In a <em>Targeted Oncology</em>™ case-based peer perspective live discussion, Sandip P. Patel, MD, associate professor of medicine at the University of California San Diego Health, discussed EGFR inhibition options for patients with non–small cell lung cancer, based on the case of a real patient.
Sandip P. Patel, MD
In aTargeted Oncology™ case-based peer perspective live discussion, Sandip P. Patel, MD, associate professor of medicine at the University of California San Diego Health, discussed EGFR inhibition options for patients with nonsmall cell lung cancer (NSCLC), based on the case of a real patient.
Targeted Oncology™: What are the potential next steps for evaluating this patient for systemic therapy?
PATEL: [We can assess] 7 NCCN [National Comprehensive Cancer Network]recommended biomarkers in this patient. Having a CNS [central nervous system] specialist and a radiation oncologist involved in the case to handle the CNS [would be wise]. Getting the patient started on the next therapy is key. Waiting for NGS [next-generation sequencing] results is also something that’s key here, so we have multiple options.
When we’re thinking about NGS, what genes should we test for in nonsquamous NSCLC in the metastatic setting?What are some of the challenges with these tests?
At the very least, [we should test for]EGFR,ALK,ROS1,BRAF, andNTRK. This is most ideally done with a multiplex-based assay. This can be done by tissue testing and by cell free DNA [cfDNA]. One of the things you often find in newly diagnosed patients with NSCLC is that there’s little tissue to begin with. These are often EBUS [endobronchial ultrasound]guided biopsies that are small, whereas in a CT-guided biopsy, we’re trying to avoid pneumothorax. In those scenarios, when they’re making the initial diagnosis of cancer…they’ll usually go through multiple [rounds of] immunohistochemistry [IHC], as well, and there’s not a lot of tissue left for molecular testing.
In fact, what I do in my clinic for these patients is order tissue-based NGS and the cfDNA at the same time up front, not with the intent of having duplicative testing done but with the idea that the cfDNA testing will come back in about a week usually, before pathology has even cut and shipped the slides. If you have your driver mutation answer in cfDNA, you know you can cancel the tissue-based NGS because you already have your answer; but if you don’t, you’re at least 1 week ahead of where you otherwise would have been. That’s something I’ve had to adopt in my workflow because even once you get the molecular answer, you still have to get the drug. In an era of high-priced drugs with high co-pays for patients, it can take some time to get started, especially for a patient with brain metastasis.
The other point is that PD-L1 expression testing should be performed in all patients, and an assay [using the 22C3 antibody] is preferred. The idea that we need to look at [not only] the tissue for PD-L1 expression, which is the only place we can get PD-L1 IHC that’s a protein-based marker on the cell surface, [but also] genomic-based markers, which can be done either in tissue or by cfDNA in a liquid biopsy, is the key concept here in terms of what to do up front.
The other point is that the results come back asynchronously. Your PD-L1 IHC will come back instantly. It may be back even before you’ve met the patient. Your NGS is likely [processing] for 2 or 3 weeks, and this is something you may have to wait for.
How often do you receive quantity-not-sufficient (QNS) test results in your practice? What percentage of the time would you say this occurs?
[On average,] about 20% to 30% of patients with NSCLC have their initial biopsy [come back as QNS] because after all these procedures, the biopsy is often too small [for a result]. That’s why I have the cfDNA workflow simultaneously, and then I’m able to cancel the tissue-based sequencing about 70% of the time.
What can a medical oncologist do to help ensure sufficient tissue quantity?
It’s difficult because we’re often the last [practitioners] to see the patient. It used to be…[that] if you [were] looking for T790M or a specific molecular aberration, you could tell the interventional radiologist and pathologist [to not perform] IHC. I already know they have cancer, and I just need the molecular testing. Nowadays in the up-front setting, you’re seeing the patient after all of that is done.
Do you ever use the liquid biopsy in situations in which tissue is delayed?
I think it’s something that has taken on a meaningful role in the care of our patients. It has both aspects of getting an answer quickly and [utility] for patients who may have had a resection 2 or 3 years ago and now they’ve relapsed in the brain or bone where you’re not going to get access to that tissue. The ability to get an answer about the cancer in its current state is quite reasonable. [Cell free DNA] in NSCLC for stage IV disease will pick up a molecular driver in about 85% of cases. About 10% to 20% of patients have nonshedding tumors. Their tumors don’t shed a lot of DNA. In those patients, a tissue-based biopsy is likely still necessary.CASE (continued)
What are the therapeutic options for this patient?
