Clinical and Translational Oncology (2021) 23:1304–1313 https://doi.org/10.1007/s12094-020-02518-0 RESEARCH ARTICLE Treatment strategy optimization for patients with non‑small‑cell lung cancer harboring EGFR mutation: a Delphi consensus D. Isla1 · J. de Castro2 · R. García‑Campelo3 · M. Majem4 · D. Vicente5 · O. Juan‑Vidal6 Received: 2 October 2020 / Accepted: 24 October 2020 / Published online: 18 November 2020 © The Author(s) 2020 Abstract Aim To stablish a consensus on the treatment strategy for advanced non–small-cell lung cancer (aNSCLC) with epidermal growth factor receptor mutation (EGFRm) in Spain. Methods After a systematic literature review, the scientific committee developed 33 statements in 4 fields: molecular diagno- sis (10 items); histologic profile and patient clinical characteristics (7 items); first-line (1L) treatment in EGFRm aNSCLC (8 items); and subsequent-line treatment (8 items). A panel of 31 experts completed 2 Delphi online questionnaires rating their degree of agreement/disagreement for each statement through a 1–9 range scale (1–3 = disagree, 7–9 = agree). Consensus was reached if 2/3 of the participants are in the median range. Results In the first Delphi round consensus was achieved for 24/33 of the statements. One of the assertions was deleted, proceeding to a second round with the eight remaining questions with no consensus or in the range of indeterminacy. Deter- mination of the EGFR status from tissue and analysis of the different biomarkers are two important variables that influenced treatment decision in patients with aNSCLC. 1L treatment should be the best therapeutic option, independently of the sub- sequent lines of treatment. For patients with the most common activating mutations osimertinib was considered the most efficient and safe 1L option. In case of disease progression, a new biopsy was needed. Conclusions A consensus document is proposed to optimize the treatment strategy for untreated patients with a NSCLC with EGFR sensitizing mutations. Keywords Delphi · Non–small-cell lung cancer · Epidermal growth factor receptor · Tirosine kinase inhibitors Introduction Non-small cell lung cancer (NSCLC) account for 85%–90% of lung cancer, being adenocarcinoma the most common * O. Juan‑Vidal subtype. Epidermal growth factor receptor (EGFR) muta- ojjuanv@seom.org tions are found in ∼10%–12% of Caucasians with adeno- 1 Medical Oncology Department, University Hospital carcinoma and are more frequent in never smokers, females Lozano Blesa, IIS Aragón, Avda. San Juan Bosco, 15, and in patients of East Asian ethnicity [1, 2]. EGFR muta- 50009 Zaragoza, Spain tion (EGFRm) testing is recommended in all patients with 2 Medical Oncology Department, University Hospital La advanced non-squamous cell carcinoma, it is not recom- Paz-IDIPAZ, Madrid, Spain mended in patients with an unequivocal diagnosis of squa- 3 Medical Oncology Unit, University Hospital A Coruña, mous cell carcinoma with the exception of never/former light A Coruña, Spain smokers (< 15 pack years). The discovery of EGFRm and 4 Medical Oncology Unit, Hospital Santa Creu I Sant Pau, development of EGFR tyrosine kinase inhibitors (TKIs) have Barcelona, Spain achieved a paradigm shift in treatment strategy of advanced 5 Medical Oncology Unit, University Hospital Virgen NSCLC (aNSCLC). Several randomized phase III studies Macarena, Sevilla, Spain have revealed that treatment with first- or second-generation 6 Thoracic Oncology Unit, Department of Medical Oncology, EGFR-TKIs results in an improved progression-free survival Hospital Universitari I Politècnic La Fe, Av. Fernando Abril (PFS) compared to standard chemotherapy in treatment Martorell 106, 46026 Valencia, Spain 13 Vol:.(1234567890) Clinical and Translational Oncology (2021) 23:1304–1313 1305 naïve patients with aNSCLC [3]. Results from direct com- mittee. In this selection of an expert panel, a balanced ter- parison of first-, second-, and third-generation EGFR TKIs ritorial representation of Spain was considered. in this population have also been reported. A significant PFS and overall survival (OS) benefit has been demonstrated for osimertinib and dacomitinib [4–6] compared with gefitinib Delphi or erlotinib. The optimal treatment sequence for patients with EGFRm aNSCLC continues to evolve, related largely Generation of Statements to an increasing number of breakthroughs and studies in this field. After a narrative literature review, 33 statements organ- Although the efficacy of EGFR-TKIs in the treatment of ized in 4 major domains were drafted by the scientific EGFRm aNSCLC patients is well established, the treatment committee. These domains aimed to cover from the key decision-making process is becoming more complex as our diagnostic criteria to different treatment lines according knowledge of EGFRm and resistance pathways grows and to the patient profile: more treatment options become available. Thus, treating physicians must consider an increasing number of factors, 1. Molecular diagnosis on aNSCLC (10 statements); facing real world challenges. Indeed, a consensus devel- 2. histologic profile and patient clinical characteristics (7 oped by experts could help to guide decisions and issues statements); that can occur in clinical practice in case there is not enough 3. 1L therapy in EGFRm aNSCLC (8 statements); evidence. 