Introduction

Lung cancer is the main cause of cancer-related mortality worldwide. Chemotherapies have reached a plateau of efficacy, and patient tolerance to them is often poor. Although targeted therapies have led to dramatic improvements in the management of selected patients, the tumors have inexorably developed mechanisms of resistance to these drugs. Thus, new strategies need to be developed to cover a larger population, to obtain better safety profiles, and to obtain more sustainable responses. Recently, although the antigen-dependent approach has seemed disappointing, antitumor immunotherapy has rapidly become integrated into the management of non-small lung cancer (NSCLC) after positive clinical trials evaluated the immune checkpoint inhibitors. Herein, we review the rational and results reported from different strategies for anticancer immunotherapy carried out at different steps during the anticancer immunity cycle in the context of lung cancer.

Antigen presentation: antitumor vaccines

Rational

The purpose of tumor-associated antigen (TAA) vaccines is to promote an antitumor immune response by enhancing the presentation of TAAs by dendritic cells to naïve T cells. Antigen-specific active immunotherapies have been tested in several clinical trials with, up to now, disappointing results. Most antigen-specific immunotherapies have incorporated a single antigen and have narrow epitope specificities, which may contribute to their lack of efficacy [1]. We briefly report on the main trials below.

Outcomes in thoracic oncology

Tecemotide, a therapeutic vaccine that targets cancer cells expressing MUC1, was tested in a phase 3 trial in unresectable stage III NSCLC, previously treated with chemoradiotherapy. Tecemotide failed to significantly prolong survival in the overall population. However, tecemotide improved survival in the preplanned group of patients who were treated with concurrent chemoradiotherapy. A further phase 3 trial is initiated in this population (START2). The TG4010 vaccine also targets MUC1 and IL2. In a phase 2 study on advanced stage NSCLC, a slight albeit non-significant difference in progression-free survival (PFS) at 6 months was observed, but there was no difference in terms of overall survival (OS). A phase 2–3 study where TG4010 was added to the chemotherapy for stage IV NSCLC (TIME trial) recently shows that TG4010 was efficacious with an acceptable safety profile in patients with stage IV NSCLC, particularly in the non-squamous population [2, 3]. The percentage of CD16 + CD56 + CD69+ cells, a phenotype of activated natural killer cells, was a potential predictor of outcome of patients who received TG4010 [4]. GSK1572932A is a vaccine that targets the melanoma-associated antigen-3 (MAGE-A3) peptide. A phase 2 study on patients with resected MAGE-A3-positive NSCLC initially suggested a trend towards improved outcomes [2, 3]. Unfortunately, these results were not confirmed in a recent large phase 3 study (MAGRIT trial), which showed no difference in terms of disease-free survival and no clear impact from the biomarkers [5].

GV1001, a peptide vaccine that corresponds to the active site of human telomerase reverse-transcriptase and GM-CSF, induced specific immune responses in 80 % of patients with unresectable stage III NSCLC [6]. A gain in median overall survival was reported among immune responders in a phase 1–2 study in which GV1001 was combined with a second telomerase peptide (I540). A phase 3 study is ongoing.

The CIMAvax EGF vaccine is designed to induce an antibody-mediated immune response [7]. However, a phase 2 study failed to demonstrate a gain in median OS although a subgroup of patients with a good antibody response appeared to derive some benefit from the drug.

The Belagenpumatucel-L vaccine (LucanixTM, NovaTx Corporation, San Diego, USA) consists of NSCLC cell lines transfected with a TGF-β2 antisense gene. Despite a promising phase 2 study [5], a recent phase 3 trial on advanced stage NSCLC was negative, although it showed a potential benefit for subgroups with adenocarcinoma and who had been previously treated with radiotherapy [8].

Tergenpumatucel-L (lung cancer cell lines transfected with a murine galactosyl-transferase gene) has been associated with an interesting median OS in a phase 2 study [9].

Altogether, the results are disappointing for the whole population to date, but some positive signals are reported in subgroups and new trials are currently being conducted in more closely selected patients.