This patient with a high PD-L1 score, to everyone’s surprise, had anEGFRexon 19 mutation. About half of patients who are considered PD-L1 high will have a driver mutation. It’s a bit of a red herring. The PD-L1 score comes back 2 weeks, sometimes 3 weeks, ahead of NGS. It’s a common conundrum we face in our clinic.
For this patient with anEGFRexon 19 mutation, we can start targeted therapy with osimertinib [Tagrisso]. It’s a third-generation EGFR inhibitor with excellent CNS penetration, but it’s equally reasonable to reach for SRS [stereotactic radiosurgery]. This patient has symptomatic CNS disease to a solitary lesion, and having that local control may be helpful.
There’s equipoise in the field on what to do here. There’s no right answer, no wrong answer, but I think using a CNS-penetrant tyrosine kinase inhibitor [TKI] like osimertinib moves the needle in favor of the patient, but you can consider SRS.
Typically in my clinic, [I’ll] see him [in conjunction with a] CNS medical oncologist. I’ll test him with steroids to see whether his symptoms resolve, and then I have time to give them osimertinib and time [for the drug to work]. The best-case scenario is that it resolves, and they don’t need any radiation. Worst-case scenario is often it’ll stay the same or maybe get a little bigger, and then they can get SRS to a resistant lesion.
What are the efficacy data associated with the approval of osimertinib in this patient population?
[Results of] the phase III FLAURA trial [NCT02296125] showed an overall survival [OS] advantage with osimertinib regardless of CNS status in terms of metastatic disease [in patients with treatment-naïve NSCLC adenocarcinoma withEGFRexon 19 or 21 mutations].1This is the study for first-line use of osimertinib versus the standard of care. These are all patients who hadEGFRexon 19 or 21 mutations. It represents about 90% of patients who haveEGFRmutations and an exon 19 or 21 mutation. [In California,] patients withEGFRmutations [account for] about 20% to 30% of patients, especially never smokers and predominately Asian woman, though you can definitely have patients who are light smokers, and you can have men with this mutation.
In this trial, osimertinib, a third-generation EGFR TKI, was compared with a erlotinib [Tarceva] or gefitinib [Iressa], [each of] which is a first-generation EGFR TKI. The comparator is not chemotherapy. The comparator is a good standard-of-care regimen up to even 2 years ago. The primary end point of the trial was progression-free survival [PFS].
Early on, osimertinib showed a separation of the Kaplan-Meier curve [for PFS], and there’s no initial crossover. The patients who got osimertinib have better PFS compared with another targeted therapy option [18.9 vs 10.2 months; HR, 0.46; 95% CI, 0.37-0.57;P<.0001].
If the use of osimertinib has an OS advantage, does that mean clinicians should use this agent in the front line, or should it be sequenced after first-generation TKIs?
I could sequence the drugs if there is an OS benefit…because [the results] indicate a durable advantage. For OS, the hazard ratio was 0.63 against another TKI. To me, that’s clinically active and quite substantial. Even at 3 years out, and even in patients with brain metastases, there is an advantage [for OS] as long as you find the right driver mutation for these patients. Even patients on the standard-of-care arm almost had a 50% survival rate at 3 years, showing robust efficacy.
Patients who progressed on a first-generation EGFR TKI were able to cross over to osimertinib on the study, so that was allowed for the [population with]EGFRT790M [mutations], but only about a third of patients crossed over. This is one of the issues and one of the mantras I have: You always want to use your best therapy first, as long as it’s appropriate, because you may never get a second chance to make a first impression on the cancer. Here you have a group of patients who are on a clinical trial, [meaning either they are] more fit than the average patient you take care of and/or they’re watched more often than the average patient you take care of in clinic in terms of MRIs and CTs that are the mandated protocol. Despite that level of detail, only a third of patients crossed over to get osimertinib as second-line therapy. Many of us switch to osimertinib for this reason because we want to make sure patients have an opportunity to have this prolonged survival.
If you look at patients with CNS metastases versus those without, how does that affect the outcomes with osimertinib?
A reasonable number of patients in the FLAURA study had CNS metastasis at baseline. You see that osimertinib, regardless of [patients’] CNS status, had a similar hazard ratio [for PFS] of about 0.46 and 0.47, which is quite substantial. Of course, osimertinib in and of itself should be able to control CNS disease. The first-generation EGFR TKIs, which are in the comparator arm, erlotinib and gefitinib, would not.
What did the CNS responses look like for these patients?