4. second and subsequent-lines treatment in EGFRm We have establish a consensus about the decision criteria aNSCLC (8 statements). for selecting the best first-line (1L) therapy, the importance of a second biopsy on disease progression to determine the It was established that these statements would be evalu- most appropriate subsequent therapies using the Delphi ated in patients with EGFRm aNSCLC with good gen- Technique method [7]. eral condition (ECOG PS 0 and 1) and without medical contraindications to receive treatment. Only reimbursed treatments in Spain are considered. Materials and methods Objectives Consensus levels The aim of this study was to establish a consensus through The expert panel completed two DR of the statements Delphi Technique on the treatment strategy options in through questionnaire in an online platform. Spain for aNSCLC harboring EGFRm. An external expert Each item was responded using Likert scale with 1–9 in Delphi methodology guaranteed the quality of the overall range (1 = “strongly disagree”; 9 = “strongly agree”) and process. the consensus level was calculated by tertiles. Range from 1 to 3 were considered values of disagreement, 4–6 range Study design neither disagree nor agree, and scores from 7 to 9 were considered values of agreement. A statement with consen- Scientific Committee sus existed when more than 2/3 of the answers are in the median range (MR), and not consensus if more than 33.3% In September 2019, a scientific committee of six experts in are outside the MR. If the MR is in 7–9, there is consensus comprehensive lung cancer management met to define and about agreement between panelists, if MR is between 4–6 lead this consensus project. The following steps were carried exist indeterminate consensus and when the MR is 1–3 out by the scientific committee: (a) selection of expert panel; there is consensus of panelists disagreement. (b) generation of clinical statements based on review of the The possible consensus levels were the following: medical literature; (c) definition of the consensus levels and agreement on methodology; (d) interpretation of the results • Consensus/agree: there is consensus between panelist, from the two Delphi rounds (DR); (e) develop of the final the statement should not be reviewed. consensus document. • Consensus/disagree: there is consensus that panelists disagree. Expert panel A panel of 31 medical oncologists special- • No consensus/agree: there are more panelists in agree- ized in lung cancer and outstanding for their clinical and ment, but there is no consensus. research experience were selected by the scientific com- 13 1306 Clinical and Translational Oncology (2021) 23:1304–1313 • No consensus/disagree: there are more answers in disa- Results greement, but there is no consensus. • Consensus/indeterminate: majority neither agree nor In the FDR (100% response rate), 33 statements were disagree. Statements should be reviewed and to decide evaluated, and a consensus was reached for 24 statements whether to delete or keep up once evaluated (no items (72.7%). One of the non-consensus statements was rejected achieved this score). after reassessed by the scientific committee, and the remain- • No consensus/indeterminate: there is no general agree- ing 8 statements were evaluated in the SDR (96.8% response ment or disagreement, and, furthermore, there is no con- rate), in which neither achieved consensus. sensus on this indeterminacy. All statements reaching consensus in the first round Reliability of the questionnaire (FDR) did not undergo the second Delphi round (SDR). A SDR was held for the 8 statements without consensus in Cronbach’s alpha coefficient and intraclass correlation coef- the FDR. ficients showed a high internal consistency for the complete In FDR, started on 30 January 2020, 31 panelists questionnaire as well as for each domain with values above answered the complete questionnaire, which supposes a 0.7 in general (except ri Domain 1 and 3) (Table 1). response rate of 100%. In SDR (23 March 2020) 30 of 31 completed the questionnaire (96.8% response rate). This Molecular diagnosis on aNSCLC study consisted of a survey of expert opinions and no patient data were collected, so no specific independent ethical or Table 2 shows the responses for the ten statements in domain research review or approval was necessary. 1 which is focused on the importance of molecular diagno- sis prior to treatment election. Eight items reached a high Statistical analysis panelist consensus (> 78%) about the agreement of these statements. Only one did not reach consensus neither in the The reliability of the questionnaire was assessed by internal FDR nor in the SDR. The other one statement was deleted consistency using Cronbach’s alpha coefficient (Cronbach’s from the SDR. α) and intraclass correlation (ri) coefficients. Both statistics Highlight the unanimity of agreement between experts were calculated for the complete questionnaire and for each (100% consensus) about the need to know the EGFR muta- domain. tion status, regardless of PD-L1/ PD-L1 expression, before Median and interquartile range (p25–p75) were calculated to establish the treatment (statement 4). for each item (none of the variables showed Gaussian dis- The one statement with no consensus asked about the tribution). As it mentioned above, the score responses were importance of disease stage versus molecular diagnosis divided in tertiles (ranges: 1–3, 4–6, 7–9). The agreement (statement 1). On the other hand, the statements with high level was calculated according to the tertile in which the MR consensus were focused on EGFR status determination of the