Priming: CTLA-4 inhibition

Rational

CTLA-4 regulates T cells, predominantly during their initial activation by dendritic cells and other antigen-presenting cells. CTLA-4 outcompetes the stimulatory receptor CD28 to bind to its ligands (CD80/CD86) due to its higher binding affinity and because it delivers a negative signal into the T cell, leading to inhibition of T cell activation and expansion [10]. Moreover, CTLA-4 is expressed by regulatory T cells (Treg) and can redirect T cells to the tolerogenic phenotype, CD25-Foxp3. The main function of these T cells is to maintain immune homeostasis and modulate immune responses, but they can also induce tumor proliferation [11]. CTLA4 is expressed on the surface of T lymphocytes but is also expressed in NSCLC tumors in 51–87 % of cases. CTLA4 expression was associated with non-squamous histology but is not significantly correlated with OS [12]. It is also associated with older age and poor tumor differentiation [13, 14].

Ipilimumab and tremelimumab are fully human monoclonal antibodies that target CTLA-4 and can lead to important, diffuse, and unspecific T cell activation.

Outcomes in thoracic oncology

Ipilimumab (Yervoy®) was first tested in metastatic melanoma before being evaluated for lung cancer. A phase 2 study for stage IIIB/IV NSCLC assessed the activity of ipilimumab plus paclitaxel and carboplatin. Patients were randomized to receive either paclitaxel with carboplatin and a placebo, or concurrent ipilimumab, or phased ipilimumab. Phased ipilimumab improved immune-related PFS (5.7 vs. 4.6 months with a placebo; HR 0.72, p < 0.05) and median OS (12.2 vs. 8.3 months with a placebo), but the difference in OS was not significant (p = 0.23). The PFS advantage was not observed in the concurrent ipilimumab arm [15]. Initial platinum-based chemotherapy is frequently efficient for small-cell lung cancer (SCLC), but usually fails to induce a durable response. A phase 1–2 study assessed nivolumab with or without ipilimumab to treat recurrent SCLC (CA209-032). Other objectives were safety, PFS, OS, and biomarker analysis. In 20 evaluable patients in the nivolumab-plus-ipilimumab arm, one patient had a complete response (CR), six (30 %) had a stable disease, and nine (45 %) had a partial response. Many other trials are ongoing, especially for SCLC. A phase 2 study is evaluating ipilimumab in addition to carboplatin and etoposide in a first-line setting to treat extensive stage SCLC (NCT01331525).

Two phase 2 studies (NCT01285609 and NCT01450761) have recently evaluated ipilimumab combined with paclitaxel–carboplatin or platinum–etoposide in patients with squamous NSCLC and SCLC. Results are awaited. Combinations with targeted therapies are also currently being tested in a phase 1b study (NCT01998126) where ipilimumab is associated with erlotinib or crizotinib for EGFR- or ALK-mutated stage IV NSCLC. A phase 2 trial is also assessing ipilimumab plus chemotherapy in a neo-adjuvant setting (NCT01820754).

Tremelimumab is another human IgG2 monoclonal antibody that targets CTLA-4. In a phase 2 open-label trial on advanced malignant mesothelioma that has relapsed after prior chemotherapy, the median PFS was 6.2 months and the median OS was 10.7 months [16]. This drug is being evaluated in combination with MEDI4736 vs. MEDI4736 alone vs. platinum-based chemotherapy as a first-line treatment for advanced or metastatic NSCLC (NCT02453282). We also await results from another phase 2b trial that is comparing tremelimumab as a monotherapy to a placebo for pretreated mesothelioma.

