Introduction

The non-small cell lung cancer (NSCLC) treatment landscape has completely changed with the coming of age of immunotherapy. The past decade has seen great advances in the treatment of patients with advanced lung cancer, with the first clinically relevant improvements in overall survival (OS) occurring in the later lines of treatment followed by significant advances in the first-line setting that have upended the treatment algorithm [1].

Upon progression to a platinum chemotherapy doublet, docetaxel or pemetrexed monotherapy was standard regimen for patients with advanced NSCLC. The addition of an anti-angiogenic agent (ramucirumab or nintedanib) to docetaxel increased OS in patients with lung adenocarcinoma who had progressed following first-line chemotherapy [2, 3]. The greatest survival improvement and reduction of tumour volume were reported in patients with particularly poor prognosis who rapidly progressed while receiving first-line chemotherapy and who had a larger tumour volume [4,5,6].

Also, in the post-chemotherapy setting, treatment with immunotherapy agents targeting PD-1/PD-L1 increased survival compared with docetaxel monotherapy, as second-line treatment (nivolumab) [7, 8], and as second- and third-line treatments (pembrolizumab and atezolizumab) [9,10,11,12]. Higher PD-L1 tumour expression correlated with a better outcome to immunotherapy in patients with adenocarcinoma [7,8,9,10, 12]. Analyses requested by the European regulatory authorities for nivolumab reported a greater risk of early mortality when receiving nivolumab compared with docetaxel in the group of patients with adenocarcinoma histology, rapid progression/refractoriness to a first-line chemotherapy, ECOG performance status (PS) 1, and low PD-L1 expression [13]. Similar analyses are not available for pembrolizumab or atezolizumab.

The greater effectiveness of immunotherapy conditioned by PD-L1 expression and tumour mutational burden (TMB) [14, 15] and the better outcomes in patients with rapid progression to first-line chemotherapy observed with anti-angiogenic agents in combination with docetaxel [2,3,4] appear to indicate that both therapeutic approaches could be complementary. Of note, there have been no randomized clinical trials directly comparing single-agent immunotherapy with the combination of an anti-angiogenic agent and docetaxel; cross-trial comparisons have been performed using the control arm of docetaxel monotherapy. This suggests that post-chemotherapy treatment and its subsequent therapeutic sequence require careful consideration.

On the other hand, several immunotherapy trials reported in 2017 and 2018 transformed the first-line treatment of metastatic NSCLC [1]. The preferred first-line treatment option for patients without targetable mutations is now an immunotherapy-based regimen alone or in combination with chemotherapy.

In patients with tumour PD-L1 expression ≥ 50%, the first-line recommendation is pembrolizumab alone, based on the KEYNOTE-024 Phase 3 study reporting a significant improvement in OS with pembrolizumab compared with chemotherapy [after median follow-up of 25.2 months, median OS [mOS] of 30.0 months with pembrolizumab, and 14.2 months with chemotherapy, HR 0.63 (95% CI 0.47–0.86)] [16, 17].

Pembrolizumab alone may also be an alternative for patients with PD-L1 expression of 1–49% who cannot receive chemotherapy, based on a significant improvement in OS with pembrolizumab compared with chemotherapy in the Phase 3 KEYNOTE-042 trial in patients with tumour PD-L1 expression of ≥ 1% [after median follow-up of 14 months, mOS of 16.4 months vs. 12.1 months, HR 0.82 (95% CI 0.71–0.93)] [18, 19]. However, this improvement in the overall population appeared to be driven by patients with PD-L1 expression ≥ 50% (mOS of 20.0 months vs. 12.2 months [HR 0.70 (95% CI 0.58–0.86)]), representing 47% of the overall population [18, 19].

The Phase 3 CheckMate 227 trial evaluating first-line ipilimumab (anti-CTLA-4) plus nivolumab vs. platinum-based chemotherapy met its co-primary endpoint of significantly prolonging PFS in patients with NSCLC with high TMB, defined as ≥ 10 mutations per megabase (mut/Mb) [15]. Median PFS (mPFS) in the immunotherapy combination arm was 7.2 months compared with 5.5 months (HR 0.58 [97.5% CI 0.41–0.81]; p < 0.001). This PFS benefit in the high TMB population was observed irrespective of PD-L1 expression. Preliminary results indicated a tendency toward improved OS in the high TMB population treated with the immunotherapy combination with a mOS of 23.0 months compared with 16.4 months in patients receiving chemotherapy [HR 0.79 (95% CI 0.56–1.10)] [20].

