NEW YORK – Updated biomarker analysis released by Genentech at the European Society for Medical Oncology Congress showed that its checkpoint inhibitor atezolizumab (Tecentriq) delayed relapse in some early-stage non-small cell lung cancer patients with PD-L1 expression in 1 to 49 percent of tumor cells, but the disease-free survival benefit from adjuvant treatment was far more modest compared to those with higher PD-L1 expression.
This analysis from the IMpower010 trial, though exploratory and not conclusive, sparked much debate among experts about how the data might impact their practice. Some oncologists at the meeting said they'll prescribe atezolizumab in the adjuvant setting for the full range of low-to-high PD-L1 expressers if regulators approve the drug based on its disease-free survival impact in IMpower010 and re-evaluate based on future overall survival readouts in the post-market setting.
Even though atezolizumab was better than standard treatments at staving off relapse in PD-L1-positive NSCLC patients, "I don't want later relapse," said Benjamin Besse, head of clinical research at Gustave Roussy in France, in reviewing the IMpower010 data. "We want to cure patients, and here we cannot guess from this data if overall survival will be positive."
Others were also skeptical and said that even in the face of a broad approval for adjuvant atezolizumab from the US Food and Drug Administration, they were unlikely to prescribe the drug to too many patients with low PD-L1 expression based on the latest biomarker subset analysis.
All experts seemed to agree, however, that PD-L1 expression is an insufficient biomarker to truly personalize immunotherapy for their patients and better strategies are urgently needed especially as checkpoint inhibitors are increasingly used in earlier treatment settings. "The pain that we're experiencing with clinicians with this lousy biomarker is sad," said Alona Zer, who heads up the thoracic oncology clinic at the Davidoff Cancer Center in Israel. "We really need better biomarkers, especially now that immunotherapy is going into the adjuvant setting."
Meanwhile, the IMpower010 data were published in The Lancet on Monday, and the ESMO presentation prompted Genentech to highlight that "Tecentriq is the first and only cancer immunotherapy to demonstrate positive Phase III results in the adjuvant lung cancer setting." The Roche subsidiary acknowledged the latest data showed a "higher magnitude of benefit from adjuvant Tecentriq" in patients with PD-L1 expression in 50 percent or more tumor cells compared to those with lower expression but added that the exploratory nature of the analysis "prevents any firm conclusions."
Genentech is pursuing FDA approval for adjuvant atezolizumab in early-stage NSCLC, though it's not clear if the company is seeking to narrow the indication based on a specific PD-L1 expression cutoff. The agency is slated to make a decision Dec. 1.
Insight into a subgroup
At the American Society of Clinical Oncology's annual meeting in June, Genentech reported interim results from IMpower010, showing that NSCLC patients with stage II to IIIA tumors with PD-L1 expression in at least 1 percent of cells had a 34 percent lower risk of death or disease recurrence from adjuvant atezolizumab following chemotherapy and surgery compared to those on best supportive care. However, the subset of patients with PD-L1 expression in 50 percent or more tumors cells had a 57 percent lower risk of death or disease recurrence on adjuvant atezolizumab compared to those on best supportive care. In contrast, patients whose tumors did not express PD-L1, comprising nearly half of the study population, didn't appear to benefit more from adjuvant atezolizumab than best supportive care.
What was missing from this earlier analysis were the outcomes for the subset of patients with PD-L1 expression in 1 to 49 percent of tumor cells, which Enriqueta Felip, head of the lung cancer unit at Vall d'Hebron University Hospital in Spain, presented at the ESMO Congress. Among patients with PD-L1 expression in this low-to-mid range, adjuvant atezolizumab reduced the risk of death or disease recurrence by 13 percent compared to best supportive care (hazard ratio .87).
"A hazard ratio of .87 in disease-free survival is not striking," said Luis Paz-Ares, chair of the medical oncology department at the Hospital Universitario in Spain. But he agreed with others that if atezolizumab is approved by regulators in patients with at least 1 percent tumor PD-L1 expression, then he will prescribe it weighing other factors. Some oncologists observed that they'd be more inclined to prescribe adjuvant atezolizumab in patients who have PD-L1 expression in 30 to 40 percent of tumor cells and look for other options for those on the lower end of the spectrum.
This data in low-to-mid-range PD-L1 expressers comes from a post hoc exploratory analysis and was not predefined like the disease-free survival analysis in the high-PD-L1 expressers, cautioned Felip. "This is the reality, [and] we have this hazard ratio of .87," she said. "We probably need more follow-up in this population." Based on the data in hand, she said adjuvant atezolizumab could still be a good option for some with low-to-mid-range PD-L1 expression after considering features of their disease, their preference, and additional actional biomarkers.
In IMpower010, investigators enrolled patients in the trial until 2018, before the benefit of adjuvant osimertinib (AstraZeneca's Tagrisso) in EGFR-mutated early-stage NSCLC patients was known from the ADAURA trial. As such, in IMpower010, around 10 percent of patients had EGFR mutations and 4 percent had ALK rearrangements, and, according to an exploratory analysis, appeared to derive less benefit from adjuvant atezolizumab. "The hazard ratio is not bad for patients with EGFR mutations," said Felip, but noted now that osimertinib is available in the adjuvant setting, such patients should probably receive that.
In addition to looking at tumor biomarkers, Besse from Gustave Roussy emphasized the importance of looking at "host" parameters, such as the patient's immune features, whether the patient has anti-drug antibodies that can degrade atezolizumab, and if the patient's age can diminish their ability to mount an immune response.
