Challenges of Immunotherapy in Stage IV Non–Small-Cell Lung Cancer

Sophie Stock-Martineau, MD; Kate Magner, MD; Kevin Jao, MD; Paul Wheatley-Price, MBChB, MD


J Oncol Pract. 2021;17(8):465-471. 

Treatment for metastatic non–small-cell lung carcinoma has seen important advances in recent years with the introduction of targeted therapies and immunotherapy. Immune checkpoint inhibitors, which target the programmed death 1 receptor and programmed death ligand-1, alone or in combination with platinum-based chemotherapy, have become standard of care in the first-line setting for patients with advanced non–small-cell lung carcinoma without targetable driver mutations. However, several clinical questions have now since emerged. Physicians treating lung cancer lack guidance when treating patients who have a poor performance status, patients who are receiving corticosteroids, and those known for pre-existing autoimmune disorders. Furthermore, data are scarce on rechallenging a patient with immune checkpoint inhibitors after the occurrence of a significant immune-related adverse event. In this review, we aim to shed light on these topics.

In the last decade, treatment for advanced non–small-cell lung cancer (NSCLC) has significantly evolved, especially with the introduction of immune checkpoint inhibitors (ICIs). KEYNOTE-024 compared pembrolizumab, a monoclonal antibody targeting the programmed death 1 (PD1) receptor, with platinum-based chemotherapy in treatment-naïve patients with metastatic NSCLC and programmed death ligand-1 (PDL1) expression of 50% and more.[1] Pembrolizumab was superior to first-line chemotherapy and emerged as a new standard of care. In KEYNOTE-189, the combination of platinum-based chemotherapy showed improved progression-free survival (PFS) and median overall survival (OS) compared with chemotherapy alone in patients with advanced nonsquamous NSCLC, independent of PDL1 expression.[2] A corresponding trial in squamous cell lung cancer, KEYNOTE-407, demonstrated similar significant advantages to the immunotherapy-containing arm.[3] Numerous additional studies, not listed here, with other immunotherapy agents such as atezolizumab, nivolumab, and durvalumab have also become standard-of-care options in NSCLC. Most trials looking at ICI treatment in stage IV NSCLC exclude specific patient populations such as those with poor performance status, those on corticosteroid treatment, and those with certain pre-existing autoimmune disorders. Clinicians treating NSCLC often encounter these types of patients in clinic. In this review, we provide a summary of recent data relevant to different challenges with ICI treatment in stage IV NSCLC, looking at special patient populations and rechallenge after an immune-related adverse event (irAE).

Patients With Poor Performance Status

Most pivotal prospective randomized clinical trials that studied ICIs in stage IV NSCLC excluded patients with performance status (PS) Eastern Cooperative Oncology Group (ECOG) two or worse. In the outpatient setting, patients with advanced NSCLC may present with high burden of disease, symptomatic disease, and poor PS. According to the National Lung Cancer Audit Annual Report from the United Kingdom published in 2018, 18% of patients with lung cancer were PS 2.[4] A clinical dilemma encountered by oncologists is whether one can extrapolate data from ICI studies that carefully selected patients with ECOG 0–1 and apply it to patients with a worse functional status.

Four prospective studies looked specifically at patients with advanced NSCLC, poor PS, and who received immunotherapy (Table 1). PePS2 is a phase II multicenter, single-arm trial in the United Kingdom.[5] It included patients with advanced NSCLC, PS 2 treatment-naïve, or pretreated and any PDL1 expression. They were treated with pembrolizumab 200 mg every 3 weeks. Forty percent of patients were treatment-naïve and were stratified according to tumor proportion score and line of therapy. Durable clinical benefit (defined as the occurrence of complete response, partial response, or stable disease that continues until at least the second computed tomography scan scheduled at 18 weeks) and toxicity were the coprimary outcomes. Durable clinical benefit was observed in 36% of the pretreated patients. The overall response rate (ORR), median PFS, and OS for pretreated patients were 31%, 4.4 months, and 10.4 months, respectively. These results are comparable with data from KEYNOTE-001, which included patients with PS 0–1.[6] This study showed an ORR, mPFS, and mOS of 18%, 3 months, and 9.3 months, respectively, for the pretreated population receiving pembrolizumab. Outcomes from PePS2 and KEYNOTE-001 were also similar in the first-line setting. The authors concluded that pembrolizumab in a population with PS2 is as effective as those with a good PS of 0–1, with the same degree of toxicity.

