NEW YORK – Researchers at MD Anderson Cancer Center are conducting further studies to explore whether neoadjuvant treatment with pembrolizumab (Merck's Keytruda) can destroy tumors in patients with localized, microsatellite unstable cancers to the extent that they can forgo the need for subsequent chemotherapy.
At the European Society for Medical Oncology Congress on Sunday, Kaysia Ludford, an assistant professor within MD Anderson's department of gastrointestinal medical oncology, said that her colleagues will explore this question in a larger cohort of patients based on the results of a 35-patient trial she presented at the meeting. That preliminary study also suggested that ctDNA testing could potentially predict which patients should go on to have surgery to remove their tumors and those who can do without it after receiving neoadjuvant pembrolizumab.
In the trial Ludford presented at the meeting, all patients had localized or locally advanced solid tumors with mismatch repair deficiency (dMMR) or high microsatellite instability (MSI-high) and were treated with pembrolizumab. The study included mostly patients with colon cancer, but several patients with rectal, pancreatic, duodenal, and other tumor types also participated.
Most gastrointestinal cancer patients notoriously don't respond well to immunotherapy, except the subset of patients who have MSI-high or dMMR tumors. Since 2017, pembrolizumab has been available in the US as a treatment for all advanced, refractory solid tumors that are dMMR or MSI-high. The anti-PD-1 drug is also approved in the US and in Europe as a first-line treatment for MSI-high or dMMR, advanced colorectal cancer.
Researchers are increasingly interested in exploring whether immunotherapy can stave off disease recurrence in earlier treatment settings. For example, in a Nature Medicine paper published last year, researchers reported that 12 out of 20 patients with dMMR colorectal cancer achieved pathologic complete response after neoadjuvant treatment with Bristol Myers Squibb's checkpoint inhibitors ipilimumab (Yervoy) and nivolumab (Opdivo).
"As we know, combination therapies can have a high rate of toxicity," Ludford said, adding, "Single-agent immunotherapy has never been studied in this setting." Patients in the MD Anderson trial received two cycles of pembrolizumab and then were evaluated for clinical benefit. If patients were benefiting from those initial doses, they received six additional doses of pembrolizumab. After a total of six months of immunotherapy, patients and their doctors decided whether to surgically remove their remaining tumor or continue to receive pembrolizumab for a year.
At the meeting, Ludford reported that out of 33 patients whose tumors were evaluable by RECIST criteria, 14 have received surgery after receiving neoadjuvant pembrolizumab, while 19 have not. In the whole study population, including the surgical and nonsurgical arms, neoadjuvant pembrolizumab shrank tumors in 26 patients, of whom nine had a complete response. Although the study was small, responses to neoadjuvant treatment didn't appear to differ according to patients' tumor mutational burden or RAS mutation status, or if they were MSI-high due to sporadic or inherited mutations.
In the surgical resection arm, 13 patients received at least three cycles of neoadjuvant immunotherapy. In that subset, Ludford reported that nine patients, or 69 percent, achieved pathologic complete response at a follow-up of nearly 10 months. The majority of patients who opted for surgery did so at around six months of treatment. Three patients went for surgery earlier due to disease progression or adverse events.
In the nonsurgical arm, Ludford reported that four patients have been on pembrolizumab for a year, opting for an entirely surgery-free management. At a median of three months of follow-up, none of the four have yet had signs of recurrence. Nine patients are still receiving the checkpoint inhibitor, one patient progressed after six weeks of treatment, and five patients have stopped taking pembrolizumab after less than a year due to various reasons, such as personal preference or toxicity. But even those five patients had some degree of benefit on neoadjuvant pembrolizumab, Ludford highlighted, such as stable disease, or a complete or partial response.
Endoscopic evaluation of 28 patients who had luminal gastrointestinal cancers and are at particularly high risk of dying from their disease showed more than half had a complete response on neoadjuvant pembrolizumab. Four patients in this subset had a near complete response. "This has implications for organ-sparing strategies," said Ludford.
In terms of adverse events, neoadjuvant pembrolizumab resulted in grade 3/4 toxicities in three patients, who had transaminitis, diarrhea, and type 1 diabetes. The patient who experienced transaminitis had to be hospitalized, but then had surgery and went on to achieve a pathologic complete response. Of the patients who opted for surgery, one had a manageable hematoma and another had a pelvic abscess.
Ludford and colleagues tested patients' ctDNA levels at the start of the study and after three weeks of pembrolizumab treatment using an in-house 70-gene liquid biopsy panel. They wanted to see if ctDNA levels could predict which patients would achieve a complete pathologic response to neoadjuvant immunotherapy. Out of 20 patients with positive ctDNA levels at baseline, the majority saw declines in the variant allele frequency of their most prevalent variant.
For example, APC variants had the highest variant allele frequency in one patient with a large abdominal tumor, but after three weeks of neoadjuvant pembrolizumab, the APC variant allele frequency was undetectable in ctDNA testing. "Five months later, the patient had a partial response [to pembrolizumab] on imaging, and at six months, the patient underwent surgery and had a complete pathologic response," Ludford said.
CtDNA testing could also predict early on the outcomes of one of the patients who ended up progressing on pembrolizumab. Ludford is hopeful that with further research liquid biopsies may become a reliable tool for identifying early which patients need surgery after neoadjuvant immunotherapy, as well as for identifying the patients who can forgo surgery and do well with just neoadjuvant pembrolizumab. "What we would not want to happen is for patients who are potentially candidates for resection to miss out on that opportunity," she said. "Identifying progressors early is one way [ctDNA testing] would be meaningful."
Using ctDNA to predict responses to neoadjuvant pembrolizumab is promising, agreed Sherene Loi, head of the Translational Breast Cancer Genomics and Therapeutics Laboratory at the Peter MacCallum Cancer Center in Australia, during her discussion of this trial. She noted the need for more accurate biomarkers to predict which patients are likely to experience pathologic complete response with neoadjuvant treatment.
"Certainly, in breast cancer, we know that the absence of invasive disease after surgery after neoadjuvant treatment is associated with a near 100 percent survival [rate]," Loi said. "And we use [pathologic complete response] now to escalate or de-escalate treatment post-surgery."
She urged researchers to explore whether pathologic complete response could similarly become a standard endpoint for deciding when to escalate or de-escalate interventions for patients with MSI-high or dMMR solid tumors receiving pembrolizumab in the neoadjuvant setting.
Ludford's group will explore the efficacy and safety of neoadjuvant pembrolizumab in a larger cohort of MSI-high or dMMR solid tumor patients. Because this initial study was small, the researchers couldn't identify any biomarkers that distinguished progressing patients from those that achieved pathologic complete response. However, Ludford noted that her team will interrogate the tumor microenvironment of responding and non-responding patients for predictive biomarkers.