Chemoimmunotherapy Transforms the Lung Cancer Armamentarium

Vignesh Narayanan, MD

Chemoimmunotherapy has truly changed the game in the treatment of patients with lung cancer, according Vignesh Narayanan, MD. However, he cautioned that the approach is not necessarily appropriate in an all-comer population. 

“If [a patient’s] PD-L1 expression is greater than 50%, single-agent immunotherapy, a well-established therapy, is the go-to option. There is not yet enough data to support this therapy when [a patient’s] PD-L1 expression is under 50%, thus, it must not be used in that setting, Narayanan said. “Further, in clinical practice, it’s tempting to choose single-agent immunotherapy for patients with a borderline ECOG performance status of 2; however, we must keep in mind that this ECOG score was not included in [the trials]. As such, there isn’t any data to support [single-agent immunotherapy in that population].”

In an interview with OncLive® during the 2020 Institutional Perspectives in Cancer webinar on lung cancer, Narayanan, a medical oncologist at Kaiser Permanente, spotlighted the growing role of chemoimmunotherapy in lung cancer, along with ongoing research in the space.

OncLive®: Could you highlight some key therapies in the NSCLC armamentarium?

Narayanan: There is an array of frontline treatment options in NSCLC, especially for patients who do not have a driver mutation. There is single-agent immunotherapy, immunotherapy in combination with a CTLA-4 antibody plus an anti–PD-L1 antagonist. Finally, there is the combination of chemotherapy and immunotherapy. Furthermore, there are also some exciting ongoing studies evaluating biomarkers.

In your own practice, how do you decide who should receive immunotherapy or chemoimmunotherapy?

A number of factors must be considered when determining who should receive single-agent immunotherapy vs chemoimmunotherapy. I personally divide them into 3 or 4 categories. Disease-related factors, particularly what histology we’re treating, has more of an impact on the chemotherapy backbone that is chosen [compared with the] immunotherapy [agent].

Currently, the best biomarker we have is PD-L1 by immunohistochemistry and tumor proportion score. We have excellent evidence supporting single-agent immunotherapy in patients with PD-L1 expression greater than 50%. For example, both pembrolizumab [Keytruda] and atezolizumab [Tecentriq] have now been approved for patients with PD-L1 expression greater than 50%.

It gets a little more nuanced in patients who have a PD-L1 expression of less than 50% or less than 1%. In this case, chemoimmunotherapy is the default option based on the KEYNOTE-189 study, which looked at chemoimmunotherapy vs chemotherapy. [The results indicated] that the outcomes were better [with chemoimmunotherapy]. We have very good response rates with patients who are treated with the combination of ipilimumab [Yervoy] and nivolumab [Opdivo] and, even in patients who are PD-L1 negative or those who have less than 1% of PD-L1 based on the CheckMate 227 study. [PD-L1] is a very important marker that I look at.

Next, there are patient-related factors, particularly the burden of disease and how quickly responses are acquired. There’s always been the concern that single-agent immunotherapy might take a little bit of time to work. But, if you look at single-agent immunotherapy studies, the median time to response was similar between chemotherapy vs immunotherapy in both single-agent studies of pembrolizumab and nivolumab. But, in clinical practice, there’s always the concern that we might not get a quick enough response. So, chemotherapy with immunotherapy is never a bad option [in a patient who presents with high disease burden or an oncological emergency] if they can tolerate it.

One of the most important determining factors tends to be performance status. Unfortunately, the ECOG performance status of 2 is not very well represented in any of these immunotherapy studies. Most studies looked at patients with a score of 0 or 1. Brain metastasis is another area of concern. All of these studies included patients who had been treated for their brain metastasis in some way; however, I still believe it’s underrepresented.

Finally, we have to look at other patient-related factors such as their ability to tolerate toxicity. We know that the combination of chemotherapy and immunotherapy is associated with more toxicity. There is also the financial impact. All these regimens typically [are given] for 2 years. As such, cost is another important factor.