responses was located, and the consensus level was (preferability from tissue samples) and analysis of the dif- evaluated by the percentage of answers in this median tertile. ferent biomarkers, which shows that both variables are key Spearman’s correlation coefficient, following Martinez and influence the treatment decision and strategy in patients et al. criteria [8], was calculated to assess the correlation with aNSCLC. between two rounds for the total Delphi questionnaire as well as between domains and for the eight statements evalu- ated in both rounds. Further, the quantitative concordance Histologic profile and patient clinical characteristics was calculated by Kappa index (by Ashby [9].) The SPSS 21.0. was used for the statistical analyses. Table 3 shows the responses for the 7 statements in domain 2 that was focused on the relevance of histologic profile and Development of the consensus manuscript patient clinical characteristics regarding the 1L therapy elec- tion. This was the domain with less consensus. The special- This project began in September 2019 and ended in July ists only coincided with the no relevance of smoking habit 2020. During this time and when this article was drafted, in establishing the 1L therapy and the importance of always newly published literature and congress presentations were perform a CNS image at diagnosis. reviewed to analyze the clinical implications of any new data in patients with aNSCLC and to provide support for the consensus statements. 13 Clinical and Translational Oncology (2021) 23:1304–1313 1307 Table 1  Evaluation of Delphi questionnaire consistency DR Cronbach’s α (p)a ri (p)a TOTAL (33 ítems) 1 0.837 (< 0.001) 0.750 (< 0.001) 2 0.834 (< 0.001) 0.746 (< 0.001) Domain 1: molecular diagnosis on aNSCLC (10 items) 1 0.761 (< 0.001) 0.679 (0.006) 2 0.751 (< 0.001) 0.668 (0.006) Domain 2: histologic profile and patient clinical characteristics (7 items) 1 0.815 (< 0.001) 0.787 (< 0.001) 2 0.829 (< 0.001) 0.808 (< 0.001) Domain 3: first-line treatment in aNSCLC with EGFRm positive (8 items) 1 0.712 (< 0.001) 0.681 (< 0.001) 2 0.712 (< 0.001) 0.681 (< 0.001) Domain 4: second and subsequent-lines’ treatment in aNSCLC with EGFRm posi- 1 0.715 (< 0.001) 0.702 (< 0.001) tive (8 items) 2 0.718 (< 0.001) 0.712 (< 0.001) aNSCLC Advanced non–small-cell lung cancer, Cronbach’s α Cronbach’s alpha coefficient, DR Delphi round, EGFRm epidermal growth factor receptor mutation, ri intraclass correlation coefficient a Acceptable values for research purposes are considered values above 0.7, between 0.7–0.9 high values of reliability and above 0.9 very high values Table 2  Domain 1: molecular diagnosis on aNSCLC a Statement DR MR Panelist in Panelist in bp1/p2 range, n(%) Consensus MR, n(%) 1. I always wait to have the staging of the disease before 1 4–6 7(22.6) 12(38.7)/12(38.7) No consensus/indeterminate requesting the molecular diagnosis 2 7–9 18(60.0) 5(16.7)/7(23.3) No consensus/disagree 2. Prior to establish a treatment in non-smoking patients, 1 7–9 27(87.1) 3(9.7)/1(3.2) Consensus/agree the EGFR mutation status must be studied in metastatic 2 N/A N/A N/A N/A or advanced NSCLC, regardless of histology 3. Prior to establish a treatment, the EGFR mutation status 1 7–9 30(96.8) 0/1(3.2) Consensus/agree must be studied in entire aNSCLC 2 N/A N/A N/A N/A 4. Prior to establish an aNSCLC treatment, I always wait 1 7–9 31(100) 0/0 Consensus/agree to definitive EGFR mutation status, regardless of the 2 N/A N/A N/A N/A percentage of PD-L1/PD-L1 expression 5. Whether there is enough tissue sample I always try to 1 7–9 30(96.8) 0/1(3.2) Consensus/agree use them for starting the EGFR mutation status analysis 2 N/A N/A N/A N/A 6. The EGFR mutation status analysis is essential. If there 1 7–9 28(90.3) 3(9.7)/0 Consensus/agree is not enough tissue sample, I consider taking a second 2 N/A N/A N/A N/A sample 7. If there is not enough tissue sample, a liquid biopsy is 1 7–9 30(96.8) 1(3.2)/0 Consensus/agree requested to determine the EGFR mutation status 2 N/A N/A N/A N/A 8. In case of result of EGFRm-negative in tissue sample, 1 1–3 17(54.8) 11(35.5)/3(9.7) No consensus/disagree I always perform a confirmation test in liquid sample 2 N/A N/A N/A N/A (plasma) 9. For establishing a treatment to the patient, I always wait 1 7–9 30(96.8) 1(3.2)/0 Consensus/agree the results of specific biomarkers for targeted therapy 2 N/A N/A N/A N/A (EGFR, ALK, ROS1) 10. It is appropriated to use of NGS platform for identifi- 1 7–9 30(96.8) 1(3.2)/0 Consensus/agree cation of genetic-based biomarkers 2 N/A N/A N/A N/A aNSCLC Advanced non–small-cell lung cancer, DR Delphi round, EGFR epidermal growth factor receptor mutation, MR median range, NA not applicable, NGS next-generation sequencing, NSCLC non–small-cell lung cancer a MR = 7–9 agree range, MR = 4–6 neither agree nor disagree rage, MR = 1–3 disagree range b p1/p2 ranges depend of MR: when MR = 1–3 range, p1 = 4–6 and p2 = 7–9; if MR = 4–6, p1 = 1–3 and p2 = 7–9; finally if MR = 7–9, p1 = 1–3 and p2 = 4–6 13 1308 Clinical and Translational Oncology (2021) 23:1304–1313 Table 3  Domain 2: histologic profile and patient clinical characteristics a Statement DR MR Panelist in Panelist ­inbp1/p2 range, n(%) Consensus MR, n(%) 1. The age of the p­ atientc is key to establish the 1L 1 1–3 19(61.3) 8(25.8)/4(12.9) No consensus (limit)/disagree therapy 2 1–3 16(53.3) 5(16.7)/9(30.0) No consensus/disagree 2. The smoking habit of the p­ atientc is key to establish the 1 1–3 24(77.4) 