Hyperprogression during anti-pd-1/pd-l1 therapy in patients with recurrent and/or metastatic head and neck squamous cell carcinoma

Transcription

1 Annals of Oncology 28: , 17 doi:1.193/annonc/mdx178 Published online 13 April 17 ORIGINAL ARTICLE Hyperprogression during anti-pd-1/pd-l1 therapy in patients with recurrent and/or metastatic head and neck squamous cell carcinoma E. Sa^ada-Bouzid 1, C. Defaucheux 2, A. Karabajakian 3, V. P. Coloma 4, V. Servois 5, X. Paoletti 6, C. Even 4, J. Fayette 3, J. Guigay 1, D. Loirat 2, F. Peyrade 1, M. Alt 2, J. Gal 7 & C. Le Tourneau 2,8 * 1 Department of Medical Oncology, Centre Antoine Lacassagne, Nice; 2 Department of Medical Oncology, Institut Curie, Paris & Saint-Cloud; 3 Department of Medical Oncology, Centre Léon Bérard, Lyon; 4 Department of Oncology, Gustave Roussy, Villejuif; 5 Department of Radiology, Institut Curie, Paris; 6 Department of Biostatistics, Institut Gustave Roussy, Villejuif; 7 Department of Biostatistics, Centre Antoine Lacassagne, Nice; 8 INSERM U9 Research Unit, Saint-Cloud, France *Correspondence to: Prof. Christophe Le Tourneau, Department of Medical Oncology, Institut Curie, 35, rue Dailly, 9221 Saint-Cloud, France. Tel: þ ; christophe.letourneau@curie.fr Background: Pembrolizumab and nivolumab are immune checkpoint inhibitors targeting PD-1 that have recently been approved in pretreated recurrent and/or metastatic head and neck squamous cell carcinoma (R/M HNSCC) patients. In the clinic, some patients seem not only not to benefit from anti-pd-l1/pd-1 agents but rather to experience an acceleration of tumor growth kinetics (TGK). Patients and methods: We retrospectively compared TGK on immunotherapy and TGK on last treatment in patients with R/M HNSCC treated with PD-1/PD-L1 inhibitors in four French centers. The TGK ratio (, ratio of the slope of tumor growth before treatment and the slope of tumor growth on treatment) was calculated. Hyperprogression was defined as a 2. Results: From September 12 to September 15, 34 patients were identified. Patterns of recurrence included exclusive locoregional recurrence in 14 patients, exclusive distant metastases in 11 patients, and both in 9 patients. No pseudo-progression was observed. Hyperprogression was observed in 1 patients (29%), including 9 patients with at least a locoregional recurrence, and only 1 patient with exclusively distant metastases. Hyperprogression significantly correlated with a regional recurrence ( 2: 9% versus : 37%, P ¼.8), but not with local or distant recurrence. Hyperprogression was associated with a shorter progression-free survival (PFS) according to RECIST (P ¼.3) and irrecist (P ¼.2), but not with overall survival (P ¼.77). Conclusions: Hyperprogression was observed in 29% of patients with R/M HNSCC treated with anti-pd-l1/pd-1 agents and correlated with a shorter PFS. It occurred in 39% of patients with at least a locoregional recurrence and 9% of patients with exclusively distant metastases. No pseudo-progressions were reported. Mechanisms and causality of hyperprogression should further be assessed through prospective controlled studies. Key words: tumor growth kinetics, HNSCC, immunotherapy, hyperprogression, immune checkpoint inhibitors Introduction Every year, over 55 new cases of head and neck squamous cell carcinoma (HNSCC) and 3 related deaths are recorded worldwide [1]. Most HNSCC are locally advanced at the time of diagnosis. Despite a multimodal treatment, more than half of these patients will recur loco-regionally or distantly [2, 3]. Median overall survival (OS) in this patient population is less than a year [4]. In vivo data have recently demonstrated the role played by the immune system to control and avoid tumor formation and progression. Mirroring this theory of immune surveillance, the immunoevasion concept has become a core hallmark of cancer [5]. PD-1 (programmed cell death protein 1) is a transmembrane immune checkpoint receptor that normally limits T cell effector functions within tissues. It is expressed on activated T cells, B cells, natural killer cells and some myeloid cells. Some tumor cells can upregulate the expression of PD-1 ligands (PD-L1 and PD- L2) and further block the antitumor immune response. PD-L1 was found to be expressed in 5% 6% of HNSCC [6]. PD-1/ VC The Author 17. Published by Oxford University Press on behalf of the European Society for Medical Oncology. All rights reserved. For Permissions, please journals.permissions@oup.com. Downloaded from on 4 February 18