CheckMate 817 is a multicohort phase IIIb or IV trial that treated patients with metastatic NSCLC with nivolumab and ipilimumab in the first-line setting.[7] Nivolumab 240 mg every 2 weeks plus ipilimumab 1 mg/kg every 6 weeks were given for two years or until disease progression or unacceptable toxicity. Cohort A had good PS (ECOG 0–1). Cohort A1 had ECOG PS 2 or ECOG PS 0–1 with one of the following: asymptomatic untreated brain metastases, hepatic or renal impairment, or HIV. The ORR and mPFS in cohort A1 PS2 population were 20% and 3.6 months, respectively. In comparison, cohort A had improved outcomes with an ORR and mPFS of 35% and 6 months, respectively. Interestingly, rates of grade 3–4 treatment-related adverse events were similar among the various cohorts, irrespective of comorbidities.

CheckMate 171 is an open-label phase II trial that included pretreated advanced squamous cell carcinomas of the lung, and patients received nivolumab 3 mg/kg every 2 weeks until progression or unacceptable toxicity.[8] Patients with PS 2 represented 13% of the population. Median OS in all treated patients was 10 months and 5.2 months in those with PS 2. In comparison, CheckMate 153 had similar inclusion criteria and treatment consisted of nivolumab, but was a phase IIIb/IV and only 9% had a PS of 2.[9] Similar to CheckMate 171, median OS in the overall population and PS 2 were 9.1 and 4.0 months, respectively.

Juergens et al[10] reported outcomes of 472 pretreated patients with metastatic NSCLC receiving nivolumab. Totally, 8.9% of patients had ECOG 2. The whole population had an mOS of 12 months. The subgroup with an ECOG 2 had a shorter mOS at 6.8 versus 12.9 months for ECOG 0–1. Median time to treatment discontinuation was similar between both groups 2.8 versus 3.7 months, suggesting equivalent tolerability. This study and CheckMate 817 suggest that although PS2 patients have poorer survival outcomes on ICI therapy, they are not more susceptible to additional toxicities compared with more fit patients.

Future randomized-controlled studies only including patients with poor PS should help clinicians in treatment decision making. The phase III eNERGY trial is underway and will randomly assign elderly or PS 2 patients with treatment-naïve stage IV NSCLC to nivolumab + ipililumab versus carboplatin-containing chemotherapy doublet.[11] IPSOS is a phase III trial that randomly assigned patients ineligible to platinum-based chemotherapy with locally advanced or recurrent NSCLC to atezolizumab versus single-agent chemotherapy. It has completed accrual and results are awaited.[12]

