Management and outcomes of non–small cell lung cancer patients with rapid progression under second‐or‐more‐line immune checkpoint inhibitors: ERORECI study (GFPC 2016‐04)

Abstract Background Immune checkpoint inhibitors (ICIs) have been approved as second‐line therapy for advanced non–small cell lung cancers (NSCLCs) progressing after platinum‐based chemotherapy. However, some patients' disease progressed rapidly and sometimes exhibited explosive tumor progression. This descriptive, prospective study aimed to assess the characteristics of nonresponders with rapid progression (RP), defined as progression‐free survival (PFS) ≤2 or 2‐4 months under ICIs. Methods This analysis included all consecutive ICI‐treated (second‐or‐more line) patients with RP ≤4 months from 1 September 2016 to 31 August 2017 and compared the clinical characteristics, treatments, and outcomes (overall survival [OS]; responses; PFS, according to treatment line) of NSCLCs that progressed after ≤2 vs 2‐4 months on ICIs. Results Comparisons of the 224 (70.2%) patients with ≤2‐month and 95 (29.8%) with 2‐ to 4‐month RP revealed the former had less frequent nonsmokers and ECOG PS = 0, more frequent stage IV disease and higher neutrophil/lymphocyte ratio. Their respective ICI PFS rates were: 1.6 [95% CI: 0.1‐2] and 2.7 [2.0‐4.2] months, with 16.5% and 11.6% having partial responses to first‐ and second‐line therapies post‐ICI chemotherapy. Their respective median OS rates were 6.0 and 9.0 months (P ≤ .009). Multivariate analysis retained only PFS of the first‐line therapy pre‐ICI and neutrophil/lymphocyte ratio at ICI onset as being significantly associated with ≤2‐month RP. Conclusion In the real‐life setting, NSCLC RP on ICI remains a challenge. New descriptive and analytic studies are needed to identify factors predictive of RP.


| INTRODUCTION
Immune checkpoint inhibitors (ICIs) represent a major advancement in the management of metastatic non-small cell lung cancers (NSCLCs). [1][2][3][4][5] As monotherapy, after progression on platinum-based chemotherapy, they significantly prolong survival with median overall survival (OS) of 9.9 [7.8-12.4] months. 1,2,5,6 Nevertheless, rapid progression (RP) characterizes 25% of these tumors, notably during the first 4 months on ICIs or even earlier. Some of those RPs can involve numerous sites, giving the impression of an explosion of metastatic disease. 7,8 The definitions of RP or hyperprogression are still being debated. 7,8 That phenomenon has also been observed for other solid tumors. 9 To define hyperprogression, the tumor growth rate between two computed-tomography (CT) scans with computer determinations has to be calculated, 7 which is difficult to do in clinical practice. The RP definition relies on the duration of progression-free survival (PFS) but no consensus has yet been reached. 8 According to some studies, 3-6 the therapeutic response was influenced by the programmed cell death-1 ligand (PD-L1)expression rate. However, some NSCLCs undergo early progression, regardless of their PD-L1 expression rate. 5 Numerous factors impact the risk of progression: nutritional status 10 , genetic alterations [11][12][13][14][15] and biological anomalies, 16 like the neutrophil/lymphocyte ratio, [17][18][19][20] which has been identified in some studies as also being a predictor of OS or PFS. [21][22][23][24] The RP definition of varies according to the authors, but the majority of studies retained 2-or 4-months following ICI onset. 25 Little is known about these patients' characteristics and those of their tumors, therapeutic options post-ICI and their outcomes. 26 This study was undertaken to describe the characteristics and management of patients given immunotherapy as second-or-more line, whose NSCLCs progressed within the 2 months after starting it, and their outcomes, and compared them to patients whose disease progressed 2-4 months post-ICI onset.

| MATERIALS AND METHODS
This multicenter, observational, prospective study included patients >18 years old, managed for stage-IIIb or-IV NSCLC treated with second-line-or-more immunotherapy, whose disease progressed within the 4 months after starting ICI. Almost all patients had progressive disease. Patients not covered by the French National Healthcare System or prisoners could not be included.
The primary outcome criterion was OS after starting ICI, with comparison between patients whose NSCLCs progressed within 2 months (first evaluation) and those whose progression occurred at 2-4 months (second evaluation). The choice of the first threshold was based on the literature 25 and the second threshold reflects routine clinical practice, because sometimes ICI is prolonged until the second evaluation, according to the clinical benefit. Secondary criteria were clinical characteristics, PFS on the different treatment lines before ICI and efficacies of post-ICI therapies.
The information was collected from medical records: clinical and histological findings, biological characteristics  16.5% and 11.6% having partial responses to first-and second-line therapies post-ICI chemotherapy. Their respective median OS rates were 6.0 and 9.0 months (P ≤ .009).
Multivariate analysis retained only PFS of the first-line therapy pre-ICI and neutrophil/lymphocyte ratio at ICI onset as being significantly associated with ≤2-month RP. Conclusion: In the real-life setting, NSCLC RP on ICI remains a challenge. New descriptive and analytic studies are needed to identify factors predictive of RP.

