Mark Socinski, MD, explained to a group of physicians at a recent <em>Targeted Oncology </em>live case-based peer perspectives presentation the treatment considerations and decisions he makes when seeing patients with non–small cell lung cancer in the clinic.
Mark Socinski, MD
Mark Socinski, MD, explained to a group of physicians at a recentTargeted Oncologylive case-based peer perspectives presentation, the treatment considerations and decisions he makes when seeing patients with nonsmall cell lung cancer (NSCLC) in the clinic. Socinski, the executive medical director of AdventHealth Cancer Institute, reviewed treatment options based on the case scenarios of 2 patients with advanced nonsquamous NSCLC.
A 63-year-old man presented to his primary care physician with intermittent cough and difficulty breathing on exertion. His medical history included hyperlipidemia that was well managed on simvastatin; hypothyroidism, which was managed on levothyroxine; and chronic obstructive pulmonary disease, which was managed on inhalers. He recently quit smoking but had a 40-pack-year smoking history. On a physical exam he had intermittent wheezing and had an ECOG performance status (PS) of 1.
His creatinine clearance was within normal limits. A chest CT revealed a 3.1-cm spiculated mass in the right upper lobe and 2 enlarged right mediastinal lymph nodes measuring 1.5 cm and 1.7 cm. Moderate emphysema was also noted. A PET scan confirmed the lung lesion and showed mediastinal lymphadenopathy without evidence of distant metastases. A brain MRI was negative.
His Pulmonary function test showed a forced expiratory volume in 1 second of 1.2 L and a diffuse capacity of the lungs for carbon monoxide of 35%. A bronchoscopy with transbronchial lung biopsy and lymph node sampling revealed adenocarcinoma, with positive nodes in stations 4R and 7; level 4L was negative. He was diagnosed with NSCLC, stage IIIA, T2aN2M0.
Based on the extent of mediastinal disease and emphysema, the patient’s cancer was deemedinoperable, and he was referred for consideration of concurrent chemotherapy and radiation.
Would you order genetic testing for this patient with locally advanced lung cancer?
I do not [perform] genomic testing for the purposes of finding a driver in patients like this. Right now, the data only support acting on that in stage IV disease. There are some ongoing trials in the adjuvant surgical setting, as well as in the chemoradiotherapy setting, trying to figure that out. The data that we have to date do not suggest that we really know how to integrate if outcomes are improved by using targeted therapy in these patients. You can make the argument that the chance that he is going to be cured is not 100%, the chance that he is going to relapse with stage IV disease in the next 1 to 2 years is high. So, we often do test these people. Sometimes if you find anEGFRmutation, it is tempting to think about what you might do differently; but the data would not support any of that.
We were very careful to point out that he had multi-stations, 4R and 7 on endobronchial ultrasound, that were positive. To me, that is the tip of the iceberg; there are probably other stations involved at least microscopically.
How would you treat this patient with that in mind?
Either carboplatin plus paclitaxel weekly or cisplatin plus etoposide is perfectly fine to use. In the United States, the last time I saw a poll of a large audience about what chemotherapy regimens are used during treatment, 47% said weekly carboplatin/paclitaxel, 47% said cisplatin/etoposide, and 6% said other regimens. I think you can go either way with these sorts of things. And it is not clear that one is better than the other. There has never been a randomized trial to address that question, but in a very famous [Veterans Health Administration] case cohort study, the outcomes were identical and weekly carboplatin/paclitaxel was less toxic.1
If this patient had stage IV disease, would you treat him systemically with weekly carboplatin/paclitaxel?
This is what bothers me with a regimen of weekly carboplatin and paclitaxel with radiation. The real reason you are giving it is to enhance the effect of radiotherapy. I am uncomfortable with that being the only systemic chemotherapy that patients get, so I give an induction therapy regimen. I give 2 cycles of carboplatin plus paclitaxel before chemoradiotherapy. I think that, depending upon how you read the literature, that is perfectly appropriate.
How do you determine which treatment strategy you are going to use?