Four randomized studies in patients with non-squamous NSCLC showed that the addition of a PD-1/PD-L1 checkpoint inhibitor to standard chemotherapy is superior to standard chemotherapy alone.

In patients without targetable mutations, first-line pembrolizumab in combination with platinum–pemetrexed was more effective than chemotherapy alone [21, 22]. In the Phase 3 KEYNOTE-189 trial of non-squamous NSCLC, OS improved in the pembrolizumab/chemotherapy arm compared with the chemotherapy alone arm (with 18.7-month median follow-up, mOS was 22.0 months vs. 10.7 months [HR 0.56 (95% CI 0.45–0.70); p < 0.00001) [23]. The OS benefit obtained with pembrolizumab/chemotherapy was independent of PD-L1 expression and was also reported in patients with tumors with PD-L1 < 1% [HR 0.52 (95% CI 0.36–0.74)]. The co-primary endpoint of PFS also improved in the pembrolizumab arm.

Three Phase 3 studies of atezolizumab also showed survival improvement independently of PD-L1 expression. In IMpower150 in patients with non-squamous NSCLC, both OS and PFS improved in the four-drug combination arm of atezolizumab plus carboplatin paclitaxel and bevacizumab (ACPB) compared with a control chemotherapy arm without atezolizumab; mOS of 19.2 months was reported in the atezolizumab arm compared with 14.7 months in the control arm (HR 0.78 [95% CI 0.64–0.96]; p = 0.02) [24]. IMpower130 evaluated atezolizumab plus carboplatin and nab-paclitaxel vs. a control chemotherapy arm without atezolizumab. OS similarly improved in the atezolizumab arm, with a mOS of 18.6 months compared with 13.9 months in the control arm (HR 0.79 [95% CI 0.64–0.98]; p = 0.033) [25]. IMpower132 evaluated the combination of atezolizumab plus carboplatin or cisplatin with pemetrexed vs. a control chemotherapy arm without atezolizumab in patients with advanced non-squamous NSCLC. PFS improved in the atezolizumab arm [mPFS 7.6 vs. 5.2 months (HR 0.6; 95% CI 0.49–0.72; p < 0.0001)], but the improvement in OS did not reach statistical significance [mOS 18.1 vs. 13.6 months (HR 0.81; 95% CI 0.64–1.03; p = 0.0797)], albeit not being fully mature [26].

Consolidation immunotherapy with durvalumab following chemoradiotherapy significantly improved the outcome of patients with locally advanced, inoperable Stage III NSCLC in the Phase 3 PACIFIC study evaluating durvalumab vs. placebo [27, 28]. OS and PFS were both significantly prolonged in the durvalumab arm compared with the placebo arm [OS HR 0.68 (99.73% CI 0.47–0.997); p = 0.0025 and mPFS 17.2 vs. 5.6 months, HR 0.51 (95% CI 0.41–0.63)].

These clinical trials changed the treatment algorithm of patients with advanced wild-type EGFR, ALK, ROS1, and BRAF lung adenocarcinoma. Immunotherapy-based regimens are standard first-line treatments for advanced disease or for chemoradiotherapy consolidation in locally advanced disease. It is important to note that all clinical trials were performed using a control arm that is no longer considered a standard regimen and all second-line trials enrolled an immunotherapy-naïve patient population that no longer exists. Consequently, there is great uncertainty in the clinic as to how to best treat patients who progress after receiving immunotherapy. The new therapeutic scenario that results from great advances in the standard of care, although having the potential to bring the best outcomes to patients for upfront treatment, does not address the best treatment to administer upon progressive disease. There is a need to optimize the sequential treatment of currently available regimens to further improve the survival outcome of patients with advanced wild-type lung adenocarcinoma [1].