Patients on Corticosteroids

ICIs represent a significant cornerstone in treating patients with stage IV NSCLC in the first-line and subsequent line setting. The concurrent use of immunosuppressive corticosteroids is thought to decrease the antitumor responses to PD1 or PDL1 inhibitors. Patients on steroids have often been excluded from ICI trials for this reason. Also, patients frequently require steroids at some point during the course of their disease for treatment of cancer-related issues, such as brain metastasis, fatigue, and dyspnea.[13–16] They may also require steroids for chronic obstructive pulmonary disease exacerbation and irAEs. Therefore, using ICIs in patients on steroids represents a real conundrum for physicians treating lung cancer. The core question is whether the use of steroids is antagonistic to the efficacy of immunotherapy, or if observed poorer outcome is merely selection bias and represents an already unfavorable patient population. Several retrospective studies have looked at this issue (Table 2). Scott and Pennell[17] studied 210 patients with advanced NSCLC on nivolumab. Sixty-six patients (31%) received concurrent steroids (> 10 mg of prednisone or equivalent). The most common causes of steroid treatment were brain metastasis (27%) and chronic obstructive pulmonary disease exacerbations (21%). Median OS was shorter for patients on steroid versus not during the first 30 days of nivolumab (4.3 v 11 months). Ricciuti et al[18] retrospectively analyzed 650 patients with stage IV NSCLC receiving an anti-PD1 or PDL1 alone or with an anticytotoxic T-cell lymphocyte–associated protein 4. They looked at three separate groups: those on < 10 mg of prednisone or equivalent (557 patients), ≥ 10 mg of prednisone for cancer-related indications (66 patients), or ≥ 10 mg of prednisone for noncancer-related indications (27 patients). Patients included in the steroid group had to be on a steroid within 24 hours of receiving an ICI. ORR, mPFS, and mOS were all worse in the overall steroid group (10.8 v 19.7%; 2.0 v 3.4 months; 4.9 v 11.2 months). Patients on ≥ 10 mg of prednisone for cancer-related indications had specifically worse mPFS and mOS versus those for cancer-unrelated indications and < 10 mg of prednisone (mPFS, 1.4 v 4.6 v 3.4 months; mOS, 2.2 v 10.7 v 11.2 months, respectively). When looking at baseline characteristics, the group on ≥ 10 mg of prednisone for cancer-related indications was more likely to have poor performance status and brain metastasis. When adjusting for ECOG and PDL1 status in a multivariate analysis, the use of ≥ 10 mg of prednisone for cancer-related indications was still associated with significantly higher mortality risk (hazard ratio [HR], 1.60; 95% CI, 1.07 to 2.39; P = .02).

Arbour et al[19] retrospectively analyzed 640 patients with advanced NSCLC treated with single-agent anti-PDL1. Ninety (14%) patients were on steroids (≥ 10 mg of prednisone equivalent) at the start of ICI therapy. Baseline steroids were associated with worse ORR, PFS, and OS. After adjusting for PS and brain metastases, multivariable analysis showed that steroids were still significantly associated with shorter PFS (HR, 1.3; P = .03) and OS (HR, 1.7; P = .001). Fuca et al[20] studied 150 patients with NSCLC treated with an ICI. Thirty-five (23%) of these were on ≥ 10 mg of prednisone equivalent early during ICI treatment, defined as at least 1 day within 28 days of therapy initiation. Peripheral blood was collected at baseline, 4 weeks, and 6 weeks. Early use of steroids was independently correlated with worse PFS (HR, 1.80; P = .003) and OS (HR, 2.60; P ≤ 0.001) after adjusting for PS and PDL1 status. Interestingly, the early use of steroid was associated with higher peripheral neutrophil counts, higher neutrophil-to-lymphocyte ratio, and lower eosinophil counts. Early steroid use–related peripheral blood cell count modulation might biologically explain worse outcomes with ICI therapy, but this needs to be prospectively validated.

In summary, retrospective studies have shown a detrimental effect for patients on steroids when starting ICIs, independent of performance status and presence of brain metastasis. When used for cancer-related issues, corticosteroid use is associated with a poorer prognosis. On the basis of the currently available data, most experts still agree on avoiding concurrent use of supraphysiological doses of steroids at baseline and early on after ICI initiation. Hopefully, future prospective trials looking at ICI use in patients on steroids will be able to clarify this question.