2 Original article PD-L1 inhibitors were reported to improve survival in several tumor types [7 9]. In recurrent and/or metastatic (R/M) HNSCC patients, the pivotal phase III CheckMate 141 trial assessed the efficacy of nivolumab, an anti-pd-1 monoclonal antibody as compared to the investigator s choice [1]. The trial was stopped early due to a survival advantage for patients treated in the experimental arm with a statistically significant hazard ratio of.7. Pembrolizumab, another anti-pd-1 monoclonal antibody, was assessed in the KEYNOTE 12 trial in pretreated R/M HNSCC [11]. An overall response rate (ORR) of 18% was reported. Both molecules recently got FDA approval in this indication. Due to their specific activity, the assessment of response to checkpoint inhibitors may be different from that of cytotoxics [12]. Pseudo-progressions, with an initial flare-up followed by tumor shrinkage, were reported in patients with metastatic melanoma [13]. Therefore, it has been advocated to continue immunotherapy after first disease progression provided that the general condition of the patient had improved [12]. Conversely, we have observed in our own experience some patients with R/M HNSCC who progressed very rapidly on immunotherapy, preventing us to safely continue the treatment. We hypothesized that some HNSCC patients might experience an acceleration of tumor growth kinetics (TGK) on immunotherapy [14]. We therefore decided to test this hypothesis by retrospectively comparing TGK on immunotherapy and on last treatment in a cohort of R/M HNSCC patients treated with PD- 1/PD-L1 inhibitors. Patients and methods Patient selection Data from R/M HNSCC patients treated with PD-1/PD-L1 inhibitors in four French centers (Antoine Lacassagne Center, Nice; Léon Bérard Center, Lyon; Curie Institute, Paris; Gustave Roussy, Villejuif) between September 12 and September 15 were retrospectively collected. Patients had a histologically confirmed R/M HNSCC treated with anti- PD-1/PD-L1 therapy. All patients received the PD-1/PD-L1 inhibitors at the recommended phase II dose. Patients had to have an available pre-baseline CT-scan in order to assess TGK before starting immunotherapy. All pre-baseline and on immunotherapy imaging were retrospectively reviewed in order to assess the ORR and the progression-free survival (PFS) according to RECIST 1.1 and irrecist, as well as TGK before starting immunotherapy (TGK PRE ) and on immunotherapy (TGK POST ). Following data were recorded: age at diagnosis, gender, primary tumor location, TNM stage at diagnosis, number of previous lines of systemic therapy, best overall response on immunotherapy, dates of pre-baseline imaging as well as on immunotherapy imaging, patterns of recurrence, location of distant metastases, HPV status, and largest diameter of target lesions. All imaging were non-blindly reviewed by physicians who performed all tumor measurements from pre-baseline imaging to on immunotherapy imaging. For each patient, target lesions were reassessed using RECIST 1.1 [15] and irrecist [12]. Assessments T PRE, T, and T POST stand for the time of pre-baseline, baseline, and first imaging, respectively. S PRE, S, S POST stand for the sum of the largest diameter of target lesions at pre-baseline, baseline, and first imaging, respectively. TGK PRE was defined as the difference of the sum of the largest diameters of the target lesions (according to RECIST) per unit of time between pre-baseline and baseline imaging: (S S PRE )/(T T PRE ). Similarly, TGK POST was defined as (S POST S )/(T POST T ). The TGK ratio ( ) was defined as the ratio of TGK POST to TGK PRE. > 1 indicated tumor growth acceleration, while < < 1and < indicated tumor deceleration and tumor shrinkage, respectively. Hyperprogression was defined as 2(Figure1). Statistical analyses Statistical comparisons were performed using v 2 or Fisher s exact tests for categorical data and student s test or Mann Whitney s test for continuous variables. Smoothing spline ANOVA model was used for evaluation of relation between the TGK PRE and the TGK POST [16]. OS and PFS were estimated and presented graphically using the Kaplan Meier method. Patients were censored at the time of death or last followup. Survival rates at various times and 95% confidence intervals were also estimated. The survival curves were compared using the log-rank