These are what the National Comprehensive Cancer Network [NCCN] lists as reasonable regimens to use [for neoadjuvant/adjuvant chemotherapy]2:
Chemotherapy Regimens for Neoadjuvant/Adjuvant Therapy category 2A
Cisplatin50 mg/m2days 1, 8;vinorelbine25 mg/m2days 1, 8, 15, 22; every 28 days x 4
Cisplatin100 mg/m, day 1;vinorelbine30 mg/m2days 1, 8, 15, 22; every 28 days x 4
Cisplatin75-80 mg/m2day 1;vinorelbine25-30 mg/m2days 1, 8; every 21 days x 4
Cisplatin100 mg/m2day 1;etoposide100 mg/m2days 1-3; every 28 days x 4
Cisplatin75 mg/m2day 1;gemcitabine1250 mg/m2days 1, 8; every 21 days x 4
Cisplatin75 mg/m2day 1;docetaxel75 mg/m2day 1; every 21 days x 4
Cisplatin75 mg/m2day 1;pemetrexed500 mg/m2day 1 for nonsquamous; every 21 days x 4
None of them have been shown to be superior to anything else. I can tell you, there’s a study called the PROCLAIM trial that looked at cisplatin plus etoposide versus cisplatin plus pemetrexed in this setting.3They were completely identical, except the course of cisplatin and etoposide was about $1000 in drug acquisition costs. To give cisplatin and pemetrexed was about $50,000 for the course of care the way they gave it. You cannot justify that when the hazard ratio was 1.0 and the curves were completely superimposed, and it was not necessarily less toxic.
These are the concurrent chemoradiotherapy regimens. This is what the NCCN calls it, realizing that there is no phase III data for carboplatin and pemetrexed.
Concurrent Chemotherapy/Radiotherapy Regimens category 2A
Cisplatin50 mg/m2days 1, 8, 29, 36;etoposide50 mg/m2days 1-5, 29-33; concurrent thoracic radiotherapy
Cisplatin100 mg/m2day 1, 29;vinblastine5 mg/m2/weekly x 5; concurrent thoracic radiotherapy
CarboplatinAUC 5, day 1;pemetrexed500 mg/m2day 1; every 3 weeks x 4; concurrent thoracic radiotherapy (nonsquamous)
Cisplatin75 mg/m2day 1;pemetrexed500 mg/m2day 1 every 3 weeks x 3; concurrent thoracic radiotherapy (nonsquamous) ± 4 cycles pemetrexed 500 mg/m2
Paclitaxel45-50 mg/m2weekly;carboplatinArea under the curve (AUC) 2; concurrent thoracic radiotherapy ± 2 cycles every 3 weeks paclitaxel 200 mg/m2+ carboplatin AUC 6
Many times, the NCCN gets it right, but remember, the NCCN is an opinion-based guideline. It is a guideline, it is not a pathway, and they get it mostly right. But sometimes they do these things, like put in a phase II regimen in this setting. All of the others are phase III tested.
Case 1 (continued):
The patient underwent therapy with cisplatin plus etoposide and concurrent thoracic radiotherapy. Follow-up imaging showed a partial response, with shrinkage of the primary and nodal lesions.
How have recent data changed how this patient would be treated?
The PACIFIC trial was designed [for] the type of patient we are talking about that got concurrent platinum-based chemotherapy; then within 6 weeks, these patients were randomized to placebo, which was the standard of care, versus a year of durvalumab [Imfinzi].4Durvalumab is an antiPD-L1 monoclonal antibody, and in this trial, it was given every 2 weeks for a year. Patients had to be randomized within 6 weeks, and you had to document that the patient had no evidence of disease progression. Now formerly, I would have waited 2 months before I would do a CT scan. This trial changed my practice because we are supposed to give it by 6 weeks, and in fact, in this dataset, the data suggest that the sooner you give it after chemoradiotherapy, the better it works.
Depending on how the patient does with chemoradiotherapy, if they have a lot of esophagitis, I will time my CT scan around when they are getting better and they can go on to get immunotherapy. If they have very little esophagitis or very little other toxicity, I will do a scan 2 weeks later and start immunotherapy right away. If they need more time to recover, I will do it at 4 weeks.
[With the results of the PACIFIC trial]I love phase III trials in which I do not need a statistician to tell me which treatment is better—the hazard ratio [for progression-free survival (PFS)] was 0.51, and that is a 50% reduction in the risk of disease progression. This is the time of PFS from the end of chemoradiotherapy. What we do not know from this study is how many people got chemoradiotherapy and never got to this randomization. You had to register patients at this point. So there may be patients that either had disease progression, their toxicity does not resolve, and their PS is not 0 to 1 by 6 weeks, so they would have never gone on this drug—but that is probably the minority of patients.