The PD-1/PD-L1 pathway: T cell expansion and activation

Rational

PD-1 is a transmembrane protein expressed on the surface of activated immune cells (T cells, B cells, dendritic cells, NK, NKT, macrophages [17]). PD-L1, its ligand, is also present on the surface of many hematopoietic (B cells, T cells, dendritic cells) and non-hematopoietic cells (epithelial cells, endothelial cells) [18]. The interaction between these two molecules constitutes a physiological checkpoint that prevents an excessive and uncontrolled immune response [19]. Ensuing inhibition of the T cell response (by limiting the secretion of cytotoxic mediators) can induce anergy or redirect T cells to a protumor T-reg FOXP3 phenotype [20, 21]. PD-L1 expression by tumor cells is a well-known mechanism of immune evasion that concerns 40–50 % of cases of NSCLC [22, 23]. This enables the tumor to be surrounded by a tolerogenic protumoral microenvironment. Blocking PD-1 or PD-L1 restores cytotoxic antitumor T cell activity and subsequently acts as an effective antitumor response [24•].

Outcomes in thoracic oncology

PD-1 inhibitors

Nivolumab (Opdivo®) was the first anti-PD1 approved for the treatment of locally advanced or metastatic squamous (EU, US) and non-squamous (US) NSCLC after prior chemotherapy [25]. This drug is a genetically engineered IgG4 monoclonal antibody that first demonstrated an interesting objective response rate (ORR) for squamous (33 %) and non-squamous (12 %) NSCLC (n = 76) in a phase 1 study. PD-L1 expression was strongly correlated with response in this study (33 vs. 0 % for PD-L1-positive or PD-L1-negative tumors, respectively) [26]. A single-arm phase 2 study, CheckMate 063, was then initiated, which tested nivolumab as a monotherapy for squamous NSCLC that had progressed after two or more previous lines of chemotherapy. An interesting objective response rate of 14.5 % was reported within this heavily pretreated population. Responses were durable, as a median response duration was not reached and 40.8 % of the population were still alive at 1 year [27].

CheckMate 017 was a phase 3 randomized trial comparing nivolumab to docetaxel in 272 pretreated squamous NSCLC patients. The ORR was significantly higher in the nivolumab arm (20 vs. 9 %). Moreover, responses appeared prolonged (9.2 months median OS vs. 6 months; median duration response was not reached in the nivolumab arm vs. 8.4 months in the docetaxel arm) and the toxicity profile was much more favorable (9 % grade 3/4 side effects vs. 71 % in docetaxel arm). PD-L1 expression (available for 83 % of patients on recent or archival tissues) was found to not be prognostic or predictive of a response, whatever the percentage of stained tumor cells in immunohistochemistry (1, 5 or 10 %). [28••]. The CheckMate 057, a study with a similar design, was then conducted on 582 cases of pretreated non-squamous NSCLC. The ORR was 19 vs. 12 % for the nivolumab and docetaxel arms, respectively, but, more interestingly, the median duration response was 17.2 vs. 5.6 months, and grade 3/4 toxicity was 10.5 vs. 53.7 %, respectively. In this study, PD-L1 expression was mostly assessed prospectively and strongly correlated with response, even when using a cutoff value of 1 % expression for tumor cells. The ORR was 9 vs. 31 % (cutoff 1 %), 10 vs. 36 % (cutoff 5 %), and 11 vs. 37 % (cutoff 10 %) for PD-L1-negative vs. PD-L1-positive tumors, respectively [29••]. CheckMate 026 (NCT02041533), a phase 3 study is recruiting patients to compare platinum-based chemotherapy and nivolumab as a first-line treatment for PD-L1-positive patients. Figure 1a, b shows an example of the response to nivolumab in a heavily pretreated patient.