For issues that frequently occur in routine clinical practice, but for which there is not enough evidence, a systematic approach that aims to generate a consensus among experts can provide reliable guidance. The Delphi technique is commonly used to develop guidelines with health professionals [29]. The Delphi process allows panelists to reach an agreement by iteratively urging them to consider a series of problematic issues about key health decisions in conditions of low-grade evidence and is considered an appropriate tool to measure agreements and delineate therapeutic options for advanced wild-type lung adenocarcinoma beyond the first-line setting.

Materials and methods

Objectives

We aimed to develop a consensus document using the Delphi technique on treatment options in Spain beyond the first-line setting for patients with advanced lung adenocarcinoma without targetable mutations (wild-type EGFR, ALK, ROS1, and BRAF). An external expert in Delphi methodology guaranteed the quality of the overall process.

Study design

Scientific committee

In April 2018, a scientific committee of five experts in comprehensive lung cancer management met to define and lead this consensus project. The following steps were carried out by the scientific committee: (1) selection of expert panel; (2) generation of clinical statements based on the extensive narrative review of the current medical literature; (3) definition of the consensus levels and agreement on methodology; (4) interpretation of the results from the Delphi rounds; (5) final wording in the consensus document.

Expert panel

A panel of 26 medical oncologists specialized in lung cancer and renowned for their clinical and research experience were selected by the scientific committee. In this selection of an expert panel, a balanced territorial representation of Spain was considered.

Delphi

Generation of statements

After an extensive narrative literature review, 33 statements belonging to 4 major domains were drafted by the scientific committee. These domains covered general aspects and three clinical scenarios according to disease stages and treatment line:

  1. 1.

    General aspects of the current treatment of advanced wild-type lung adenocarcinoma (four statements)

  2. 2.

    Durvalumab as consolidation therapy after platinum-based chemoradiotherapy: impact on subsequent treatment lines (six statements)

  3. 3.

    First-line immunotherapy ± chemotherapy: impact on optimal treatment sequencing in second- and further treatment lines (11 statements)

  4. 4.

    Treatment after platinum-based chemotherapy without immunotherapy: options and optimal therapeutic sequencing (12 statements)

It was established that these statements would be evaluated in patients with wild-type lung adenocarcinoma with a good general condition (ECOG PS 0–1) without contraindications to any type of treatment. Non-immunotherapy treatment regimens included in the statements were limited to those approved and reimbursed by the Spanish National Health System. These prerequisites were included to simplify the interpretation and allow for a better clinical applicability of any consensus.

Consensus levels

The expert panel completed two Delphi rounds of the statements through an online platform. In the first round, the 26 panelists voted using the following answers: (1) “strongly disagree”; (2) “basically disagree”; (3) “doubtful”; (4) “basically agree”; and (5) “strongly agree”. A consensus existed if ≥ 70% of the responses were classified as “strongly agree” or “strongly disagree”. If the consensus was 100%, it was considered “unanimous”. All statements reaching consensus in the first round did not undergo the second Delphi round. The statements that had ≥ 15% of “doubtful” responses were reassessed by the scientific committee, to discern if this was due to the ambiguity of the statement itself, in which case it was reformulated, or due to the issue being controversial in itself, in which case the statement remained unchanged in the second Delphi round.

A second Delphi round was held for all statements without consensus in the first round. In this second round, the following answers were available: (1) “strongly disagree”; (2) “basically disagree”; (3) “basically agree”; and (4) “strongly agree”. The reduction to four answers in this second round is contemplated in the Delphi methodology to prompt experts to decide in favour or against a statement and avoids the tendency to give a neutral answer. A consensus was defined if ≥ 70% of the responses were in categories 1 and 2 (consensus in the disagreement) or in categories 3 and 4 (consensus in the agreement), and the consensus was considered unanimous if 100% was reached in either of the two categories. A majority was defined when the percentage of agreement or disagreement was ≥ 60% but < 70%. A dissent was defined as < 60% agreement.

All 26 panelists responded to both Delphi rounds. The first round took place from 11 July 2018 to 21 August 2018 and the second round took place from 7 September 2018 to 17 October 2018. This study consisted of a survey of expert opinions and no patient data were collected, so no specific independent ethical or research review or approval was necessary.

Statistical analysis

Categorical and ordinal variables were summarized with counts and percentages tabulated by round. The SPSS v25 (IBM) program package was used for the statistical analyses.