This was published initially by Scott Antonia, MD, PhD, who used to be at Moffitt Cancer Center and who is now at Duke Cancer Institute, in theNew England Journal of Medicine.4Looking at the various subsets, patients withEGFRmutations were included in this trial, but there were very small numbersonly 6% of the patient population—but they were not excluded from the trial. No matter what baseline demographic you had, all of this favored the use of durvalumab.
What we learned last September at the 19th World Conference on Lung Cancer in Toronto, Canada, was the overall survival [OS] data. This was again published in theNew England Journal of Medicine.5This is a trial where you do not need a statistician to tell you that the hazard ratio [for OS] is 0.68, and that amounts to about a third in the reduction of the risk of death over time.
In my opinion, this is a little premature. I am interested in 4- and 5-year survival, because to me that translates into a cure. I viewed stage III disease as potentially curable with chemoradiotherapy, although we probably only cured 1 out of 4 patients. The interesting thing is what is this going to be at the end of 4 to 5 years. The curves are only out to 3 years or so.
The other thing that is remarkable is that the median survival on the control arm was actually 28.7 months. That is actually pretty good for the control arm, which was placebo. Now remember, these are patients [who] got chemoradiotherapy, did not progress, had a PS of 0 to 1 within 6 weeks after radiotherapy and then went on. So it is not all the patients who get chemoradiotherapy, but it is probably a majority of them. The median survival on the investigational arm, the durvalumab arm, has not yet been reached.
Is PD-L1 status important in this setting?
PD-L1 status was not a requirement for being on the trial, but they did look at it. From a PFS point of view, all of it was as we have seen in almost every immunotherapy trial to date. The stronger you stain for PD-L1, the greater the benefit is from the immunotherapy.
What was a little complicated is that in the United States, if you look at the package insert for durvalumab, the FDA did not use language about PD-L1 status. The European Medicines Association [EMA] asked for an unplanned retrospective analysis based on PD-L1, and in the PD-L1negative patients, placebo looks a little better. Now with the EMA label, patients must be PD-L1 positive [to receive durvalumab], but that is based on a retrospective, unplanned analysis of this trial, which we are all taught not to do. But when a regulatory agency says we want to see this data, this is what you come up with. In the United States, there is no language about PD-L1 status; in Europe, you have to be PD-L1 positive to get durvalumab as consolidation therapy.
I am very reluctant to buy into the PD-L1 analysis that was unplanned and retrospective. It may be giving us a false message. We know in stage IV disease that the combination of chemotherapy and immunotherapy has benefit in PD-L1negative patients, so I am not willing to write that off at this point.
How long are you willing to wait after radiotherapy before starting immunotherapy?
The trial said 6 weeks. When they first opened the trial, you had to be randomized within 2 weeks of completing radiotherapy. No one was accruing to the trial, so they changed it to 6 weeks. If you are ≤14 days [from last radiation before randomization], the hazard ratio is 0.39; if you are >14 days, the hazard ratio is 0.63, almost a doubling of the benefit if you get it earlier.6So this is why I try to do it as early as possible, because it is great if you can do it early. You do have to get that CT scan, theoretically, so if the insurance company calls…you can confirm that the patient has not had progressive disease. Although the CT is not going to look fine, it is going to show radiation changes and it is going to be confusing, but I cannot help that. You just have to say in your note: “I reviewed the CT scan and there is no evidence of disease progression; there is a lot of radiation changes.’ “
If you talk to [the lead investigator on the PACIFIC trial], Scott Antonia, he creates an argument that you have a window to light the immune responsefor lack of a better phrase—and that window is immediately after radiation finishes. I am not enough of an immunologist to defend that and tell you why, but Scott believes that sooner is better, and that is why this was looked at. Is it just those patients [who] are robust patients? They have no esophagitis, they do well, they get through. Is it better patients or better treatment? It is hard to know. All things being equal, I try to get patients on immunotherapy as quickly as possible.
What was the safety profile seen with this regimen?
The toxicity profile was not bad. Immune-related toxicities were fine. I think the one thing that many of us were concerned about in this trial was that you were giving an agent that causes pneumonitis after a modality that also causes pneumonitis, and would that cause an increase in pneumonitis? Results showed that there was a slight increase in grade 1/2 pneumonitis, but I think the important thing is grade 3/4, which occurred in about 4% of patients in both arms. Discontinuation rates were 4% to 6% on both arms, so I think there was little heightened concern after these data were known about the regimen.
