In advanced NSCLC studies, immunotherapy was initially compared with docetaxel after failure of initial platinum-based chemotherapy. FDA-approved immune checkpoint inhibitors for NSCLC include nivolumab, pembrolizumab, and atezolizumab. Across these PD-1 and PD-L1 inhibitors, response rates were in the 12% to 20% range, with superior PFS and OS compared with chemotherapy.

The other part of the story is that a subset of patients had quite prolonged responses. This drove further research into which patients with NSCLC are most likely to benefit from immunotherapy, including tumors with higher PD-L1 expression, and whether combining immunotherapy and chemotherapy could improve efficacy. This led to clinical trials investigating combinations of first-line chemotherapy with immunotherapy for NSCLC.

KEYNOTE-189 was a randomized phase III study that evaluated the addition of pembrolizumab to standard platinum-based chemotherapy with pemetrexed in patients with previously untreated, stage IV nonsquamous NSCLC (N = 616).⁶⁷ Of note, the study allowed patients with any level of PD-L1 expression to be enrolled. Patients were not allowed to have actionable EGFR or ALK alterations, because those patients should receive targeted therapy as first-line treatment.

Patients were randomized to platinum-based chemotherapy with or without pembrolizumab. After 4 cycles, the chemotherapy-only arm continued on pemetrexed maintenance with placebo, while the pembrolizumab arm received maintenance with pemetrexed plus pembrolizumab. Treatment was continued until progressive disease or unacceptable toxicity; patients in the placebo arm could crossover to get pembrolizumab. The coprimary endpoints were OS and PFS.

KEYNOTE-189 was an impressively positive study. In the overall population, the HR for OS was 0.49, favoring the addition of pembrolizumab to standard platinum-based chemotherapy (P < .001).⁶⁷ An important part of the story is that the OS benefit was seen at every level of PD-L1 expression. We expected patients with ≥ 50% PD-L1 expression to benefit with the addition of pembrolizumab, because second-line trials showed that patients with high PD-L1 expression had the strongest benefit from pembrolizumab. However, there was a statistically significant OS benefit across PD-L1 expression levels, including in those with medium expression (1% to 49%; HR: 0.55) and even in patients with < 1% expression (HR: 0.59). These results from KEYNOTE-189 established chemotherapy plus pembrolizumab as the standard of care for first-line treatment of nonsquamous NSCLC.

IMpower150 was a randomized phase III trial asking the question of whether the addition of the PD-L1 inhibitor atezolizumab to standard platinum-based chemotherapy plus bevacizumab would improve outcomes in advanced NSCLC.^68-70 This study enrolled patients with stage IV or recurrent, chemotherapy-naive, nonsquamous NSCLC (N = 1202).

Patients in this study were randomized to 1 of 3 arms consisting of the standard carboplatin and paclitaxel chemotherapy backbone plus bevacizumab 15 mg/kg or atezolizumab 1200 mg, or both atezolizumab and bevacizumab. As maintenance therapy, each arm received the same antibodies as in the treatment phase. Atezolizumab was continued until progression or loss of benefit and bevacizumab was continued until progression. The coprimary endpoints were PFS and OS. Secondary endpoints included ORR, safety, and pharmacokinetics.

Results from IMpower150 showed a statistically significant improvement in both PFS and OS with atezolizumab plus carboplatin, paclitaxel, and bevacizumab vs carboplatin, paclitaxel, and bevacizumab alone, leading to the approval of this regimen as a first-line alternative for patients with nonsquamous NSCLC.⁷⁰ The median PFS was 8.3 months with atezolizumab vs 6.8 months with chemotherapy alone (HR: 0.59; P < .0001), and the median OS was 19.8 months and 14.9 months, respectively (HR: 0.76). 

Of interest, unlike KEYNOTE-189, IMpower150 did allow patients with EGFR and ALK aberrations as long as they had already progressed on appropriate targeted therapy. With the caveat that this was a small subset of patients (n = 104), there was a reasonably strong trend toward an OS benefit (HR: 0.54). This was the first phase III trial to include these patients and to show benefit from an immunotherapy. Bevacizumab was included because it is known to be active in patients with NSCLC and EGFR mutations. However, the FDA approval in the first-line setting does not include patients with EGFR or ALK aberrations.