Fig. 1
figure 1

Examples of tumor responses before (a) and after (b) four courses of nivolumab. Pulmonary toxicity from anti-PD1 (nivolumab) (c, d)

Full size image

Pembrolizumab (MK-3475, KEYTRUDA®) is another highly selective humanized IgG4 antibody that targets PD1 and has been approved by the FDA since October 2015 to treat PD-L1-positive metastatic NSCLC after failure of platinum-based chemotherapy. KEYNOTE-001 was a large phase 1 trial that enrolled 495 patients (394 pretreated, 101 not previously treated). The ORR was 19.5 % (18 % in the pretreated cohort, 24.8 % for untreated patients). Median duration of response was 23.3 months for previously untreated patients vs. 10.4 months for pretreated patients. Grade 3/4 adverse effects were reported in 9.5 % of patients (1.8 % pneumonitis). Strong PD-L1 expression, defined as expression of at least 50 % of tumor or tumor-infiltrating immune cells, strongly correlated with response to pembrolizumab, with a response rate of 45.2 % [30]. This justified the integration of PD-L1 expression for FDA approval. Clinical trials are ongoing to precisely identify the indications of pembrolizumab. KEYNOTE 010 is a phase 3 study comparing docetaxel versus pembrolizumab in previously treated patients. In total, 1034 patients have been enrolled: 345 received pembrolizumab 2 mg/kg, 346 pembrolizumab 10 mg/kg, and 343 docetaxel. The median OS was significantly longer for pembrolizumab 2 mg/kg (10.4 months) versus docetaxel (8.5 months) (HR 0.71, 95 % CI 0.58–0.88; p = 0.0008) and for pembrolizumab 10 mg/kg (12.7 months) versus docetaxel (HR 0.61, 0.49–0.75; p < 0.0001). Furthermore, for patients with strong PD-L1 expression (at least 50 % of tumor cells expressing PD-L1), OS was significantly longer with pembrolizumab 2 mg/kg than with docetaxel (median 14.9 vs. 8.2 months; HR 0.54, 95 % CI 0.38–0.77; p = 0.0002) and with pembrolizumab 10 mg/kg than with docetaxel (17.3 vs. 8.2 months; 0.50, 0.36–0.70; p < 0.0001). Treatment-related adverse events were less frequent with pembrolizumab than docetaxel [31]. KEYNOTE 024 (NCT02142738) is recruiting PD-L1 patients to receive either pembrolizumab or platinum-based chemotherapy as a first-line treatment for NSCLC. Finally, KEYNOTE 091 (NCT02504372) will compare pembrolizumab to a placebo in an adjuvant setting.

PD-L1 inhibitors

Atezolizumab (MPDL3280A), an antibody targeting PD-L1, showed encouraging results in a phase 1 study, leading to an ORR of 23 % and durable responses in pretreated patients. PD-L1 was predictive of a response in this study (39 % ORR and 88 % disease-control rate vs. 13 % ORR and 41 % disease-control rate in PD-L1-positive and PD-L1-negative patients, respectively) [32]. A phase 2 study (POPLAR) was then conducted, including 287 pretreated patients (interim analyses [33, 34]) randomized to receive either MPDL3280A or docetaxel. Patients were stratified by PD-L1 status (expression in immune cells and tumor cells), histology, and prior lines of treatment. Patients with the lowest PD-L1 expression (TC0 and IC0) did not derive any benefit from the treatment (ORR 8 %; median PFS 1.9 months). In contrast, strong immuno-histochemistry (IHC) staining (of TC3 or IC3) was strongly predictive of a response to atezolizumab, with an ORR of 38 vs. 13 % for docetaxel. Median PFS in this cohort was 9.7 months, and the median OS has not been reached. Atezolizumab was also better tolerated, with 11 % grade 3/4 side effects vs. 56 % in the docetaxel arm. A phase 3 study of a similar design has been recently completed (OAK, NCT02008227). Atezolizumab is also currently being evaluated in two single-arm phase 2 studies (FIR and BIRCH, NCT01846416 and NCT02031458) in PD-L1-positive patients. Preliminary results from the BIRCH trial were reported at the European Cancer Congress 2015 [35]. The ORR was 26, 24, and 27 % for tumors showing strong expression to PD-L1 (TC3 or IC3) in first-, second-, and third-line treatments, respectively (ORRs were 19, 17, and 17 % for TC2/3 or IC2/3 PD-L1 expression). The median response duration has not been reached, as most responses are still ongoing. Treatment-related overall adverse effect rate is 65 %, with only 11 % grade 3/4 adverse events.