5(16.1)/2(6.5) Consensus/disagree 1L therapy 2 NA NA NA NA 3. ECOG PS of the p­ atientc is key to establish the 1L 1 4–6 9(29.0) 10(32.3)/12(38.7) No consensus/indeterminate therapy 2 4–6 8(26.7) 11(36.7)/11(36.7) No consensus/indeterminate 4. The comorbidities of the ­patientc is key to establish the 1 4–6 6(19.4) 10(32.3)/15(48.4) No consensus/indeterminate 1L therapy 2 4–6 9(30.0) 10(33.3)/11(36.7) No consensus/indeterminate 5. The common EGFR mutation subtypes (del19 and 1 4–6 7(22.6) 15(48.4)/9(29.0) No consensus/indeterminate L858R) are key to establish a 1L therapy 2 1–3 17(56.7) 5(16.7)/8(26.7) No consensus/disagree 6. The presence of brain metastases in the p­ atientc is key 1 7–9 18(58.1) 7(22.6)/6(19.4) No consensus/agree to establishing the 1L therapy 2 7–9 19(63.3) 4(13.3)/7(23.3) No consensus (limit)/agree 7. The ­patientc always undergo to CNS imaging (MRI/ 1 7–9 29(93.5) 2(6.5)/0 Consensus/agree CT) at diagnosis 2 NA NA NA NA CNS Central nervous system, CT computed tomography, DR Delphi round, EGFR epidermal growth factor receptor mutation, 1L first line, MR median range, MRI magnetic resonance imaging, NA not applicable a MR = 7–9 agree range, MR = 4–6 neither agree nor disagree rage, MR = 1–3 disagree range b p1/p2 ranges depend of MR: When MR = 1–3 range, p1 = 4–6 and p2 = 7–9; if MR = 4–6, p1 = 1–3 and p2 = 7–9; finally if MR = 7–9, p1 = 1–3 and p2 = 4–6 c patient with aNSCLC and EGFRm 1L therapy in patients with aNSCLC and EGFRm to assess properly the patient, and the inclusion in a clinical positive trial would be a priority. To perform a new biopsy was rel- evant in case of progression, regardless of the type of TKI Table 4 shows the responses for the eight statements in used on the front-line. Finally, in case of disease progression domain 3 which was focused on the therapeutic election as to osimertinib or in absence of T790M mutation, the best 1L therapy in EGFRm aNSCLC patients. Outcomes show option was to use chemotherapy. there is homogeneous therapeutic criteria, with consensus in all the evaluated items (74.2–96.8%). Correlation analysis Osimertinib was considered the best 1L therapy choice in patients with common activating mutations (del19, L858R) In order to assess the correlation between the two rounds due the safety profile, PFS and OS of this drug compared to as much the total questionnaire, as each domain (Table 6a) other EGFR-TKI, and for patients with uncommon EGFR and of the eight items which were included in both rounds mutations. (Table 6b), Spearman’s coefficient and Kappa index were calculated. The Kappa index values corroborate Spearmans`s Second and subsequent‑lines treatment in aNSCLC coefficient estimators in all cases. with EGFRm positive Table 6a shows a strong quantitative correlation between two DR Table 5 shows the responses for the eight statements in (Spearmans’s coefficient range 0.76–1 = strong or domain 4 which was focused on the second and subsequent- very strong correlation) both in total questionnaire and lines therapeutic options in patients with EGFRm aNSCLC. by domains. However, the coefficients are overestimated As in the front-line, the specialists showed homogeneous because the values of the items which reaches consensus at therapeutic criteria with consensus in 6/8 of the statements, the first round are repeated in the SDR. Kappa index indi- highlighting the 100% consensus in three of them. cates a good qualitative concordance in total questionnaire In patients treated with osimertinib in front-line, if oli- as well as in each domain. goprogression occured during the therapy, this TKI must The items between the two phases have moderate and be maintained adding local treatment. In case of progres- strong correlation values, except for item 1 from Domain 1 sion, evaluation of PD-L1 expression was needed in order with weak correlation (Table 6b). 13 Clinical and Translational Oncology (2021) 23:1304–1313 1309 Table 4  Domain 3: First-line treatment in EGFRm advanced NSCLC a Statement DR MR Panelist in Panelist in Consensus MR, n(%) bp1/p2 range, n(%) 1. The best therapeutic option is chosen as 1L therapy, regardless of the subse- 1 7–9 29(93.5) 1(3.3)/1(3.2) Consensus/agree quent lines 2 NA NA NA NA 2. The first-generation of EGFR-TKI (gefitinib, erlotinib) are the most effec- 1 1–3 25(80.6) 6(19.4)/0 Consensus/disagree tive and safe 1L therapy in patients with common activating mutations (del19, 2 NA NA NA NA L858R) 3. The second-generation EGFR-TKI (afatinib) is the most effective and safe 1L 1 1–3 23(74.2) 8(25.8)/0 Consensus/disagree therapy in patients with common activating mutations (del19, L858R) 2 NA NA NA NA 4. The 3rd-generation EGFR-TKI (osimertinib) is the most effective and safe 1L 1 7–9 30(96.8) 0/1(3.2) Consensus/agree therapy in patients with common activating mutations (del19, L858R) 2 NA NA NA NA 5. In case of the patient has uncommon EGFR mutations, the treatment choice is 1 7–9 24(77.4) 5(16.1)/2(6.5) Consensus/agree different 2 NA NA NA NA 6. The benefit of osimertinib in progression-free survival compared to other 1 7–9 29(93.5) 2(6.5)/0 Consensus/agree EGFR-TKI is relevant and justifies its use as 1L therapy 2 NA NA NA NA 7. The benefit that osimertinib provides in overall survival compared to other 1 7–9 29(93.5) 2(6.5)/0 Consensus/agree EGFR-TKI is relevant and justifies its use as 1L therapy 2 NA NA NA NA 8. The safety profile of osimertinib is relevant to support its use as 1L therapy 1 7–9 29(93.5) 1(3.2)/1(3.2) Consensus/agree 2 NA