So now the NCCN guidelines have durvalumab as a category 1 agent in patients who get chemoradiotherapy and have no evidence of disease progression. They get consolidation therapy with durvalumab in this particular setting, so this is kind of a new standard of care in this setting.
Discuss the role of chemotherapy induction and consolidation therapy in patients who will receive radiotherapy.
One of the surprising things to me, in talking to a lot of oncologists, was this belief that consolidation chemotherapy had a great benefit. There has never been a randomized trial that said giving 2 cycles of chemotherapy afterwards has a benefit. Consolidation chemotherapy, which we all used to do, should only be given to those patients who have a contraindication to immunotherapy. Twenty-eight percent of the patients on the PACIFIC trial got induction therapy. I am a fan of induction therapy, and the reason is 3-fold:
1) It takes the pressure off the radiation oncologist. We see these patients [who] hope to get in to see the radiation oncologist the same day. The patient wants to get started…but the physicist needs to get a plan ready, plan the CT, and get the physics done. All that takes time. If you give 2 cycles of chemotherapy, they have 6 weeks to get this done.
2) There is always some degree of tumor reduction with 2 cycles of chemotherapythe tumor volume shrinks. I spent 16 years at the University of North Carolina and did about 5 or 6 investigator-initiated trials with induction therapy, and one of the things that we showed in that cumulative experience was that the volume of tumor the patient had when they started radiation mattered most for long-term survival or cure. We know that radiation is a volume-dependent exercise. If you think you are treating microscopic disease in the adjuvant setting, they give a lower dose than the stage III setting where you have a bulky tumor. You know that radiation is more likely to treat a smaller tumor. Intuitively, that makes sense. I like that the volume reduction may allow the radiation oncologist to treat a smaller tumor. Remember, the toxicity of radiation is not related to the total dose of radiation, so you can give a very high dose; that is stereotactic body radiation therapy. The toxicity of radiotherapy is related to an excessive dose to a normal organ. At what radiation dose is the lung essentially dead? [That dose is] 25 Gy. So if you are giving 60 to 66 Gy, whatever part of the lung got over 25 Gy is no longer exchanging gas and the surgeon might as well have resected it. That is why radiation oncologists want to treat smaller fields.
3) By giving induction therapy, you have the option to shrink it. We often do this in small cell [lung cancer]. If they have a big tumor and the radiation oncologist says give them a cycle of chemotherapy, the tumor shrinks, they can treat a smaller volume, and there is less normal tissue in the field. Those are the 3 reasons why I like induction therapy.
Now that there may be more pressure after radiation to put patients on immunotherapy earlier, how do you handle that?
In a placebo-controlled trial, there was no real difference in grade 3/4 pneumonitis between placebo and durvalumab, so I think it is a problem that we are not going to have. I think there is going to be the occasional patient where you have to make a judgment call. If the patient has severe radiation pneumonitis, chances are they did not have a PS 0 to 1 and they would not have gone on this trial.
I guess the question we are trying to get at is: If durvalumab is supposed to be administered within 6 weeksthat is what the trial did—would you give it at 10 weeks? Let us say you had a patient [who] had pneumonitis, you had to give them steroids, and you weened them off. Then they had esophagitis that got complicated by hepatitis and went on for 4 or 5 weeks, but the patient is slowly getting better. Would you give it at 10 weeks or 12 weeks? After 12 weeks, I think it might be too long. Maybe the patient should not get consolidation immunotherapy in this setting.
I always say stage III is a team sport. I think every oncologist should know from the radiation oncologist what is the volume of lung getting radiation and how much of a risk is radiation pneumonitis going to be so you can prepare your management strategy for these sorts of things. It is not a perfect science, but if the volume of lung getting more than 20 Gy is >35%, the risk of radiation pneumonitis goes up.
How comfortable are you giving steroids for toxicity with immunotherapy?
Very comfortable. I think that some of the mistakes we make with managing patients on immunotherapy is we do not give them enough steroids for toxicity. We are concerned that maybe the steroid will blunt the effect of the immunotherapy, so we have been conservative. I am very aggressive and get the toxicity under control.
There are some data to suggest…maybe the toxicity that you are seeing in normal ordinance reflects the toxicity that the tumor is feeling in terms of immune response to the tumor. My advice is to be aggressive with steroids for toxicity, get it under control, and keep it a grade ≤2 and do not worry about the immunotherapy.