Of the patients treated on FLAURA who received osimertinib, the vast majority had complete responses, and several had partial responses. Of course, 2 patients [did not respond]; one had stable disease, and one had some growth. Typically, if I have a patient with CNS disease who was not doing SRS up frontwhich is somewhat more common nowadays, in the sense that these patients are living longer and some of the longer-term sequelae of SRS are trying to get better in terms of radiation necrosis—I get MRI 1 month after therapy…which lets us go straight to SRS in case this is one of the few patients who is not going to have shrinkage in terms of their initial CNS response. If you are going to avoid using local therapy to the brain, make sure to watch those patients carefully. I typically don’t wait longer than a month to get that initial scan.
Objective response rates for those who were evaluable for response are quite high. The vast majority of patients will have CNS control.
What do you think about using the IMpower150 (NCT02366143) regimen in a patient such as this?
This is a regimen of carboplatin, paclitaxel, bevacizumab [Avastin], and atezolizumab [Tecentriq]atezolizumab being a PD-L1 inhibitor—versus conventional chemotherapy. It’s a bit of a complicated 3-arm study. Arm A is atezolizumab with carboplatin and paclitaxel and atezolizumab. Then you have the 4-drug combination in arm B. In arm C is the control of carboplatin, paclitaxel, and bevacizumab.
These are patients with nonsquamous NSCLC who either areEGFRorALKnegative or have received appropriate EGFR or ALK therapy previously and now need a chemotherapeutic option. That’s what is unique about this study. Patients who had prior EGFR and ALK inhibitors were allowed on this study. The randomization was 1:1:1 across these 3 arms.
There’s a survival benefit with a hazard ratio of about 0.76 [for arm B versus arm C]. One of the caveats of this study is that paclitaxel, which is the chemotherapy backbone for IMpower150, is not used as commonly in nonsquamous NSCLC. Typically, many of us use pemetrexed because it has a significantly improved tolerability profile relative to taxanes.
Patients with liver metastases and those withEGFRandALKmutations, though there are some caveats to this analysis, did better with the 4-drug combination featuring bevacizumab and atezolizumab with chemotherapy compared with chemotherapy and bevacizumab alone.
What does the efficacy of this regimen look like across PD-L1 strata?
In patients who had tumors with high PD-L1 expression, the hazard ratio for OS was 0.70. In patients with PD-L1low tumors and those who were negative, there was benefit, though these hazard ratios all crossed 1 in these subgroup analyses. What you see here is that although there may be a hint more benefit, it’s a wide confidence interval, so there is benefit to all-comers, but it’s really not dichotomized based on the PD-L1 stratus.
I think the most important point…about the IMpower150 data is that if you have a patient with anEGFRmutation or anALKrearrangement, that patient is best served by targeted therapy first. Exhausting all their targeted therapy options forALK, we arguably have 2 to 3 different next-generation options before switching to one of these regimens.
Would you ever initiate immunotherapy in a patient before their mutation testing returns?
This goes back to the excellent question about PD-L1 status. The driver mutations that we typically test for in NSCLC [have concurrent expression of] PD-L1 >50% at high rates. ForEGFR, about 29% have high PD-L1 expression, 40% inALK, 56% inBRAF, 60% inROS1, 50% inRET, and almost 50% inMET. PD-L1 can become a bit of a red herring because it shows up first. You get that result initially because it’s an IHC test. It takes a day to come back. NGS takes 2 to 3 weeks. It’s reasonable to think of starting them on immunotherapy, especially for a patient who is in extremis, but if you wait for the NGS, you’ll find that this patient has an oncogene-addicted tumor for which their best therapy may be targeted therapy.2
This is important because of this study and some of the related data that show the pneumonitis risk happens if you start someone on an immune checkpoint blockade agent first and then switch to an EGFR inhibitor. Now, what’s interesting is that if you start the TKI and then go to immunotherapy, there tends not to be a big pneumonitis risk, for reasons that aren’t clear. If you start with an immunologic agent first and then switch the EGFR-targeting TKI, you’re more likely to have pneumonitis.
What factors do you consider when choosing an EGFR-targeted TKI?
Many clinicians use osimertinib, and this is [in] the NCCN guidelines.3It’s a sensitizing mutation, and if it’s discovered prior to first-line therapy, the options are erlotinib, afatinib [Gilotrif], gefitinib, dacomitinib [Vizimpro], erlotinib, [and] ramucirumab [Cyramza].
If they’re on a chemotherapy-alone regimen [when anEGFRmutation is discovered], you can typically take them to their first scans. I typically do 2 doses of chemotherapy and get a scan to see how they’re doing. If they’re doing well, you can continue it and then just plan on giving them osimertinib when they have progression.