Importantly, the biggest factor right now that’s occupying all of our minds is the coronavirus disease 2019 pandemic and its impact on our patients. I do believe there’s [anxiety] among doctors and patients alike with regard to chemotherapy because of the impact that it has on immunity. Generally, avoiding chemotherapy is a preferred practice in the field of oncology; however, we cannot do that in patients with lung cancer that express less than 50% of PD-L1, so we must be careful.

Five-year updated overall survival (OS) data from the KEYNOTE-024 trial were presented during the 2020 ESMO Virtual Congress. Could you highlight the updated data, along with the significance of this research?

KEYNOTE-024 was a groundbreaking study that enrolled patients with previously untreated NSCLC who did not have a driver mutation. The study was enriched by selecting patients who expressed high levels of PD-L1 of 50% or more. Patients were randomized to receive pembrolizumab monotherapy or platinum-based chemotherapy alone. Crossover was allowed from the chemotherapy arm to the single-agent immunotherapy arm at the time of progression.

The primary end point was progression-free survival [PFS] [because of the anticipated] effect of crossover on OS benefit. At the initial data read-out, we saw about a 4-month PFS benefit in favor of pembrolizumab vs chemotherapy. Notably, the objective response rate [ORR] of 45% was one of the highest single-agent monotherapy response rates we’ve seen in the frontline setting in lung cancer.

The updated analysis had about 25 months of follow-up and showed an OS benefit [with pembrolizumab that was] twice that of chemotherapy. The OS was about 30 months with pembrolizumab and about 14 months with chemotherapy alone. We now have 5-year survival data indicating that the 5-year survival probability seems to be double that with single-agent immunotherapy vs chemotherapy, at 32% versus 16%. The median duration [of response] was also nearly 5 times as much, at 30 months vs 6 months.

We also evaluate the safety profile closely, especially in terms of grade 3 or higher toxicities. [The rate of grade 3 or higher toxicities] was substantially lower with single-agent immunotherapy compared with chemotherapy. So, if a patient presents to my clinic with PD-L1 expression greater than 50%, we have great data to support single-agent immunotherapy with pembrolizumab or atezolizumab.

What are the differences between pembrolizumab and nivolumab?

KEYNOTE-024 and CheckMate 026, were similar in terms of enrollment. These were previously untreated patients with lung cancer who were randomized to receive single-agent immunotherapy or platinum-doublet chemotherapy. There were some differences. In CheckMate 026, the PD-L1 cutoff was 5% or higher. That study did not show the impressive responses seen with pembrolizumab as a single agent, which was quite a surprise. This was true even in patients with PD-L1 greater than 50%.

The obvious concern is that there is a market discrepancy, even though both of these drugs do the exact same thing and they are both PD-1 antagonists. I believe it has more to do with how patients are selected, rather than a true difference in terms of how these drugs work. In the nivolumab study, fewer patients were randomized to the single-agent nivolumab arm who expressed PD-L1 greater than 50% because of the lower PD-L1 cutoff.

We also use PD-L1 as bottom-up approach when selecting patients for these studies. When the investigators included patients with PD-L1 of 5% or more, this valued the benefit that was seen in patients with very high PD-L1 expression, of 70% or 90%.

We also faced a problem in the KEYNOTE study that evaluated pembrolizumab because we don’t know the exact percentages of patients who had very high PD-L1 expression. They were all lumped together as those with greater than 50%. We don’t have any evidence supporting that [a certain level of] PD-L1 expression [above 50%] has a better advantage in terms of outcomes. With that being said, I believe it has more to do with how patients are selected, rather than a true difference between the drugs.

Could you highlight the CheckMate 227 and MYSTIC trials? What did we learn about immunotherapy combinations from those studies?

CheckMate 227 and MYSTIC were similar in terms of choosing a combination of PD-L1 antagonists along with an anti-CTLA-4 antibody. CheckMate 227 looked at the combination of ipilimumab with nivolumab vs chemotherapy. It was a two-part study with somewhat of a complex randomization, based on PD-L1 expression. For patients that had PD-L1 expression greater than 1%, there was an additional arm looking at nivolumab monotherapy. In patients with PD-L1 less than 1%, there was a nivolumab plus chemotherapy arm. Survival benefit was discovered during this trial.