NA NA NA DR Delphi round, EGFR-TKI epidermal growth factor receptor tyrosine kinase inhibitor, 1L first line, MR median range, NA not applicable a MR = 7–9 agree range, MR = 4–6 neither agree nor disagree rage, MR = 1–3 disagree range b p1/p2 ranges depend of MR: When MR = 1–3 range, p1 = 4–6 and p2 = 7–9; if MR = 4–6, p1 = 1–3 and p2 = 7–9; finally if MR = 7–9, p1 = 1–3 and p2 = 4–6 Table 5  Domain 4: Second and subsequent-lines treatment in EGFRm advanced NSCLC a Statement DR MR Panelist Panelist in bp1/p2 Consensus in MR, n range, n (%) (%) 1. In case of NSCLC oligoprogression during 1L therapy with osimer- 1 7–9 31(100) 0/0 Consensus/agree tinib, osimertinib is maintained plus local treatment 2 NA NA NA NA 2. In case of progression to osimertinib treatment, PD-L1 expression 1 4–6 9(29.0) 12(38.7)/10(32.3) No consensus/indeterminate is relevant for taking future decision 2 1–3 15(50.0) 7(23.3)/8(25.8) No consensus/disagree 3. In case of progression to osimertinib treatment, the priority is to 1 7–9 30(96.8) 0/1(3.2) Consensus/agree include the patient in a clinical trial 2 NA NA NA NA 4. In case of progression to osimertinib treatment, to perform a new 1 7–9 28(90.3) 3(9.7)/0 Consensus/agree biopsy is relevant 2 NA NA NA NA 5. In case of progression to treatment with a first-generation TKi (gefi- 1 7–9 31(100) 0/0 Consensus/agree tinib, erlotinib), to perform a new biopsy is relevant 2 NA NA NA NA 6. After progression to treatment with a second generation TKi 1 7–9 31(100) 0/0 Consensus/agree (afatinib), to perform a new biopsy is relevant 2 NA NA NA NA 7. In case of progression to osimertinib or a TKi treatment in absence 1 7–9 21(67.7) 7(22.6)/3(9.7) Consensus (limit)/agree of T790M, the best option is to use chemotherapy 2 NA NA NA NA 8. In case of progression to osimertinib or a treatment in absence of 1 7–9 17(54.8) 11(35.5)/3(9.7) No consensus/agree T790M mutation, the best option is to use chemotherapy + antiangio- 2 4–6 13(43.3) 3(10.0)/14(46.7) No consensus/indeterminate genic therapy + anti-PD-L1 antibody DR Delphi round, 1L first line, MR median range, NA not applicable, NSCLC non–small-cell lung cancer, PD-L1 programmed death-ligand 1, TKI tyrosine kinase inhibitor a MR = 7–9 agree range, MR = 4–6 neither agree nor disagree rage, MR = 1–3 disagree range b p1/p2 ranges depend of MR: When MR = 1–3 range, p1 = 4–6 and p2 = 7–9; if MR = 4–6, p1 = 1–3 and p2 = 7–9; finally if MR = 7–9, p1 = 1–3 and p2 = 4–6 13 1310 Clinical and Translational Oncology (2021) 23:1304–1313 Table 6  Delphi Rounds correlation analysis A. Correlation of the total questionnaire and of the domains between two Delphi Rounds Spearman’s c (p)a Kappa index (p)b TOTAL (33 ítems) 0.909 (< 0.001) 0.880 (< 0.001) Domain 1. Molecular diagnosis on aNSCLC (10 items) 0.954 (< 0.001) 0.939 (< 0.001) Domain 2. Histologic profile and patient clinical characteristics (7 items) 0.856 (< 0.001) 0.829 (< 0.001) Domain 3. First-line treatment in aNSCLC with EGFRm positive (8 items) 1 (< 0.001) 1 (< 0.001) Domain 4. Second and subsequent-lines treatment in aNSCLC with EGFRm positive (8 0.849 (< 0.001) 0.867 (< 0.001) items) B. Correlation of the statements between the two Delphi Rounds Statement Spearman’s c (p)a Kappa index (p)b Domain 1. Molecular diagnosis on aNSCLC Item 1 0.318 (0.125) 0.359 (0.137) Domain 2. Histologic profile and patient clinical characteristics Item 1 0.892 (< 0.001) 0.860 (< 0.001) Item 3 0.902 (< 0.001) 0.899 (< 0.001) Item 4 0.738 (0.006) 0.708 (0.008) Item 5 0.561 (0.162) 0.569 (0.156) Item 6 0.856 (< 0.001) 0.842 (< 0.001) Domain 3. First-line treatment in aNSCLC with EGFRm positive Domain 4. Second and subsequent-lines treatment in aNSCLC with EGFRm positive Item 2 0.662 (0.016) 0.641 (0.022) Item 6 0.692 (0.019) 0.715 (0.011) aNSCLC Advanced non–small-cell lung cancer, EGFRm epidermal growth factor receptor mutation, Spearman’s c Spearman’s correlation coef- ficient a Cuantitative correlation score: [0—0.25] poor correlation; [0.26–0.50] weak correlation; [0.51–0.75- moderate/strong; [0.76–1] strong/very strong correlation b Cualitative correlation score: < 0.00 no agreement; 0.00–0.20- poor concordance; 0.21–0.40- weak concordance; 0.41–0.60- moderate concord- ance; 0.61–0.80- good concordance; 0.81–1- very good concordance Finally, the coefficient of variation was much less than a specific therapeutic decision process to optimize manage- 10%. Such a low value, which did not even reach 1%, sup- ment of EGFRm aNSCLC patients. ports the decision not to carry out a new round, since it Regarding molecular diagnosis, it is important to high- would not provide variation. light the unanimity of agreement among experts about the need to know the definitive EGFRm status, regardless of PD-L1 expression, or patient characteristics like age or smoking habit, before starting aNSCLC treatment. This Discussion recommendation follows the most relevant and current guidelines [10–12]. Although tumor tissue is the most rel- The current Delphi study shows that a high degree of con- evant source of information, unfortunately in up to 25% of sensus exists among experts on the relevance to obtain cases [13], the initial tissue biopsy is inadequate for preci- molecular diagnosis before starting any treatment in sion oncology. In this context, the use of liquid biopsy is aNSCLC, how these patients should be treated in the 1L an efficient diagnostic alternative which can be considered setting according to the presence of EGFRm, and how the at the time