma or lung cancer brain metastasis population, where the survival gains are larger. Still, the annual cost of pembrolizumab treatment runs into six figures in the United States, and insurance coverage, prior authorization requirements, and access to comprehensive neuro-oncology centers all affect which patients actually receive it.