MEDI4736 (durvalumab) is another PD-L1 inhibitor that achieved a 13 % ORR in a phase 1 study [36]. This drug is still undergoing ongoing phase 3 evaluation (ATLANTIC, NCT 02087423) as a monotherapy for heavily pretreated patients (at least two lines), with multiple arms, based on PD-L1 level of expression and mutational profile. The PACIFIC trial (NCT 02125461) is a phase 3 study that includes randomized patients receiving either durvalumab or a placebo after chemoradiation for stage III unresectable NSCLC. There are strong signals for a synergistic effect in this context, as radiation therapy may upregulate PD-L1 expression of tumor cells as a mechanism of resistance, which can be overcome by the PD-L1 blockade [37].

Table 1 summarizes the main results reported in clinical trials that have evaluated these anti-PD1 and anti-PD-L1 drugs.

Table 1 Results reported in the main clinical trials that have evaluated the PD1 and PD-L1 inhibitors
Full size table

Combinations of drugs

Even though immunotherapies have activities as single agents, more impressive activities can be seen when they are combined with other agents.

Chemotherapies and immunotherapies

Several conventional chemotherapies inducing immunogenic tumor-cell death can enhance a strong adaptive immune response. Indeed, tumor cells are converted into an anticancer vaccine (release of endogenous tumor antigens), and their effects may be enhanced by immune checkpoint inhibitors [38]. Numerous clinical trials are ongoing and are testing such combinations for SCLC and NSCLC.

Anti-PD-1/PDL-1 and anti-CTLA4 antibodies

Because of distinct levels of intervention, T cell activation, and expansion in the periphery of CTLA4 and CTL-effector functions in the tumor microenvironment of PD-1, it seems logical to combine these two drugs to enhance antitumor immune activity [39]. This has been demonstrated in melanoma. In lung cancer, the combination of ipilimumab plus nivolumab has shown response rates of 11–33 % depending on histology, but induces frequent and sometimes severe immune-related adverse effects [40].

Immunotherapy and targeted therapy

Activating the patient’s immune system during the time of tumor reduction and remission may be the best way to ensure that responses are converted into long-term and durable benefits. Unlike conventional chemotherapies, targeted therapies for EGFR-mutated or ALK-rearranged adenocarcinoma may achieve rapid and significant tumor shrinkage without the need for immunosuppression induced by chemotherapies. Moreover, oncogenic EGFR signaling remodels the tumor microenvironment to trigger an immune escape and mechanistically links the treatment response to PD-1 inhibition [41]. A first-line therapy combining nivolumab and erlotinib showed excellent response rates, even in EGFR–TKI pretreated patients, but with a grade 3–4 incidence of adverse events of 24 % [27]. Other trials that have combined, for example, gefitinib and tremelimumab in EGFR-mutated patients who progress under EGFR–TKI, or alectinib and atezolizumab in ALK-positive patients, are ongoing.

Biomarkers

Only a minority of patients currently benefit from active immunotherapy. Patient selection is therefore crucial. Strong arguments exist to consider PDL-1 expression in IHC analysis as a potential predictive biomarker for the response to drugs that target the PDL-1/PD-1 checkpoint. In the MPDL3280A phase 1 study, ORR was 46 % (6/13) in patients with a PDL-1 IHC score of 2 or 3, and 83 % (5/6) in those with a PDL-1 IHC score of 3 [42]. MEDI4736 response rate was also strongly correlated with PDL-1 expression (39 vs. 5 % ORR [36]). Response rates in the KEYNOTE-001 study (pembrolizumab) were 37, 17, and 10 % for strong-positive, weak-positive, or negative PDL-1 expression, respectively [43]. PD-L1 expression was found to be potent at discriminating responders from non-responders in the landmark phase 1 trial with nivolumab, and in the CheckMate 057 trial (for details, see above). The predictive value of PD-L1 expression was predefined as an additional endpoint in the CheckMate 017 and 057 studies, but the assay was not standardized and was conducted retrospectively on recent or archival tissue.