A 59-year-old Caucasian man presented with symptoms of chest pain, cough, and dyspnea. He had a medical history of hypertension that was managed on a calcium channel blocker and osteoarthritis. He was also a former smoker with a 10-pack-year smoking history.
A CT of the chest and abdomen revealed a 9.0-cm spiculated mass in left lower lobe, a loculated pleural effusion in left hemithorax, diffuse liver nodules, and right adrenal metastases. A PET/CT scanshowed18fludeoxyglucoseuptake in left lung mass, pleura, liver, and right adrenal gland. A brain MRI was negative for intracranial metastases.
A physical exam was notable for decreased breath sounds in the left lung base and no hepatomegaly. He had an ECOG PS of 1. Image-guided biopsy of the liver lesion showed a poorly differentiated adenocarcinoma of the lung.
Molecular testing was performed, including next-generation sequencing (NGS), which was negative for molecular aberrations inEGFR, ROS1, BRAF, andALK.Immunohistochemistry showed that he had PD-L1 expression in 2% of cells.
What are your initial impressions of this case?
Why this patient needed a PET scan, I do not know, because you had a CT scan that showed liver metastases. So you already knew he had stage IV disease. Brain MRI was negative for metastases. A biopsy of the liver [was performed] so you can prove it is stage IV disease. In my opinion, this is the right thing to do.
PET is a great tool, but you use that tool for the job that you need to define. If you already know they have stage IV disease, why are you doing the PET scan? A bone scan is much cheaper.
NGS is negative for the 4 [genetic aberrations outlined by the NCCN for testing], and PD-L1 expression is 2%.
What data impact the treatment options for this patient?
KEYNOTE-189 took everyone’s favorite doublet, carboplatin and pemetrexed [Alimta], and added pembrolizumab [Keytruda].7What we did in this trial was take first-line therapy and second-line therapy and give it at the same time. And this was another a trial where I do not need a statistician to tell me what is better. [The] hazard ratio [for OS] was 0.49 here, [which represents a] 50% reduction in death. Every baseline subgroup analysis favored pembrolizumab, including those with brain metastases. For PD-L1 expression, as I said before, the stronger you stain, the greater the benefit. But even if you don’t stain, or there is less than 1%, there is still a benefit. Not as great a benefit, but still there.
For PFS, there is the same impact. The hazard ratio is about the same at 0.52. In the negative expressers, there is not quite the same benefit; the hazard ratio still 0.75, but the confidence intervals still overlap 1. But we all care about OS more than PFS.
Response rate more than doubled with pembrolizumab, 18.9% versus 47.6%. And then like we have seen in almost all the immunotherapy trials, the duration of response is much longer for immunotherapy versus chemotherapy. Response by PD-L1 status is still higher for PD-L1 expressers.
But what if the patient is eligible for bevacizumab (Avastin)?
This trial, [the IMpower150 trial] would justify it. This trial that essentially took patients who were eligible for bevacizumab and used [them] as the control armarm C, [which was based on the regimen from] ECOG 4599 of carboplatin, paclitaxel, and bevacizumab—an FDA-approved regimen.8And then 2 different strategies were tested: an add-on strategy, so a 4-drug regimen with atezolizumab [Tecentriq], or arm B, and then a substitution strategy [to] replace bevacizumab with atezolizumab. The OS favored arm B versus arm C, with a hazard ratio of 0.78, so a 22% reduction.
One of my pet peeves about this is that it is very incorrect to make crosstrial comparisons between KEYNOTE-189 and IMpower150. In IMpower150, the patients had to be eligible for bevacizumaband on a clinical trial, there are lots of reasons not to give bevacizumab—so these patients are different patients than those on KEYNOTE-189. In KEYNOTE-189, you had to be eligible to get carboplatin and pemetrexed. There are some bevacizumab patients in there, but there were a lot of patients in KEYNOTE-189 [who] would not have been eligible for IMpower150. So there is a significant benefit to OS in that patient population.
This was the only trial in whichEGFRmutations andALK-positive patients were allowed to be in the cohort. The wild-type OS, with patients withoutEGFRorALKissues, [was 19.2 months with atezolizumab versus 14.7 months without]. All those patients in that setting [had a median OS of 19.8 months with atezolizumab versus 14.9 months without]. There were 2 special key subgroups we looked at: liver metastases andEGFR/ALKmutations. If you had anEGFRmutation, you actually did pretty well with chemotherapy, and these wereEGFR-mutant patients who had received prior tyrosine kinase inhibitors [TKIs] and progressed. They received the standard of care if they had a sensitizing mutation. We knew from ECOG 4599 that bevacizumab with chemotherapy works better in patients with liver metastases than chemotherapy alone. This actually shows that if you combine immunotherapy with bevacizumab, then patients with liver metastases benefit more than from bevacizumab alone. There is a little clinical vignette here in that regard.