K E Y W O R D S
immune checkpoint inhibitors, non-small cell lung cancer, rapid progression, second-line treatment (neutrophil and lymphocyte counts) at NSCLC diagnosis and then at ICI onset, treatments before and after immunotherapy. Each PFS was calculated from initiation of the treatment line until the disease progressed, according to RECIST v1.1 criteria, which were assessed by each local investigating team, including a radiologist. No central review was done because this was a real-world study. Statistical analyses were computed with StatView v5.0 (SAS Institute Inc). They included descriptive analyses: frequencies of qualitative variables, and means ± standard deviation (SD), medians [95% confidence intervals (CIs)] for quantitative variables. Between-group comparisons (≤2-vs 2-to 4-month RP) used chi 2 or Fisher's exact tests. Univariate analyses of dichotomized variables used ANOVA. Survival analysis was estimated with Kaplan-Meier curves. Factors predictive of survival and RP ≤2 vs 2-4 months were identified with a Cox backward step-by-step logistic-regression model comprised of variables achieving P < .25 in univariate ANOVA.
In accordance with French law, the study was approved by Limoges University Ethics Committee on 23 March 2017.

| RESULTS
Between July 2016 and July 2017, 20 GFPC (French Lung Cancer Group) centers prospectively included 319 patients: median age: 64.3 years; 70.8% men; 91.6% smokers or exsmokers; 92.9% with Eastern Cooperative Oncology Group performance status (ECOG PS) PS = 0/1, 7.1% PS = 2; predominantly (63.6%) adenocarcinomas and 50.2% with >1 metastatic site at ICI onset (see Table 1 for other characteristics). Because PD-L1 testing was not generalized in France until September 2017, its status had been obtained for only 9% of the patients. All the patients had received nivolumab, the only ICI available at that time, initially as compassionate therapy, then after its marketing authorization had been obtained. Data about oncogenic drivers were not available for all patients, depending on the study period in France. However, 82 (27.5%) patients had Kirsten rat-sarcoma viral oncogene mutations, with the other biological markers being scarce: nine epidermal growth factor receptor mutations, four the v-RAF murine sarcoma viral oncogene homolog-B mutations, four human epidermal growth-factor-receptor-2 overexpressions, and four tyrosine kinase c-Met protooncogene mutations.
The cohort's PFS was 1.8 [95% CI: 0.2-4.2] months. This short duration is explained by cohort constitution (all patients had progressive disease). Among the 319, 167 progressed within 2 months after starting ICI; their PFS lasted 1.6 [95% CI: 0.2-2] months vs 2.7 [95% CI: 2.2-4.2] months for those with RP 2-4 months. The more rapid progressors had significantly worse ECOG PS, more advanced stage NSCLCs, and higher neutrophil/lymphocyte ratios at ICI onset and lower rates at the end of ICIs. Their PFS was significantly shorter on first-and second-line therapies before ICI than for those with RP 2-4 months. The cohort received a median of 4 [95% CI: 1-10] immunotherapy cycles: 3.99 and 5 for RP ≤ 2 and 2-4 months, respectively.
Progression-free survival rates on first-, second-, and third-line therapies, as a function of RP on ICI, are reported in Table 2. Those results clearly showed that PFS for the two first lines pre-ICI was longer for 2-to 4-month RP group than those progressing within ≤2 months. PFS3 is difficult  Figure 1B). Multivariate analysis retained only pre-ICI PFS after firstline therapy and the neutrophil/lymphocyte ratio at ICI onset as being significantly associated with RP ≤ 2 months (Table 4).

| DISCUSSION
This multicenter, observational, prospective study was designed to determine the clinical, biological, and evolutionary characteristics of patients with NSCLC progression and compared as a function of their RP ≤ 2 or 2-4 months. Its results showed that a shorter PFS after first-line therapy and marked inflammation at ICI onset were significantly associated with the earlier RP. Those factors were accompanied by shorter survival. However, it was possible to treat these patients after progression on ICI, with some of them achieving satisfactory survival. Hyperprogression among immunotherapy-treated patients is a real challenge for oncologists. And its precise mechanisms are still being debated: primary resistance, PD-1 expression on T-regulatory cells, compensatory T-cell exhaustion, modulation of tumor-promoting cells, aberrant inflammation, and activation of an oncologic pathway. 7 This investigation concerned patients with metastatic NSCLCs that progressed rapidly, defined as a very short interval between ICI onset and diagnosis of progressive disease. 25 We did not focus on patients with recently defined hyperprogression, 8