PD-L1 expression can vary with the microenvironment of the tumor. Thus, PD-L1 expression at a single time point may not represent a dynamic immune response. Overall, PD-L1 expression appears to be an imperfect marker, as the optimal test has not yet been clearly defined although constitutes an interesting basis for selecting patients. PD-L1 screening should be probably used in the future to select patients, especially in first-line settings, as chemotherapy (which can be associated with bevacizumab and maintenance therapy) has a response rate and a PFS that can reach 35 % [44] and 7.4 months [45], respectively. In contrast, in heavily pretreated patients who have less therapeutic options, biomarkers may be dispensable, as the response rates among PD-L1-negative patients remain higher than those observed in second- or multiple-line settings.

Other molecular biomarkers could also be used to help select the best candidates. Exome analysis of tumors from patients treated with another PD-1 inhibitor, pembrolizumab, showed that the best responses to PD-1 blockage were observed when there was a high mutation burden [46]. High molecular transversion associated with smoking is particularly associated with improved efficiency of pembrolizumab. These correlations with tobacco and the KRAS mutation have been confirmed among all stages of combined populations [47, 48]. This can be explained by the increase in tumor-specific antigens, such as ACE, KRAS, p53, and hTERT, from the frequently overexpressed or mutated genes observed in lung cancer, with these constituting targets for an efficient immune antitumor response.

Toxicity

Targeting immune checkpoints induced the emergence of a new form of toxicity. The side effects are less frequent and severe than the ones observed with chemotherapy and essentially concerned the lung (pneumonitis), the skin (rash), the gastrointestinal tractus (diarrhea and colitis), the liver (hepatitis), the kidneys (renal insufficiency), and the endocrine glands (hypophysitis and hypothyroidism) [49]. A rare but potentially fatal inflammatory pneumonitis has been observed in a few cases: pneumonitis was reported in 6 % of a group of patients (8/129) [50]. Figure 1c, d shows an example of lung toxicity. Pulmonary toxicity needs to be recognized and treated early. There is no validated recommendation for pneumonitis management, which is usually guided by clinical experience and observational reports [15, 51, 52]. Patients presenting with signs and symptoms of pneumonitis (dyspnea, cough, hypoxia, and interstitial radiological syndrome) must have CT scan. This scan can confirm pneumonitis (extensive, patchy, bilateral, peri-bronchial consolidation) or orient towards a differential diagnosis (infection, heart failure, tumor progression, pulmonary embolism). A bronchoscopy with an oriented broncho-alveolar lavage should be considered, including a cell count, analysis of T cell subsets, and microbial analyses. For grade 1 pneumonitis (according to CTCAE 4.0), it is possible to consider using steroids and to repeat the CT scan 3 weeks later. For grade 2 pneumonitis, it is recommended to delay immunotherapy and introduce prednisolone (1–2 mg/kg) [53]. For grade 3/4 pneumonitis, immunotherapy should be discontinued permanently and high doses of intravenous steroids (methylprednisolone 1 g/day) should be administered along with oxygen and ventilatory support. In cases of failure after 48 h or serious evolution, prophylactic antibiotics and additional immunosuppressive medication (infliximab, mycophenolate mofetil, or cyclophosphamide) can be discussed even if data are lacking to validate the use of such drugs [53].

Conclusion

After decades of frustration with negative results from antitumor vaccines for lung cancer, antitumor immunotherapy has finally entered the therapeutic arsenal to treat lung cancer after the positive results from clinical trials that have assessed immune checkpoint inhibitors. The PD1 and PD-L1 inhibitors are becoming a new standard of care for previously treated and advanced NSCLC. Nevertheless, if these drugs are well tolerated and offer durable responses, only a limited population can benefit. More fundamentally, translational and clinical research is needed to improve the selection of patients (biomarkers) and to find the best combinations (monotherapy, with chemotherapy, with a targeted therapy) to further improve their efficiency.