test. All statistical analyses were performed in 5% alpha risk or 95% confidence interval using R software on WindowsVR, and gss R package for smoothing spline ANOVA model [16]. Results Patient characteristics Annals of Oncology Among 64 identified R/M HNSCC patients treated with anti-pd- 1/PD-L1 blockers, 34 patients were eligible. The remaining 3 patients were not included in the final analysis because of the absence of available pre-baseline imaging (13 patients), the absence of available post treatment imaging (12 patients), no variation in the sum of target lesions between pre-baseline imaging and baseline imaging (3 patients), and non-measurable disease on prebaseline imaging (2 patients). Median age was 63 years (Table 1). Primary tumor location was oral cavity, oropharynx, larynx, and hypopharynx in 11 (32%), 12 (35%), 5 (15%), and 6 (17%) patients, respectively. Patterns of recurrence included exclusive loco-regional recurrence in 14 patients (41%), exclusive distant metastases in 11 patients (33%), and both in 9 patients (26%). No delays, interruption, or discontinuation of the anti-pd1/pd-l1 agents have been registered for the 34 patients. During the pre-baseline period (pre-baseline to baseline point), 19 patients did not have a systemic treatment (median wash-out duration: 45 days, range days) and 15 patients got a systemic treatment (median prebaseline period: 68 days, range days) without any delay, interruption, or discontinuation. Response to treatment Median follow-up was 1.3 months (95% CI 5 1). Best overall response was complete response in 1 patient (3%), partial response in 3 patients (9%), stable disease in 13 patients (38%), and progressive disease in 17 patients (5%). Median PFS according to RECIST was 2.5 months (95% CI ), while median PFS according to irrecist was 4.9 months (95% CI ). Median OS was 8 months (95% CI ). 166 Sa^ada-Bouzid et al. Volume 28 Issue 7 17 Downloaded from on 4 February 18

3 Annals of Oncology TGK POST =2 =1 Original article Hyperprogression TGK acceleration TGK decelaration TGK PRE Tumor shrinkage Figure 1. Assessment of the tumor growth kinetics ratio ( ) based on TGK on immunotherapy (TGK POST ) and TGK before starting immunotherapy (TGK PRE ). Table 1. Patients characteristics, N (%) 2, N (%) P value Table 1. Continued, N (%) 2, N (%) P value Total 24 1 Immune checkpoint inhibitor.23 PD-1 inhibitor 18 (75) 5 (5) PD-L1 inhibitor 6 (25) 5 (5) Number of previous lines of treatment in the R/M setting.7 2 (8) 1 14 (58) 6 (6) 2 8 (34) 4 (4) Gender 1 Female 5 (21) 2 () Male 19 (79) 8 (8) Tobacco 1 Yes (83) 8 (8) No 4 (17) 2 () HPV status 1 p16 negatif 1 (42) 2 () p16 positif 4 (17) 1 (1) Unknown 1 (42) 7 (7) Continued Previous neck irradiation 1 Yes 23 (96) 1 (1) No 1 (4) Local recurrence.46 Yes 11 (46) 3 (3) No 13 (54) 7 (7) Regional recurrence.8 Yes 9 (38) 9 (9) No 15 (62) 1 (1) Distant metastases.7 Yes 15 (62) 5 (5) No 9 (38) 5 (5) Tumor size at baseline (S).7 Median 11 (46) 6 (6) >Median 13 (54) 4 (4), tumor growth kinetics ratio; PD-1, programmed death 1 receptor; PD-L1, programmed death ligand 1 receptor; R/M, recurrent and/or metastatic; HPV, human papilloma virus. Volume 28 Issue 7 17 doi:1.193/annonc/mdx Downloaded from on 4 February 18

4 Original article A 1 8 < < B Annals of Oncology 5 C 8 8 D 8 < < TGK 6 R Figure 2. Sum of the largest diameter of target lesions over time for patients with (A) a regional recurrence, (B) a local recurrence, (C) distant metastases, and (D) exclusively distant metastases. TGK assessment The evolution of the sum of the largest diameter of target lesions are represented for patients with regional recurrence, local recurrence, distant metastatic recurrence, and exclusively distant metastases are depicted in Figure 2. Hyperprogression was observed in 1 patients (29%), but in only one out of the 11 patients (9%) with exclusively distant metastases and in 9 out of the 23 patients (39%) with at least a loco-regional recurrence (Figure 3). No pseudo-progression was reported. Predictors of hyperprogression Hyperprogression was more frequent in case of metastatic cervical nodes at diagnosis (42% versus 26%), although it did not reach statistical significance. Hyperprogression did not correlate with tumor size at baseline, T stage, M stage at diagnosis or tumor burden at baseline (Table 1). Hyperprogression occurred in 4/19 (21%) patients who did not receive a systemic treatment during the pre-baseline period and 6/15 (4%) patients who got systemic treatment during the pre-baseline period (P ¼.2). Hyperprogression significantly correlated with the presence of a regional recurrence (9% versus 37%, P ¼.8), but not with the presence of a local recurrence. Prognostic value of hyperprogression Hyperprogression was associated with a shorter PFS according to RECIST (2.5 versus 3.4 months, P ¼.3) (Figure 4A), and irrecist (2.9 versus 5.1 months, P ¼.2) (Figure 4B). Hyperprogression was associated with shorter OS (6.1 versus 168 Sa^ada-Bouzid et al. Volume 28 Issue 7 17 Downloaded from on 4 February 18