of initial diagnosis in patients who need tumor 1L therapy decision in these population may influence sub- molecular profiling, specially when tumor tissue is scarce, sequent therapeutic approaches. This consensus document unavailable, or a delay greater than 2 weeks in obtaining complements the information in the European Society of tumor tissue is expected [14, 15]. The one statement with Medical Oncology (ESMO) [10], Sociedad Española de no consensus was related to the importance of disease stage Oncología Médica (SEOM) and Sociedad Española de versus molecular diagnosis. After the ADAURA trial, that Anatomía Patológica (SEAP) [11], American College of has shown significant results in terms of disease free sur- Pathologists (ACP) [12] and aims to provide physicians with vival with osimertinib in the adjuvant setting in EGFRm 13 Clinical and Translational Oncology (2021) 23:1304–1313 1311 stage IB-IIIA population [16], probably changing the current Finally, in case of disease progression to osimertinib or after treatment paradigm in early NSCLC EGFRm, all patients first or second generation TKI in absence of T790M muta- will need to be tested for EGFRm regardless of the stage tion, the best option was to use chemotherapy. of disease. Although this consensus document aims to help therapeu- Moving forward the histologic profile and patient clini- tic decision-making process, it has several limitations. First, cal characteristics, the experts agreed with the no relevance this Delphi project has been developed under the premise of smoking habit and age in establishing the 1L therapy of patients with EGFRm aNSCLC with a good general con- and the importance of CNS imaging at diagnosis, given dition (ECOG PS 0 and 1) and without medical contrain- the high incidence of brain metastasis in EGFRm aNSCLC dications. This may limit its potential applicability to all [17]. Regarding the type of EGFRm, the two most com- patients with EGFRm aNSCLC. Also, as the treatment of mon EGFRm subtypes are a deletion in exon 19 (Del19) aNSCLC further evolves, it is important to underline that the or point mutations in exon 21 (L858R) that constitute SDR took place before the communication of some relevant approximately 90% of activating mutations. Although scientific information such as the results of osimertininb in subgroup analysis from several clinical trials have shown adjuvant setting in stage IB-IIIA NSCLC [16]. that a Del19 mutation is predictive of better efficacy than Despite these limitations, the study shows the consistency L858R mutation,the EGFRm subtype is not a factor to be of the questionnaire created to meet the proposed objectives considered in the EGFR-TKI selection process in this con- statements selected: Cronbach’s alpha coefficient and intra- sensus [3]. On the contrary, patients with aNSCLC with class correlation coefficients showed a high internal con- uncommon EGFRm may experience variable responses to sistency for the complete questionnaire as well as for each the currently available EGFR-TKIs [18, 19]. In patients with domain with values. The correlation analysis showed strong uncommon EGFRm, there was a consensus that the treat- quantitative correlation and a good qualitative concordance ment of choice had to be different. between two DR in total questionnaire as well as in each Regarding the 1L therapy option in EGFRm, osimer- domain. The Kappa index values corroborate Spearmans’s tinib was considered the best front-line choice in patients coefficient estimators in all cases. with common activating mutations (del19, L858R). The This consensus document has been generated due to FLAURA trial results support this recommendation: the need to answer key clinical questions such as the best osimertinib demonstrated improvement in median PFS sequencing approach for EGFRm aNSCLC. We would (18.9 months vs. 10.2 months; hazard ratio (HR) 0.46; 95% encourage to use this document to help stimulate discussion CI, 0.37 to 0.57; p < 0.001) and a more favorable toxicity on future real-world studies that could be carried out to sup- profile due to its lower affinity for wild-type EGFR. Further- port or question the consensus statements. more, osimertinib has also improved efficacy against brain metastases [4]. Very recently published OS results have Acknowledgements The authors would like to thank the panel experts for their participation in this Delphi study: Edurne Arriola (Hospital del shown a median of 38.6 months in the osimertinib group Mar, Barcelona); Ángel Artal (Hospital Universitario Miguel Servet, and 31.8 months in the comparator group (HR, 0.80; 95.05% Zaragoza); Reyes Bernabé (Hospital Virgen del Rocío, Sevilla); Carlos CI, 0.64 to 1.00; P = 0.046). At 3 years, 79 of 279 patients Camps (Hospital General Universitario de Valencia); Enric Carcereny (28%) in the osimertinib group and 26 of 277 (9%) in the (Hospital Germans Trias i Pujol, Badalona); Manuel Cobo (Hospital Universitario Carlos Haya, Málaga); Pilar Diz (Complejo Asistencial comparator group were continuing to receive a trial regimen Universitario de León); Manuel Dominé (Hospital Universitario Fun- [5]. Grade ≥ 3 adverse events were reported in 42% of the dación Jiménez Díaz, Madrid); Enriqueta Felip (Hospital Universitario patients in the osimertinib group and in 47% of those in the Vall d’Hebron, Barcelona); Pilar Garrido (Hospital Ramón y Cajal, comparator group. These data favor osimertinib for tolerabil- Madrid); Ignacio Gil Bazo (Clínica Universidad de