Development of the consensus manuscript

All five members of the scientific committee wrote the consensus manuscript, which was then enriched with feedback from three external experts (Enriqueta Felip, Pilar Garrido, and Luís Paz-Ares).

This project began in April 2018 and ended in March 2019. During this time and when this article was drafted, newly published literature and congress presentations were reviewed to analyze the clinical implications of any new data in patients with advanced NSCLC and to provide support for the consensus statements.

Results

In the first Delphi round, 33 statements were evaluated, and a consensus was reached for 2 statements (6%). In the second Delphi round, the remaining 31 statements were evaluated, including 1 statement that had been reformulated: in 7 statements, there was unanimity in the agreement and in 22 statements, there was consensus in the agreement (overall consensus: 94%). For one statement, there was majority in disagreement and in one statement, there was dissent.

General aspects of the current treatment of advanced wild-type lung adenocarcinoma

Table 1 shows the responses for the four statements in this domain. There was a high degree of consensus (92–100%) on the fact that the incorporation of new highly effective immunotherapy-based treatments as first-line treatment of advanced wild-type lung adenocarcinoma has generated uncertainty regarding optimal treatment upon disease progression. It is not possible to generate high-level scientific evidence because the second-line Phase 3 studies leading to regulatory approvals will not be repeated with patient populations that have received the current standard of first-line immunotherapy. In this case, expert opinion and clinical experience are necessary to define what is the best therapeutic sequence to offer patients who have reported disease progression.

Table 1 General aspects of the current treatment of advanced wild-type lung adenocarcinoma
Full size table

Durvalumab as consolidation therapy after platinum-based chemoradiotherapy in Stage III wild-type lung adenocarcinoma: impact on subsequent treatment lines

Table 2 shows the responses for the six statements in this domain. There was high agreement (85–96%) that the following variables may define the optimal regimens to be administered upon progression to durvalumab in the locally advanced setting: progression-free interval, clinical aggressiveness of the progression (early progression and high symptomatic tumour load), accumulated toxicity from previous treatments, tumour PD-L1 expression, and TMB.

Table 2 Durvalumab as consolidation therapy after platinum-based chemoradiotherapy in Stage III wild-type lung adenocarcinoma: impact on subsequent treatment lines
Full size table

Patients treated with platinum-based chemoradiotherapy progressing during durvalumab consolidation, can be classified as refractory to platinum when the interval from the end of chemoradiotherapy to progression is less than 6 months. For this reason, the combination of docetaxel and nintedanib could be an appropriate therapeutic option. If this disease-free interval is longer than 12 months, during treatment with durvalumab or when it has already ended, a platinum-based chemotherapy would be an appropriate therapeutic option. If the interval is between 6 and 12 months, the decision could be based on the proximity to 6 vs. 12 months, the clinical aggressiveness of the progression (high symptomatic tumour burden) and reports of previously accumulated toxicity.

There is not much evidence that rechallenging patients with immunotherapy-based treatment is effective [30,31,32]. Moreover, there is no evidence for rechallenging patients after chemoradiotherapy consolidation with durvalumab, so patients should be encouraged to participate in clinical trials. In any case, an immunotherapy rechallenge might be justified with a longer interval between the end of durvalumab and the progression of the disease and with higher tumour PD-L1 expression and high TMB.

First-line immunotherapy ± chemotherapy in advanced wild-type lung adenocarcinoma: impact on optimal treatment sequencing in second- and further treatment lines

Table 3 shows the responses for the 11 statements in this domain. A high degree of consensus (88–100%) was established, confirming the overall acceptance that new first-line immunotherapy-based treatments in routine clinical practice will reduce its use in later lines. There was a shared concern for the resulting economic impact.

Table 3 First-line immunotherapy ± chemotherapy in advanced wild-type lung adenocarcinoma: impact on optimal treatment sequencing in second- and further treatment lines
Full size table

After progression to first-line platinum, pemetrexed, and pembrolizumab, the administration of docetaxel and nintedanib may be the most appropriate therapeutic option. After progression to first-line carboplatin, paclitaxel, bevacizumab, and atezolizumab, the combination of docetaxel and nintedanib could be a therapeutic option. Single-agent pemetrexed is also a reasonable therapeutic option as it has demonstrated efficacy in the second-line setting, but with a more favourable safety profile than docetaxel alone [33] and may be a preferred option for disease with low aggressiveness. In patients experiencing early disease progression to first-line treatment and/or a high symptomatic tumour load, the combination of docetaxel and nintedanib may be more appropriate. On the other hand, after progression to a first-line immunotherapy regimen without chemotherapy, the most appropriate second-line therapeutic option would be platinum-based chemotherapy.