5 Annals of Oncology Original article A B C 3 D Figure 3. TGK after initiation of PD-1/PD-L1 inhibitors according to pre-treatment TGK in patients with (A) a local recurrence, (B) regional recurrence, (C) distant metastases, and (D) exclusive distant metastases. Graph of smooth curve is represented in blue, the Bayesian confidence intervals at 95% in red lines and the linear regression defining hyperprogression in dot lines. TGK PRE : TGK before starting immunotherapy; TGK POST : TGK on immunotherapy. 8.1 months, P ¼.77), although it did not reach statistical significance (Figure 4C). Discussion In our series, 29% of patients with R/M HNSCC experienced hyperprogression as defined by a exceeding two during anti-pd-1/pd-l1 therapy. This phenomenon was observed in 39% of patients with at least a loco-regional recurrence, and in 9% of patients with exclusively distant metastases. The occurrence of hyperprogression correlated with shorter PFS. No pseudo-progression was reported. Similar results were recently reported in patients with various tumor types treated with PD-1/PD-L1 blockers [17]. The rate of hyperprogressors in this latter study was 9% but authors discuss why their methodology, despite a higher number of included patients, may have led to an underestimation of the incidence of the phenomenon. Hyperprogression also correlated with a poor prognosis in both reports. Three main hypotheses can be made to explain the acceleration of tumor growth. First, PD-1/PD-L1 inhibitors may cause a major immune reaction in some cases. The most frequent occurrence of hyperprogression in patients with cervical node recurrence supports this hypothesis. Mutations or polymorphisms in genes that encode immune modifiers may also explain the variability in the occurrence of the hyperprogression phenomenon. Second, PD-1/PD-L1 inhibitors may promote in some cases tumor proliferation. Indeed, the immune system is known to play a dual role and may contribute to cancer development through direct (i.e. induction of DNA damage by the generation of free radicals) and indirect mechanisms (i.e. angiogenesis and tissue remodeling promotion by the production of growth factors and matrix metalloproteinases induced by inflammation) Volume 28 Issue 7 17 doi:1.193/annonc/mdx Downloaded from on 4 February 18

6 Original article A 1 B 1 Annals of Oncology 8 : : 8 : : Survival rate (%) 6 4 p.value=.35 (Logrank test) Survival rate (%) 6 4 p.value=.23 (Logrank test) Months Months C 1 8 : : Survival rate (%) 6 4 p.value=.77 (Logrank test) Months Figure 4. PFS of hyperprogressors and non-hyperprogressors according to (A) RECIST and (B) irrecist. (C) OS of hyperprogressors and non hyperprogressors. [18 22]. Tumor proliferation may also be promoted by regulatory T cells that suppress antitumor T-cell responses. In this setting, the effect of PD-1 and PD-L1 inhibitors on innate immune cells is unknown. Third, previous irradiation might play a role since almost all cases of hyperprogression occurred in patients who had at least a locoregional recurrence in an irradiated field. Mechanisms behind this hypothesis are unknown. Our study has several limitations. First, our study was retrospective without any control arm. It would have been interesting to perform a similar analysis in a cohort of patients treated in the same setting with other drugs than checkpoint inhibitors. In the absence of a control group, one cannot assess the causality of the hyperprogression to the checkpoint blockade and must evoke non immune related hypothesis but also tumor intrinsic or systemic features in this specific population. However, the hyperprogressions we report here are in line with some of the patients reported on the spider plot of the KEYNOTE-12 who seem to progress very rapidly [11]. While we cannot exclude that the phenomenon