Patients With Pre-existing Autoimmune Disorders

ICIs pose a risk of potentially severe autoimmune toxicity affecting diverse organ systems. To limit the risk of autoimmune symptom activation or adverse events, clinical trials of ICIs have conventionally excluded patients with pre-existing autoimmune disease.[21] The impact of this practice has been evaluated in large data sets, suggesting that 14%-25% of patients diagnosed with lung cancer concomitantly have an autoimmune disease.[21–23] Patients with autoimmune disease are more likely to be female, older, and more comorbid and have earlier stage lung cancer. The most common autoimmune diseases in these patients were rheumatoid arthritis, psoriasis, and polymyalgia.[21–23] Chen et al sought to understand the prevalence of autoimmune disease among patients receiving ICIs in real-world clinical care using data from ASCO's oncology database. The prevalence of autoimmune disease among patients receiving ICIs was 27%. The most common autoimmune diseases were glucocorticoid deficiency, rheumatoid arthritis, and sacroiliitis.[23]

The efficacy and safety of ICIs in patients with lung cancer and in patients with different solid tumors have been assessed in observational studies. Kehl et al reviewed outcomes in patients with all cancer types treated with ICIs from 2011 to 2017. Of 4,438 patients, pre-existing autoimmune conditions were present among 4%-6% depending on the evaluation criteria used. Pre-existing autoimmune disease, defined by both strict and relaxed criteria, was associated with hospitalization secondary to an irAE (for strict criteria: HR, 1.81; 95% CI, 1.21 to 2.71) and with corticosteroid treatment (for strict criteria: HR, 1.93; 95% CI, 1.35 to 2.76), but not with hospitalization for any cause (for strict criteria: HR, 1.27; 95% CI, 0.998 to 1.62).[24]

In a retrospective cohort study of 56 patients with NSCLC and an autoimmune disease who received a PD1 inhibitor, 38% of patients developed an irAE, including 11% who developed grade 3 or 4 irAEs. Flares of existing autoimmune disease occurred in 23% of the cohort. Overall, 14% required permanent anti-PD1 treatment discontinuation because of the development irAEs.[23] This is slightly higher than discontinuation rates (between 3% and 8%) reported in clinical trials that excluded patients with autoimmune conditions. The ORR to immunotherapy in this population was 22%.[25]

Similarly, in a retrospective cohort study of patients with NSCLC who received an ICI, Khozin et al[22] found that the time to treatment discontinuation, time to next treatment, real-world PFS, and OS did not differ between patients with or without an underlying autoimmune disease.

Not all patients with autoimmune disorders at baseline are excluded from receiving ICIs. The US Food and Drug Administration published a post hoc analysis including 22 trials with NSCLC patients with baseline autoimmune disorders who received an ICI.[26] Five hundred fifty-two patients were included, and none required steroids at baseline. Most frequent autoimmune diseases included thyroid disorder (n = 188), psoriasis (n = 70), and vitiligo (n = 44). Flares of pre-existing autoimmune disorders were found in 6%-16% of patients. Two cases of grade 4 hyperglycemia in diabetic patients; three cases of grade 3 hypothyroidism, psoriasis, and interstitial lung disease and one case of grade 3 ankylosing spondylitis were identified. A small proportion 8%-9% of patients required oral steroids. In all, ICI treatment in patients with thyroid, psoriasis, and vitiligo, among other less severe autoimmune diseases, seems safe. However, detailed data on exact type of autoimmune condition, diagnostic methods, and severity were missing in a large proportion of patients in this series.

Using ICIs in patients with pre-existing autoimmune disease is probably reasonable, especially with more benign conditions not requiring high-dose steroids and particularly in view of advanced malignancy with few other treatment options. These cases should be presented and discussed at multidisciplinary lung cancer rounds given their complexity. Close monitoring is required in this population since up to half might be expected to experience an autoimmune flare or an irAE. Most of these toxicities are mild and usually manageable without treatment discontinuation.[27]