Upon progression to second-line platinum-based chemotherapy after first-line immunotherapy, a third-line regimen should consider treatment with the combination of docetaxel and nintedanib, or single-agent docetaxel or pemetrexed. The following variables are important when selecting third-line treatment: previous treatments, aggressiveness of disease progression (early progression and a high symptomatic tumour load), and accumulated toxicity. Higher aggressiveness of the disease would favour the use of the docetaxel and nintedanib combination.

In this clinical context, there is no scientific evidence that the reintroduction of an immunotherapy-based treatment is effective, so it should be considered only in a clinical trial setting.

Treatment after platinum-based chemotherapy without immunotherapy in advanced wild-type lung adenocarcinoma: options and optimal therapeutic sequencing

Table 4 shows the responses for the 12 statements in this domain. A high degree of consensus was attained in 10 of the 12 statements (88–100%). One statement required reformulation, one statement had a majority in disagreement, and one statement was in dissent; reflecting the uncertainties of treating patients in this clinical scenario.

Table 4 Treatment after platinum-based chemotherapy without immunotherapy in advanced wild-type lung adenocarcinoma: options and optimal therapeutic sequencing
Full size table

Statement 1 (“tumour PD-L1 expression is a fundamental criterion to help therapeutic decision making in a patient previously treated with first-line platinum-based chemotherapy”) obtained a high proportion of “doubtful” responses (23.1%) in the first Delphi round. The scientific committee considered that the word “fundamental” could lead to confusion, so it was eliminated in the second Delphi round, achieving consensus.

The following variables should be considered when deciding second-line treatment after a first-line chemotherapy regimen: aggressiveness of disease progression (early progression and a high symptomatic tumour load), tumour PD-L1 expression, previous treatment, and accumulated toxicity from previous treatments.

In patients with aggressive disease progression with low or negative tumour PD-L1 expression, the combination of docetaxel and nintedanib could be a reasonable therapeutic option as a second-line treatment. On the other hand, second-line immunotherapy could be a reasonable therapeutic alternative in patients with a negative or unknown tumour PD-L1 expression in patients with slow disease progression.

At the time of progression to second-line immunotherapy, the third-line combination of docetaxel and nintedanib can be considered. Similarly, after a progression to second-line chemotherapy, the administration of third-line immunotherapy should be evaluated considering the aggressiveness of the progression to previous chemotherapy and tumour PD-L1 expression.

Statement 9 (“upon progression to second-line chemotherapy [docetaxel and nintedanib, docetaxel alone, or pemetrexed alone], one could consider administering immunotherapy as third-line treatment, but exclusively in patients with positive tumour PD-L1 expression”) obtained a majority in disagreement (61.5%). The point of disagreement was interpreted to be related to “exclusively patients with positive tumour PD-L1 expression”. If the disease progression to a second-line chemotherapy was not aggressive, the administration of third-line immunotherapy could be evaluated independently of tumour PD-L1 expression.

Statement 12 (“in a patient with negative tumour PD-L1 expression treated with first-line platinum-based chemotherapy plus bevacizumab attaining a disease control greater than 18 months, one could consider administering a second-line anti-angiogenic therapy [docetaxel and nintedanib] because of the benefit from the previous antiangiogenic”) obtained a dissent (42% disagree vs. 58% agree). Experts who frequently use a first-line combination of chemotherapy with bevacizumab to achieve a prolonged PFS, especially when tumour PD-L1 expression is negative, prefer to continue treatment at the time of progression with an anti-angiogenic with a different mechanism of action, such as the combination of docetaxel and nintedanib [34]. On the other hand, experts without extensive experience with bevacizumab value that when such a prolonged PFS is obtained with first-line chemotherapy, the most reasonable therapeutic option in the second-line setting is immunotherapy regardless of the negative tumour PD-L1 expression. The results from the LUME-Lung 1 study of docetaxel and nintedanib in patients with advanced NSCLC support the efficacy in all patients with adenocarcinoma; however, there is lower benefit in long-term progressors, particularly those patients who progressed at least 12 months after initiation of first-line therapy.