The IMpower150 investigators conducted a prespecified subgroup analysis of patients with liver metastases.⁷⁰ This stems from an interesting finding from the ECOG 4599 trial that patients with liver metastases appeared to have a greater magnitude of benefit with the addition of bevacizumab to chemotherapy. In IMpower150, results of this subgroup analysis showed an HR strongly favoring the quadruple combination over carboplatin, paclitaxel, and bevacizumab alone in patients with liver metastases: The median OS was 13.2 vs 9.1 months, respectively (HR: 0.54). As mentioned above, a similar magnitude of benefit was seen when looking at OS in patients with EGFR or ALK aberrations (HR: 0.54).

IMpower130 is an ongoing randomized phase III study of atezolizumab in nonsquamous stage IV NSCLC (N = 724).⁷¹ Unlike IMpower150, IMpower130 used nab-paclitaxel instead of paclitaxel with carboplatin because the albumin-bound formulation is thought to be more tolerable for some patients than paclitaxel.

Patients in IMpower130 were randomized 2:1 to chemotherapy with or without atezolizumab, followed by atezolizumab maintenance for patients in the atezolizumab arm and either observation or pemetrexed maintenance in the chemotherapy-only arm. Patients on the chemotherapy-only arm were allowed to cross over to atezolizumab upon progression. The primary endpoints were PFS plus OS in patients with wild-type EGFR and ALK. Secondary endpoints included PD-L1 expression, OS over time and by PD-L1 level, and time to symptom deterioration.

IMpower130 did show a benefit regarding PFS and OS favoring the addition of atezolizumab to chemotherapy. The median PFS was 7.0 vs 5.5 months, respectively (HR: 0.64; P < .001), and the median OS was 18.6 vs 13.9 months (HR: 0.79; P = .033). At 1 year, 63% of patients in the atezolizumab arm remained alive vs 56% of those receiving chemotherapy only.⁷¹

As a result of these findings, in December 2019, the FDA approved atezolizumab in combination with nab-paclitaxel and carboplatin for the first-line treatment of nonsquamous NSCLC with no EGFR or ALK aberrations.⁷² This is now another option for the initial treatment of advanced NSCLC.

For squamous cell disease, where the chemotherapy combinations of choice are taxane and platinum based, KEYNOTE-407 was the first study to show a benefit with the addition of pembrolizumab.⁷³ This phase III study randomized patients with previously untreated stage IV squamous NSCLC to receive carboplatin and the investigator’s choice of paclitaxel or nab-paclitaxel, with or without pembrolizumab (N = 559). Brain metastases were not allowed. Pembrolizumab was continued up to 31 cycles and patients on the placebo arm were allowed to cross over to single-agent pembrolizumab upon progression. The primary endpoints were PFS and OS.

Similar to KEYNOTE-189 in nonsquamous NSCLC, KEYNOTE-407 showed a PFS benefit with the addition of pembrolizumab for the entire population that was impressive, with an HR of 0.56 (P < .001); the median PFS was 6.4 vs 4.8 months, respectively. Moreover, there was a PFS benefit at every level of PD-L1 expression, with HRs for < 1%, 1% to 49% and ≥ 50% PD-L1 expression of 0.68, 0.56, and 0.37, respectively.⁷³

Similarly, the OS was also superior with the addition of pembrolizumab, with an HR of 0.64 (P = .0008) in the overall population; the median OS was 15.9 months vs 11.3 months, respectively. As with PFS, the OS curves clearly separated at every level of PD-L1 expression.⁷³

Results from KEYNOTE-407 led to carboplatin, a taxane, and pembrolizumab becoming the standard of care for first-line treatment of squamous NSCLC. Of note, a subgroup analysis showed there was good benefit with either paclitaxel or nab‑paclitaxel, so both of these taxanes are options if appropriate.