MYSTIC looked at the combination of durvalumab [Imfinzi] with tremelimumab and compared this with chemotherapy. Impressively, CheckMate 227 showed responses; however, in MYSTIC, there was a statistical analysis change at the time of initial enrollment. All patients with PD-L1 greater than 1% were to be included in the analysis. But then, based on results from other studies, there was a protocol amendment to include and look at patients with PD-L1 expression of greater than 25%. As such, only about 44% of the population could be analyzed for results. Because of this, the study lost power and the results were less impressive and not statistically significant.

That aside, both CheckMate 227 and MYSTIC looked at tumor mutation burden [TMB] as an exploratory biomarker. The only difference is that CheckMate 227 looked for it in tumor tissue, whereas MYSTIC looked for it in peripheral blood based on a cell-free DNA analysis, since the turnaround time would be shorter. In CheckMate 227, multiple combinations of TMB with PD-L1 were analyzed but the sample sizes started becoming much smaller. As such, it’s hard to bank on those results, statistically speaking. Irrespective of TMB greater than 20 mutations/Mb, or less than 20 mutations/Mb, there seems to be a benefit in the subgroup analysis with the combination of ipilimumab plus nivolumab.

MYSTIC did not show a statistically significant response but, numerically, the trend did favor patients with high TMB, along with peripheral blood as a predictor of response for the combination of durvalumab and tremelimumab. Both of these studies show us that patient selection, and how we select cutoffs, are very important in how results pan out in the end.

With the number of chemoimmunotherapy combinations available, how do you navigate these options? What are some of the key deciding factors?

For patients with PD-L1 between 1% and 50% or 1% and 49%, based on the KEYNOTE-189 study, the go-to option, which is now a well-established standard, is chemoimmunotherapy.

We now have data from KEYNOTE-407, demonstrating that patients with squamous cell carcinoma also do well with this combination. We also have data from CheckMate 9LA that looked at the combination of a CTLA-4 antibody with a PD-1 antagonist and 2 cycles of chemotherapy. Both of these studies showed a substantial benefit in terms of OS and PFS, as well as ORR for chemoimmunotherapy.

We must be careful when stratifying patients with PD-L1 greater than 50% vs less than 50%. I believe there is good evidence to support single-agent immunotherapy in those who express high levels of PD-L1.

How we choose between several different existing combinations of chemotherapy plus immunotherapy depends on several factors. Some of the patient-related factors include how quickly patients acquire a response, toxicity, as well as the financial toxicities.

If you look at the IMpower150 study, which evaluated atezolizumab plus a VEGF inhibitor, bevacizumab [Avastin], in combination with chemotherapy, toxicities and cost results were higher. There was also modest improvement in survival benefit. These factors must be taken into consideration.

How could other biomarkers of response to immunotherapy come into play?

TMB and PD-L1 have been studied systematically in only 1 study. I believe this area represents an unmet need in the space, especially in terms of choosing the right biomarker.

Not only could this help us determine who will benefit from immunotherapy, which is how current biomarkers are designed to predict benefit, but they could also help predict a lack of benefit.

Microsatellite instability is another biomarker that could come into play, as well as T-cell surface markers. These are important biomarkers to keep our eyes on.

Could you provide some background on the INSIGNIA trial? What will we learn from this research?

This is a big area of need for us in terms of determining how to choose between single-agent immunotherapy v chemoimmunotherapy. The benefit of chemotherapy in patients with PD-L1 greater than 50% has not yet been defined because there hasn’t been a comparison of pembrolizumab monotherapy vs the combination of pembrolizumab plus chemotherapy.

As such, the INSIGNIA trial is a 3-arm study that is evaluating 1 arm of pembrolizumab plus histology-directed chemotherapy. The other arm consists of pembrolizumab monotherapy and, upon progression on that arm, patients are going to be stratified into 2 further arms. One arm will receive chemotherapy alone, and the other arm will receive chemoimmunotherapy upon progression.

This study touches on an area where we don’t have clear answers yet, which is how to deal with patients who are progressing on single-agent immunotherapy. The standard currently has been to switch to chemotherapy alone, although, would there be a benefit in continuing immunotherapy and then adding chemotherapy? That would be an interesting study topic.