Navarra, Pam- plona); Amelia Insa (Hospital Clínico Universitario de Valencia); Mar- ity and safety, except for the slightly higher rate of interstitial tín Lázaro (Complejo Hospitalario Universitario de Vigo); Pilar Lianes lung disease, but which was nonetheless manageable [20]. (Hospital de Mataró, Barcelona); Bartomeu Massuti (Hospital General Focusing on the second and subsequent treatment lines de Alicante), Ernest Nadal (Hospital Duran i Reynals, Hospitalet); in EGFRm aNSCLC there was a consensus to maintain TKI Luis Paz Ares (Hospital Universitario 12 de Octubre, Madrid); San- tiago Ponce (Hospital Universitario 12 de Octubre, Madrid); Mariano plus local treatment in case of NSCLC oligoprogression dur- Provencio (Hospital Puerta de Hierro, Madrid), Noemí Reguart (Hos- ing 1L therapy. This statement has been validated in small pital Clinic de Barcelona), Josefa Terrasa (Hospital de Son Espases, trials showing the benefit of this strategy specially with the Palma de Mallorca), Nuria Viñolas (Hospital Clínic de Barcelona); use of sterotactic body radiation therapy (SBRT) [21, 22]. In Jordi Remón (Hospital HM Delfos, Barcelona); Delvys Rodriguez (Hospital Universitario Insular de Gran Canaria); José Manuel Trigo case of progression to osimertinib, the inclusion in a clinical (Hospital Virgen de la Victoria, Málaga). trial and to perform a new biopsy was considered to be rel- evant. To perfom a new biopsy was also considered to be rel- Funding The project was funded by ICAPEM (Asociación para evant in case of progression to first or second generation TKI la Investigación de Cáncer de Pulmón en Mujeres). The opinions in order to determine the acquired resistance mechanism. expressed are those of the authors. The funding party did not influence 13 1312 Clinical and Translational Oncology (2021) 23:1304–1313 any aspect of the study design, collection, analysis or interpretation of 3. Lee CK, Davies L, Wu YL, Mitsudomi T, Inoue A, Rosell R, data, or decision to submit the manuscript for publication. et al. Gefitinib or Erlotinib vs Chemotherapy for EGFR Mutation- Positive Lung Cancer: individual patient data meta-analysis of overall survival. J Natl Cancer Inst. 2017. https:​ //doi.org/10.1093/ Compliance with ethical standards jnci/djw27​9. 4. Soria JC, Ohe Y, Vansteenkiste J, Reungwetwattana T, Chewasku- Conflict of interest The authors would like to disclose the following: lyong B, Lee KH, et al. Osimertinib in untreated EGFR- Dolores Isla: Personal financial interests: Consulation Honoraria: Ab- mutated advanced non-small-cell lung cancer. N Engl J Med. bVie, Astra-Zeneca, BMS, Boehringer Ingelheim, F. Hoffmann-La 2018;378(2):113–25. https​://doi.org/10.1056/NEJMo​a1713​137. Roche, MSD, Pierre Fabre, Pfizer, Takeda. Speaker Honoraria: Astra- 5. Ramalingam SS, Vansteenkiste J, Planchard D, Cho BC, Zeneca, BMS, Boehringer Ingelheim, F. Hoffmann-La Roche, MSD, Gray JE, Ohe Y, et al. Overall survival with osimertinib in Pierre Fabre,Pfizer. Institutional financial interests: Clinical Trials: untreated, EGFR-mutated advanced NSCLC. N Engl J Med. AbbVie, Astra-Zeneca, Boehringer Ingelheim, BMS, F. Hoffmann- 2020;382(1):41–50. https​://doi.org/10.1056/NEJMo​a1913​662. La Roche, Janssen, MSD, Novartis, Research Grant: BMS. Javier 6. Wu YL, Cheng Y, Zhou X, Lee KH, Nakagawa K, Niho S, et al. de Castro: Educational grants: Astra-Zeneca, Boehringer Ingelheim, Dacomitinib versus gefitinib as first-line treatment for patients BMS, MSD, Novartis, Roche. Scientific consulting: Astra-Zeneca, with EGFR-mutation-positive non-small-cell lung cancer Boehringer Ingelheim, BMS, MSD, Novartis, Pfizer, Roche, Takeda. (ARCHER 1050): a randomised, open-label, phase 3 trial. Lan- Rosario García-Campelo: Consulting, advisory role or speaker’s bu- cet Oncol. 2017;18(11):1454–66. https​://doi.org/10.1016/s1470​ reau: Boehringer Ingelheim, BMS, Astra-Zeneca, Roche, MSD, Sa- -2045(17)30608​-3. nofi, Takeda, Pfizer. Margarita Majem: Grants: BMS. Personal fees: 7. McMillan SS, King M, Tully MP. How to use the nominal group BMS, MSD, Boehringer Ingelheim, Astra-Zeneca, Roche, Kyowa and Delphi techniques. Int J Clin Pharm. 2016;38(3):655–62. https​ Kyrin, Pierre Fabre. Non-financial support: MSD, Boehringer Ingel- ://doi.org/10.1007/s1109​6-016-0257-x. heim, Astra-Zeneca, Roche. Others: Astra-Zeneca, Roche. David Vi- 8. Martínez Ortega RM, Tuya Pendás LC, Martínez Ortega M, Pérez cente: Honoraria: Astra-Zeneca, Merck, BMS, Roche, MSD, Pfizer. Abreu A, Cánovas AM. El coeficiente de correlación de los rangos Consulting or advisory role: Astra-Zeneca, Merck, BMS, Roche, de Spearman caracterización. Rev Haban Cienc Méd. 2009;8:0. MSD, Pfizer. Expert testimony: MSD. Travel, accomodations, ex- 9. Ashby D. Practical statistics for medical research. Douglas G. penses: Astra-Zeneca. Oscar Juan-Vidal: Honoraria or Advisory role: Altman, Chapman and Hall, London, 1991. No. of pages: 611. Boehringer Ingelheim, Bristol-Myers Squibb, Merck Sharp & Dohme, Price: £32.00. Statistics in Medicine. 1991;10(10):1635–6. doi: Roche/Genetech, Astra-Zeneca, Pfizer, Eli Lilly, Abbvie, Takeda. https​://doi.org/10.1002/sim.47801​01015​. Travel, accomodations, expenses: Boehringer Ingelheim, Merck Sharp 10. Planchard D, Popat S, Kerr K, Novello S, Smit EF, Faivre-Finn & Dohme, Roche/Genetech. C, et al. Metastatic non-small cell lung cancer: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Ethical approval This study consisted of a survey of expert opinions Oncol. 