Results from studies of single-agent pembrolizumab as second-line treatment for NSCLC have shown that some patients will enjoy excellent durable responses. This brings up the question of whether certain patients can receive pembrolizumab alone (ie, deferring chemotherapy) as first-line therapy for NSCLC.

KEYNOTE-024 was an open-label, randomized, phase III study of pembrolizumab vs platinum doublet chemotherapy in patients with previously untreated stage IV NSCLC and high PD-L1 expression (N = 305).^74,75 Patients had a PD-L1 tumor proportion score of at least 50%, no actionable EGFR or ALK aberrations, either squamous or nonsquamous histology, and CNS metastases were allowed if treated. Pembrolizumab was given at 200 mg every 3 weeks for up to 35 cycles, and crossover from chemotherapy to pembrolizumab was allowed on progression. The primary endpoint was PFS. 

KEYNOTE-024 was stopped early because the OS signal was so strong. As shown here, the median PFS was improved by approximately 4 months with first-line pembrolizumab vs chemotherapy (10.3 vs 6.0 months, respectively; HR: 0.50; P < .001). The 12-month PFS rate was robust with pembrolizumab at 48% vs 15% with chemotherapy.^74,75

The median OS was doubled with the use of pembrolizumab immunotherapy over chemotherapy: 30.0 vs 14.2 months (HR: 0.63; P = .002). The OS rate at 24 months was high for pembrolizumab: More than one half of patients in the pembrolizumab arm remained alive vs approximately one third of the patients who received chemotherapy. Clearly, pembrolizumab is a very reasonable choice for first-line use as a single agent in patients with advanced NSCLC and tumors with high PD-L1 expression.

KEYNOTE-024 used a PD-L1 expression threshold of 50% or higher. KEYNOTE-042 was a randomized phase III study of first-line pembrolizumab in a broader swath of the NSCLC population: those with at least 1% PD-L1 expression (N = 1274).⁷⁶ Patients could have either squamous or nonsquamous stage IV NSCLC, no previous treatment for advanced disease, no untreated brain metastases, and no actionable EGFR or ALK mutations.

Patients were randomized to either pembrolizumab 200 mg every 3 weeks for up to 2 years or chemotherapy. Of note, on this international study, crossover to pembrolizumab was not allowed from the chemotherapy arm. The primary endpoint was PFS, with secondary endpoints including ORR, OS, and safety.

Overall, as demonstrated in the lower OS graph, the study was positive: In all patients with a PD-L1 expression of at least 1%, the HR was 0.81 and statistically significant for favoring the use of pembrolizumab over chemotherapy (P = .0018). Similarly, the other 2 OS graphs showed that for patients with at least 20% or at least 50% PD-L1 expression, pembrolizumab also demonstrated a significant OS advantage over chemotherapy.⁷⁶

The investigators conducted an exploratory analysis of OS in patients with 1% to 49% PD-L1 expression (ie, patients who were not included in KEYNOTE-024), and found an HR of 0.92, basically at unity.^76,77 This showed no clear benefit of pembrolizumab over chemotherapy alone in this group and the OS benefit seen in the overall population was mainly driven by the group with PD-L1 expression ≥ 50%.

In my opinion, it is not correct to say that pembrolizumab was just as good as chemotherapy with less toxicity because these survival curves overlap. Instead, I would argue that since crossover to pembrolizumab as second-line treatment was not allowed, the patients on the chemotherapy arm actually had suboptimal therapy, to which pembrolizumab had similar survival. For patients with low PD-L1 expression (1% to 49%), I feel that chemoimmunotherapy is a stronger choice and more effective, despite the FDA approval of pembrolizumab monotherapy in this setting.

In May 2020, the FDA also approved atezolizumab monotherapy as first-line treatment of metastatic NSCLC with high PD-L1 expression and no EGFR or ALK aberrations.

To aid in treatment planning, Clinical Care Options has developed a very useful interactive decision support tool for NSCLC treatment, which can be found at clinicaloptions.com/Lungtool. A user selects a combination of patient and disease characteristics through a series of questions, and the tool then provides treatment recommendations based on the entered characteristics from myself and 4 other experts in NSCLC care.