Felip also reported at the meeting that as of Jan. 21 this year, around 29 percent of patients with PD-L1 expression in at least 1 percent of tumor cells had relapsed on atezolizumab, while 45 percent had relapsed on best supportive care. There were no differences in the patterns of relapse between the two treatment arms regardless of whether patients had PD-L1-positive tumors.
Further post hoc analysis showed that PD-L1-positive patients on atezolizumab had a longer disease-free period before relapse than those on best supportive care. The same time-to-relapse advantage was not seen between treatment arms in all randomized patients, regardless of PD-L1 status, with stage II to IIIA tumors and in the even broader intent-to-treat population with stage IB to IIIA tumors.
Although regulatory bodies accept disease-free survival in adjuvant therapy trials like IMpower010, overall survival is the gold standard, and in Besse's view, most oncologists would rather cure patients than delay recurrence. Earlier this year, the Phase III PACIFIC trial showed that half of the stage III NSCLC patients who received AstraZeneca's immunotherapy durvalumab (Imfinzi) after platinum chemotherapy and radiation were alive at four years; the benefit was seen in patients across PD-L1 expression groups.
Besse is encouraged by this but would like to see more data from ongoing studies and the overall survival analysis in IMpower010 before making a final decision on which patients to treat with adjuvant atezolizumab.
Once the overall survival data are mature in IMpower010, the field will also have greater insight into the risk-benefit profile of adjuvant atezolizumab in early-stage NSCLC. Immunotherapy can lead to hyper-progression of metastatic tumors, and based on very early survival data in IMPower010, atezolizumab doesn't appear to drive hyper-progression of micro-metastatic disease in early-stage tumors, Besse reflected.
Never-ending debate over PD-L1
The question of how well atezolizumab treats micro-metastatic lesions is an important one in adjuvant treatment trials and helps put into focus why experts want better biomarkers in this setting. In adjuvant immunotherapy studies, researchers are looking at PD-L1 expression in resected early-stage tumors. However, the goal of adjuvant treatment is to stop the undetected micro-metastatic disease that might still remain in the body despite surgery and therapy and cause relapse.
Micro-metastatic disease "is a disease you cannot touch," Besse said. "What is the rate of PD-L1 expression in micro-metastatic disease? You will never know. So, you need something else."
Using liquid biopsy tests to track ctDNA and monitor minimal residual disease could be that "something else," many experts agreed. "Patients who are minimal residual disease-positive after surgery might still have micro-metastasis," Besse said. "It's a very nice tool to escalate or de-escalate adjuvant treatment."
AstraZeneca is looking at this very premise in the Phase III MERMAID-1 trial, in which researchers are comparing adjuvant durvalumab plus chemotherapy against chemotherapy in early-stage NSCLC patients who have minimal residual disease after surgery.
In the IMpower010 trial, Genentech also collected blood samples, which may allow for future analysis of MRD as a tool for monitoring relapse after adjuvant atezolizumab.
In IMpower010, the PD-L1 subgroup analysis is particularly controversial because Genentech used different immunohistochemistry assays — Roche's Ventana PD-L1 SP142 and SP263 tests. "The test that is used to stratify the patients, SP142, is different from the test that is used to give you the results of the benefit according to PD-L1 testing," said Besse.
Solange Peters, current ESMO president and chair of thoracic oncology at the University Hospital in Lausanne, Switzerland, further elaborated that Genentech chose to stratify patients according to PD-L1 expression on immune and tumor cells using SP142 but analyzed study results by looking at PD-L1 expression just on tumor cells using the SP263 test. "This will be a never-ending debate, particularly because of all the blurred parameters, which we have to deal with," she said, wondering whether the test switch puts into question the validity of the PD-L1 stratification in IMpower010.
The switch does makes it "hugely problematic" for oncologists to parse the data and home in on which patients are likely to respond and which are not, suggested Keith Kerr, a consultant pathologist at Aberdeen University in the UK. "I think that's why the switch has taken place," he reflected.
He further observed that patients in the low-to-mid-range PD-L1 expression group are a biologically heterogeneous bunch, with most patients falling into the low end of the spectrum. "We need to bear in mind that there is a whole load of difference biologically between someone who has 2 percent PD-L1 expression in tumor cells and someone who has 40 percent PD-L1 expression in tumor cells," Kerr said.
Left with this ambiguity, most oncologists are eagerly waiting to see how regulators will interpret the data from IMpower010, though they don't always provide clarity. Paz-Ares from the Hospital Universitario hopes that US and European regulators will align their review of IMpower010 and come to the same decision about which NSCLC indication of adjuvant atezolizumab to approve.
Based on the PACIFIC trial, the FDA approved AstraZeneca's durvalumab after chemo and radiation in unresectable stage III NSCLC patients, regardless of PD-L1 status. However, European regulators restricted their approval to patients with PD-L1 expression in at least 1 percent of tumor cells based on a small post hoc analysis that failed to show a survival advantage in PD-L1-negative patients.
It doesn't help oncologists standardize best practices when regulatory bodies interpret data from pivotal drug trials differently. In the case of IMpower010, "I would really like the FDA and EMA not to do what they did with the PACIFIC trial," Paz-Ares said. "Approve [adjuvant atezolizumab] and ask for a big trial [involving patients] in the 1 to 49 [percent] PD-L1 group."