Rechallenging With ICIs After an irAE

Patients using ICIs are at risk of developing irAEs. Most of these are mild to moderate in severity and resolve after discontinuation of the ICI and treatment with steroids.[28] Oncological guidelines suggest that discontinuation of treatment can be temporary for grade 2–3 irAEs but should be permanent for those patients who experience grade 4, or severe, irAEs.[29–31]

This approach is supported by cohort study data (Table 3), including a study by Santini et al,[32] which identified 38 patients with NSCLC who were retreated with an ICI after previously discontinuing treatment because of an irAE. Of these patients, 18 (48%) had no recurrence of irAEs. Ten (26%) experienced recurrence of the same irAE, and another 10 (26%) developed a new irAE. Recurrent or new irAEs were more likely if the initial event required hospitalization. The grade of the initial adverse event and the time to retreatment did not, however, significantly affect the risk of irAEs on rechallenge. Of those who experienced a recurrent or new irAE, 58% were mild (grade 1–2). There were two treatment-related deaths.

Similarly, Dolladille et al conducted a retrospective cohort study of irAEs in patients with cancer undergoing rechallenge with a broad range of ICI regimens. They report that 28.8% (95% CI, 24.8 to 33.1) of patients experienced recurrence of the same irAE that had led to their initial treatment discontinuation. In this study, colitis, hepatitis, and pneumonitis were associated with a higher recurrence rate.[28]

Pollack et al retrospectively examined 80 patients with metastatic melanoma who were rechallenged with an anti-PD1 agent after discontinuation for a variety of irAEs. The median follow-up was 14.3 months, and the same irAEs occurred in 18% after the rechallenge. Any type of irAE occurred in 39% of patients.[33]

Simonaggio et al[34] retrospectively analyzed the incidence of a second irAE in 93 patients with cancer, among whom 40 were rechallenged with an anti-PD1 or anti-PDL1 inhibitor after an initial grade 2 or higher irAE. Of these 40 patients, 17 (42.5%) experienced a recurrence of the same type of adverse event, and five (12.5%) experienced different irAEs. Notably, the second irAEs were not more severe than the first events that had led to the initial treatment discontinuation.

These studies suggest that retreatment with ICIs can be safely considered in many patients who have experienced an irAE, although up to 50% may either get a recurrence of the same irAE or a new one, but not necessarily at the same severity. irAEs occur and recur frequently but are typically manageable.

In summary, in those who discontinue treatment with ICIs because of irAEs, treatment rechallenge can be safe and appropriate. Haanen et al suggested two scenarios: the first is to rechallenge with the same therapy after complete resolution of the initial adverse event, and the second is to concomitantly treat with steroids when resuming ICI therapy.[31] Data to support these approaches are scarce. Most studies are retrospective and have small number of patients with NSCLC and thus must be interpreted with caution. More work is underway in this area: a phase II study of durvalumab treatment in patients who discontinued prior checkpoint inhibitor therapy because of immune-related toxicity is currently open to accrual and aims to determine if patients who discontinued immunotherapy because of severe side effects related to treatment can safely be retreated with durvalumab at the time of disease progression.[35] The benefit of prophylactic steroids will be investigated, as will the response to retreatment. In the meantime, irrespective of the approach taken, proper patient selection for a rechallenge with immunotherapy is important and may depend on initial treatment response, the type and severity of the initial adverse event, and the availability of alternative treatment options.[31]

In conclusion, ICIs are standard of care in the treatment of advanced NSCLC, whether alone or in combination with chemotherapy, and their use is expanding. Studies looking at patients with a poor PS treated with ICIs in NSCLC are heterogenous, and thus, it still remains unclear if these patients benefit enough to warrant ICI use. Retrospective data looking at concurrent use of corticosteroids and ICIs tell us we should, in general, try to avoid overlapping the two, but that steroid use is not an absolute contraindication to ICI use. Initiating ICIs in patients with autoimmune disorders and rechallenging patients with irAEs can be feasible and safe, but higher rates of adverse events should inform the treating clinician to perform close follow-up.