Discussion

The current Delphi study shows that a high degree of consensus exists among experts on how standard first-line immunotherapy-based regimens may be affecting decisions regarding subsequent treatment of patients with advanced wild-type lung adenocarcinoma. This consensus document complements the information in the European Society of Medical Oncology (ESMO), Sociedad Española de Oncología Médica (SEOM), National Institute for Health and Care Excellence (NICE), and other guidelines [35,36,37], and aims to provide oncologists with a specific therapeutic decision process to optimize management of patients when their disease has progressed after receiving immunotherapy.

As the treatment of NSCLC further evolves, it is important to note that the following relevant scientific information was communicated after the second Delphi round took place:

  1. 1.

    The use of durvalumab as a consolidation therapy after platinum-based chemoradiotherapy was limited by the European Medicine Agency (EMA) to patients with tumours expressing PD-L1 ≥ 1%. This restriction is a cause for concern among experts as it is judged to be based on non-robust analysis [38].

  2. 2.

    In an updated analysis of CheckMate 227 trial, differences in OS between the nivolumab and ipilimumab arm and the chemotherapy arm were not statistically significant in the subgroup of patients with TMB ≥ 10 mut/Mb, with a HR of 0.77 (95% CI 0.56–1.06) [39].

  3. 3.

    Two Phase 3 first-line studies of atezolizumab plus chemotherapy in non-squamous NSCLC reported improved PFS outcomes compared with the control arms [25, 26]. OS improved in one study [25], but not the other one, although not being yet fully mature [26].

After careful analysis of this new information, the scientific committee agreed that, despite its relevance, it did not affect the consensus statements or the elaboration of this document.

Although this consensus document aims to help therapeutic decision making, it has several limitations. First, this Delphi project has been developed under the premise of patients with advanced wild-type lung adenocarcinoma with a good general condition (ECOG PS 0 and 1) and without medical contraindications. This may limit the potential applicability of the Delphi consensus to all patients with advanced wild-type lung adenocarcinoma. This consensus document has been generated due to the lack of scientific evidence and with the realistic assumption that completed Phase 3 studies will not be repeated in a population that has received immunotherapy. We would encourage to use this document to help stimulate discussion on future real-world studies that could be carried out to support or question the consensus statements.

Several studies have suggested that treatment with immunotherapy does not condition the efficacy of subsequent chemotherapy. On the contrary, these studies seem to suggest an increase in the efficacy of chemotherapy [40,41,42,43,44], particularly when combined with anti-angiogenic agents [45,46,47], when administered after treatment with immunotherapy. Synergistic activity in both preclinical [48, 49] and clinical models [21, 22, 25, 26] support a hypothesis of a possible chemosensitization after prior exposure to immunotherapy. The changes induced by prior immunotherapy, as well as the “prolonged tissue half-life” of immunotherapy agents [50], could increase the efficacy of chemotherapy. The efficacy of chemotherapy in this setting, may be further increased when combined with an anti-angiogenic agent [51, 52].

There is little scientific evidence that rechallenging patients with immunotherapy-based treatment is effective [30,31,32]. Currently, immunotherapy rechallenge should be considered only in the setting of a clinical trial.

In summary, due to the lack of scientific evidence, a panel of experts has developed a consensus document via a Delphi process that can help therapeutic decision making for the optimal management of second-line treatment as well as subsequent treatment lines of patients with advanced wild-type lung adenocarcinoma. These decisions should carefully evaluate previous treatment(s) administered and the PFS obtained, aggressiveness of the disease progression (early progression and a high symptomatic tumour burden), accumulated toxicity, tumour PD-L1 expression, and TMB (Table 5). We hope that dedicated research efforts can generate missing real-world evidence to fill data gaps in the clinical frameworks covered in this consensus document.

Table 5 Variables for informing treatment decisions in patients with wild-type lung adenocarcinoma
Full size table