Earlier, we discussed the effectiveness of targeted therapy in patients with NSCLC and EGFR mutations, including the inevitability of progression and the question of when to use immunotherapy.

This forest plot shows the benefit from single-agent checkpoint inhibitors as second-line therapy in patients with NSCLC and EGFR mutations.⁷⁸ In all 4 of the major studies shown here, the PD-L1 or PD-1 inhibitor was clearly favored over docetaxel in the EGFR wild-type population. By contrast, in patients whose tumors had EGFR mutations, checkpoint inhibitors were not superior to chemotherapy.

Of note, EGFR mutation was the only clinical variable not associated with benefit in this meta-analysis. So, while there are some patients with NSCLC and EGFR mutations who can derive benefit from pembrolizumab or other PD-L1/PD-1 inhibitors, generally we will save these drugs for later-line use when other opportunities are exhausted.

Another potential biomarker of interest is the TMB. This graphic shows that different tumors have a different number of somatic mutations per megabase.⁸² Melanoma is known to have very high TMB, presumably because it represents an ongoing assault of ultraviolet radiation in the vast majority of patients. Lung cancer also has a relatively high TMB.

One question is whether patients with greater TMB may derive more benefit from immunotherapy. In a study we will discuss next, CheckMate 227, where TMB was defined at the cut point of 10 mutations per megabase, it is interesting to note that there was no correlation between TMB and PD-L1 expression.^83,84

CheckMate 227 was a complex, randomized, open-label phase III study of immunotherapy vs chemotherapy in patients with stage IV or recurrent NSCLC and no EGFR or ALK alterations (N = 1739).^83,85 Patients were initially distinguished by PD-L1 expression: those with ≥ 1% PD-L1 expression (n = 1189) were randomized to receive nivolumab plus ipilimumab, histology-based chemotherapy, or nivolumab alone; those with < 1% PD-L1 expression (n = 550) were randomized to receive nivolumab plus ipilimumab, histology-based chemotherapy, or nivolumab plus chemotherapy. This strategy was based on the idea that single-agent nivolumab is not appropriate for patients who do not express PD-L1.

The coprimary endpoints were PFS in the TMB-selected population and OS in patients with ≥ 1% PD-L1 expression.

Results from CheckMate 227 showed that PFS with the immunotherapy combination of nivolumab plus ipilimumab was superior to chemotherapy in patients with a high TMB, with an impressive HR of 0.58 (P < .001).⁸³ This was particularly evident in patients with PD-L1 expression < 1%, where only patients with a high TMB benefited from the combination (HR: 0.48 vs 1.17 with low TMB).⁸⁶ However, a similar OS benefit was reported with the combination vs chemotherapy in both high and low TMB populations.

This study was intended to help answer the broader question of whether dual immunotherapy can improve OS over chemotherapy, particularly in a patient population with PD-L1 expression of 1% or higher. As shown here, results from CheckMate 227 did favor the use of dual immunotherapy over chemotherapy, with an HR of 0.79 (P = .007) for OS.⁸⁵ The median OS in PD-L1 expressors was 17.1 months with immunotherapy vs 14.9 months with chemotherapy. The OS at 2 years was 40% with immunotherapy vs 33% with chemotherapy.

Although the primary endpoint of this study was specific to patients with PD-L1 expression of 1% or greater, the investigators also reported OS in patients who were nonexpressors (ie, PD-L1 < 1%). Of interest, even in this patient population, the combination of nivolumab and ipilimumab improved OS compared with chemotherapy, with a median OS of 17.2 months vs 12.2 months, respectively. At 2 years, 40% of patients in this group receiving immunotherapy remained alive vs 23% in the chemotherapy group.