2018;29(Suppl 4):iv192–237. https​://doi.org/10.1093/ and no patient data were collected, so no specific independent ethical annon​c/mdy27​5. or research review or approval was necessary. 11. Garrido P, Conde E, de Castro J, Gómez-Román JJ, Felip E, Pijuan L, et al. Updated guidelines for predictive biomarker testing in Informed consent For this type of study formal consent is not required. advanced non-small-cell lung cancer: a National Consensus of the Spanish Society of Pathology and the Spanish Society of Medical Oncology. Clin Transl Oncol. 2020;22(7):989–1003. https​://doi. Open Access This article is licensed under a Creative Commons Attri- org/10.1007/s1209​4-019-02218​-4. bution 4.0 International License, which permits use, sharing, adapta- 12. Kalemkerian GP, Narula N, Kennedy EB, Biermann WA, Don- tion, distribution and reproduction in any medium or format, as long ington J, Leighl NB, et al. Molecular testing guideline for the as you give appropriate credit to the original author(s) and the source, selection of patients with lung cancer for treatment with tar- provide a link to the Creative Commons licence, and indicate if changes geted tyrosine kinase inhibitors: American Society of Clinical were made. The images or other third party material in this article are Oncology Endorsement of the College of American Patholo- included in the article’s Creative Commons licence, unless indicated gists/International Association for the Study of Lung Cancer/ otherwise in a credit line to the material. If material is not included in Association for Molecular Pathology Clinical Practice Guideline the article’s Creative Commons licence and your intended use is not Update. J Clin Oncol. 2018;36(9):911–9. https​://doi.org/10.1200/ permitted by statutory regulation or exceeds the permitted use, you will jco.2017.76.7293. need to obtain permission directly from the copyright holder. To view a 13. Zehir A, Benayed R, Shah RH, Syed A, Middha S, Kim HR, copy of this licence, visit http://creat​iveco​mmons​.org/licen​ses/by/4.0/. et al. Mutational landscape of metastatic cancer revealed from prospective clinical sequencing of 10,000 patients. Nat Med. 2017;23(6):703–13. https​://doi.org/10.1038/nm.4333. 14. Rolfo C, Mack PC, Scagliotti GV, Baas P, Barlesi F, Bivona TG, References et al. Liquid biopsy for advanced non-small cell lung cancer (NSCLC): a statement paper from the IASLC. J Thorac Oncol. 1. Herbst RS, Morgensztern D, Boshoff C. The biology and manage- 2018;13(9):1248–68. https​://doi.org/10.1016/j.jtho.2018.05.030. ment of non-small cell lung cancer. Nature. 2018;553(7689):446– 15. Remon J, García-Campelo R, de Álava E, Vera R, Rodríguez- 54. https​://doi.org/10.1038/natur​e2518​3. Peralto JL, Rodríguez-Lescure Á, et al. Liquid biopsy in oncol- 2. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal ogy: a consensus statement of the Spanish Society of Pathology A. Global cancer statistics 2018: GLOBOCAN estimates of inci- and the Spanish Society of Medical Oncology. Clin Transl Oncol. dence and mortality worldwide for 36 cancers in 185 countries. 2020;22(6):823–34. https:​ //doi.org/10.1007/s12094​ -019-02211-​ x. CA Cancer J Clin. 2018;68(6):394–424. https​://doi.org/10.3322/ 16. Herbst RS, Tsuboi M, John T, Grohé C, Majem M, Goldman JW, caac.21492​. et al. Osimertinib as adjuvant therapy in patients (pts) with stage IB–IIIA EGFR mutation positive (EGFRm) NSCLC after com- plete tumor resection: ADAURA. Journal of Clinical Oncology. 13 Clinical and Translational Oncology (2021) 23:1304–1313 1313 2020;38(18_suppl):LBA5-LBA. doi: https​ ://doi.org/10.1200/ Saf. 2018;17(12):1239–48. https ​ : //doi.org/10.1080/14740​ JCO.2020.38.18_suppl​.LBA5. 338.2018.15492​22. 17. Yang B, Lee H, Um SW, Kim K, Zo JI, Shim YM, et al. Inci- 21. Al-Halabi H, Sayegh K, Digamurthy SR, Niemierko A, Piotrowska dence of brain metastasis in lung adenocarcinoma at initial diag- Z, Willers H, et al. Pattern of failure analysis in metastatic EGFR- nosis on the basis of stage and genetic alterations. Lung Cancer. mutant lung cancer treated with tyrosine kinase inhibitors to 2019;129:28–34. https​://doi.org/10.1016/j.lungc​an.2018.12.027. identify candidates for consolidation stereotactic body radia- 18. Yang JC, Sequist LV, Geater SL, Tsai CM, Mok TS, Schuler M, tion therapy. J Thorac Oncol. 2015;10(11):1601–7. https​://doi. et al. Clinical activity of afatinib in patients with advanced non- org/10.1097/jto.00000​00000​00064​8. small-cell lung cancer harbouring uncommon EGFR mutations: 22. Guo T, Ni J, Yang X, Li Y, Li Y, Zou L, et al. Pattern of recurrence a combined post-hoc analysis of LUX-Lung 2, LUX-Lung 3, analysis in metastatic EGFR-mutant NSCLC treated with osimer- and LUX-Lung 6. Lancet Oncol. 2015;16(7):830–8. https​://doi. tinib: implications for consolidative stereotactic body radiation org/10.1016/s1470​-2045(15)00026​-1. therapy. Int J Radiat Oncol Biol Phys. 2020;107(1):62–71. https​ 19. Cho JH, Lim SH, An HJ, Kim KH, Park KU, Kang EJ, et al. Osi- ://doi.org/10.1016/j.ijrob​p.2019.12.042. mertinib for patients with non-small-cell lung cancer harboring uncommon EGFR mutations: a multicenter, open-label, phase II Publisher’s Note Springer Nature remains neutral with regard to trial (KCSG-LU15-09). J Clin Oncol. 2020;38(5):488–95. https​ jurisdictional claims in published maps and institutional affiliations. ://doi.org/10.1200/jco.19.00931​. 20. Mezquita L, Varga A, Planchard D. Safety of osimertinib in EGFR-mutated non-small cell lung cancer. Expert Opin Drug 13