There is one issue here: The Genentech-Roche antibody is a little less sensitive than all the other antibodies, so it is hard to compare because they used the Genentech-Roche antibody for this trial.
What were the outcomes for patients who received atezolizumab in place of bevacizumab in the standard regimen?
Now, there was no difference in outcomes when comparing it with the substitution strategy. If you compared carboplatin, paclitaxel, and bevacizumab with carboplatin, paclitaxel, and atezolizumab [arm A], the hazard ratio is 0.88 with aPvalue of 0.2not significantly different. Maybe there is a little bit of separation late, but from a log-rank, Kaplain-Meier point of view, it was not statistically significant.
When you look in the patients with liver metastases, again, it was the interplay between anti-VEGF therapy and antiPD-L1 therapy that made the difference. That was arm C versus arm B. If you look at bevacizumab versus atezolizumab, there was no difference. And the same is true of theEGFR-mutant population. It was really this anti-VEGF, antiPD-L1 effect in theEGFR-mutant patients. These were patients [who] had failed TKIs versus the arm C where there was no difference between the substitution drug.
People talk about how VEGF is very immunosuppressive on multiple parts of the immune system. It turns out that patients withEGFR-mutant tumors with liver metastases are exactly those patients [who] have high VEGF states. When you combine anti-VEGF with antiPD-L1, you see the benefit here in the patients withEGFRmutations and liver metastases. It is something to think about in terms of selecting regimens.
Discuss the results of the IMpower130 trial. Do you think this regimen has a place in the setting of first-line nonsquamous NSCLC?
Here is a trial in nonsquamous [NSCLC with] carboplatin plus nab-paclitaxel [Abraxane]. Remember, that is an FDA-approved regimen for nonsquamous disease.9In my opinion, there is not a shred of evidence that pemetrexed is a better drug than a taxane in lung cancer in general, whether it is squamous or nonsquamous disease, [and this trial employed a pemetrexed maintenance cycle of the control arm following induction treatment with chemotherapy]. Well, in squamous, we know taxanes are better. There was a benefit in this trial from adding atezolizumab to carboplatin and nab-paclitaxel for OS.
Now, occasionally you have those patients who have an estimated glomerular filtration rate of 35 mL/min. Pemetrexed is theoretically contraindicated in those patients because it is more toxic. Nab-paclitaxel is not. There are a lot of studies with nab-paclitaxel showing that patients with borderline renal function do not have any increased toxicity; it is not true with pemetrexed.
I think IMpower130 does contribute a little bit to it, although it is not commonly used in the nonsquamous population.
If the patient experiences disease progression, what would your choice be for a second-line therapy?
There were some data from the CheckMate 227 trial to suggest that nivolumab [Opdivo] and ipilimumab [Yervoy] did better than chemotherapy in patients with high tumor mutational burden [TMB] for PFS.10The overall response rate was better, too, in those patients who had ≥10 mutations per megabase, if you are comparing nivolumab and ipilimumab to chemotherapy. Of course, that is not comparing it with chemotherapy plus immunotherapy, and this is only PFS.
As you know, Bristol-Myers Squibb withdrew the FDA application because the survival is no different, so there is a disconnect between PFS and OS.11
Right now, based on CheckMate 227, I do not know how to use TMB and I do not know where to use nivolumab plus ipilimumab in this setting. We are still waiting on more data to convince us one way or the other.
Now the plot thickens because we have atezolizumab’s answer to KEYNOTE-189. KEYNOTE-189 was carboplatin and pemetrexed, plus or minus pembrolizumab; [the IMpower132 trial is of] carboplatin or cisplatin and pemetrexed, plus or minus atezolizumab.12And like everything in life, things get messed up. The PFS is positive; the OS was not. For OS, .0797 is thePvalue, and the hazard ratio is 0.81. It was not the 0.50 we saw in KEYNOTE-189, but again, 2 trials with carboplatin and pemetrexed plus or minus immunotherapy, one positive with pembrolizumab, and one kind of negative with atezolizumab.