As we know from melanoma, the use of immunotherapy combination clearly carries a higher risk of toxicity than single immunotherapy, even without the use of chemotherapy. In CheckMate 227, grade 3/4 AEs were  increased with nivolumab/ipilimumab vs nivolumab alone (31.2% vs 18.9%). Arguably, this is similar to the 36.1% of patients in the chemotherapy arm who developed grade 3/4 AEs, although the types of AEs are different.⁸³

As expected, there were more immune-related AEs in the immunotherapy arms of this trial. All-grade rash was seen in 16.7% of patients receiving nivolumab plus ipilimumab vs 5.1% with chemotherapy. All-grade diarrhea and pruritis were also experienced by more patients receiving the immunotherapy combination vs those in the chemotherapy arm. Hematologic AEs were more frequent in the chemotherapy arm.

Some patients with advanced NSCLC will have a very long duration of benefit, potentially years, from single-agent immunotherapy, immunotherapy/immunotherapy, or immunotherapy/chemotherapy combinations. However, there are patients who are immune-refractory, either de novo or who develop resistance. There are myriad studies looking at different combinations including immune checkpoint inhibitors in the first-line and second-line settings, and beyond.

This table highlights many different novel approaches being investigated for treating immune-refractory NSCLC, with agents including different checkpoint inhibitors, hypomethylating agents, targeted agents, vaccines, and even combinations including immunotherapy and radiation.⁸⁷ I hope we will be able to improve the immune response in a way that is tolerable for our patients.

As promising as immunotherapies are, they also introduce the need for clinics to address immune-related AEs. Every organ of the body can be affected by immune-related toxicities.^88-91 The majority of these are mild to moderate and manageable, but a few can be life threatening. Because these are not toxicities associated with the therapies that we have used previously, they require foresight and training to recognize, diagnose, and treat.

The onset is variable; immune-related AEs can occur on Day 1 of therapy, or long after cessation of therapy—a high threshold of suspicion is necessary. Inflammation of any organ of the body is possible, and management of those toxicities requires interdisciplinary collaboration with other specialists, as well as our care teams. Most immune-related AEs can be reversed with steroids, but some will require treatment discontinuation. It is critical that clinicians be educated in identifying the signs and symptoms of immune-related AEs, know how to treat them appropriately, and then share that information with patients and their caregivers.

This graph shows how variable the time to onset of immune-related AEs can be in patients with NSCLC receiving immune checkpoint inhibition with nivolumab.⁹² Trends include gastrointestinal and skin reactions occurring earlier in the treatment course (within 6 weeks), whereas endocrine and pulmonary AEs might occur later, even after more than 2 years. These are generalizations and we have much to learn. The key point is to have a low threshold of suspicion for toxicities during therapy, and even after therapy is discontinued.

There are fundamental principles that generally pertain to treating immune‑related AEs across organ systems.⁸⁸ For grade 1 AEs, immunotherapy can often continue with closer monitoring and symptomatic care. However, as AEs become grade 2 or higher, we have to consider holding immunotherapy temporarily, and discontinuing it in the case of grade 3/4 events.

Grade 2 AEs may merit the use of low‑dose corticosteroids, but for higher grades, patients may need higher doses or IV doses, or even hospitalization for more careful monitoring and management. We are also finding that some of the immune‑related AEs are refractory to traditional steroids like prednisone and methylprednisolone, and may require infliximab or other immunosuppressive therapy. This typically requires the cooperation of a multidisciplinary team. Those with grade 3 AEs should discontinue therapy and start steroid doses of 1-2 mg/kg/day. Patients with grade 4 immune-related AEs should permanently discontinue that therapy.

Some immune‑related AEs can be grade 5, which adds emphasis to the recommendation to teach patients to promptly report symptoms that are out of the ordinary so they can be identified and treated. As shown here, there are patient education resources available on managing immune-related AEs from various sources.

The National Comprehensive Cancer Network^® (NCCN^®) and Clinical Care Options have developed a very useful interactive decision support tool for managing immune checkpoint inhibitor–related toxicities, which can be found at clinicaloptions.com/immuneAEtool. A user selects the affected organ system and severity through a series of questions, and the tool then provides a consensus-based recommendation for management that reflects what is found in the NCCN guidelines. This is particularly useful for clinicians who identify an unusual symptom to guide them toward the appropriate therapy.