of hyperprogression also exist with non-immunotherapeutic agents, the hyperprogressions we describe here in 29% of patients is clinically significant and warrants further exploration. The threshold of 2 for, that we use to define hyperprogression, is very high, since a doubling in size in one dimension means 8 in volume. However, the methodology of a case control retrospective study would not be unbiased. Interestingly, in the pivotal randomized trial CheckMate 141 [1], the OS curves are superimposed during the first 3 months of treatment and then separate significantly. The PFS curves tend to favor the chemotherapy arm during a similar period, until they separate, suggesting that a subset of patients in the nivolumab arm progressed or died earlier than expected. A cost effective way to firmly define the phenomenon of hyperprogression would be the calculation of the in both the nivolumab and chemotherapy arms of the CheckMate 141 patients. Second, we used the RECIST criteria. The selection of target lesions may not reflect perfectly the evolution of the whole tumor burden. It remains to be determined whether the assessment of response and therefore PFS according to RECIST is a good surrogate for OS, as 161 Sa^ada-Bouzid et al. Volume 28 Issue 7 17 Downloaded from on 4 February 18

7 Annals of Oncology demonstrated with chemotherapeutic agents. Third, the small size of our series and the multiple tests performed objectively weaken the consistency of some results. Finally, we hypothesized that TGK would be stable over time without treatment, which might not hold true [23]. However, by comparing pretreatment and on treatment TGK, the better considers the natural history of the disease and mitigate the impact of inter-tumor TGK disparity. Because of all these limitations, our study is only hypothesis-generating and does not allow to firmly attribute the causality of hyperprogression to checkpoint blockade or conclude that hyperprogression significantly affected OS. Our results highlight the necessity to identify predictors of hyperprogression, in order not to treat these patients who might be harmed by immune check point inhibitors. At least, continuing immunotherapy beyond progression should be avoided except in carefully selected patients who really experience a clear clinical improvement during immunotherapy. Biopsies at the time of progression would clearly help understanding the biological mechanisms of hyperprogression. Funding None declared. Disclosure The authors have declared no conflicts of interest. References 1. Jemal A, Bray F, Center MM et al. Global cancer statistics. CA Cancer J Clin 11; 61: Bernier J, Domenge C, Ozsahin M et al. Postoperative irradiation with or without concomitant chemotherapy for locally advanced head and neck cancer. N Engl J Med 4; 35: Blanchard P, Baujat B, Holostenco V et al. Meta-analysis of chemotherapy in head and neck cancer (MACH-NC): a comprehensive analysis by tumour site. Radiother Oncol 11; 1: Vermorken JB, Mesia R, Rivera F et al. Platinum-based chemotherapy plus cetuximab in head and neck cancer. N Engl J Med 8; 359: Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 11; 144: Zandberg DP, Strome SE. The role of the PD-L1:PD-1 pathway in squamous cell carcinoma of the head and neck. Oral Oncol 14; 5: Original article 7. Weber JS, D Angelo SP, Minor D et al. Nivolumab vs chemotherapy in patients with advanced melanoma who progressed after anti-ctla-4 treatment (CheckMate 37): a randomised, controlled, open-label, phase 3 trial. Lancet Oncol 15; 16: Borghaei H, Paz-Ares L, Horn L et al. Nivolumab vs docetaxel in advanced nonsquamous non-small-cell lung cancer. N Engl J Med 15; 373: MotzerRJ,EscudierB,McDermottDFetal.Nivolumabvseverolimus in advanced renal-cell carcinoma. N Engl J Med 15; 373: Ferris RL, Blumenschein G, Jr, Fayette J et al. Nivolumab for recurrent squamous-cell carcinoma of the head and neck. N Engl J Med 16; 375; Seiwert TY, Burtness B, Mehra R et al. Safety and clinical activity of pembrolizumab for treatment of recurrent or metastatic squamous cell carcinoma of the head and neck (KEYNOTE-12): an open-label, multicentre, phase 1b trial. Lancet Oncol 16; 17: Wolchok JD, Hoos A, O Day S et al. Guidelines for the evaluation of immune therapy activity in solid tumors: immune-related response criteria. Clin Cancer Res 9; 15: Weber JS, O Day S, Urba W et al. Phase I/II study of ipilimumab for patients with metastatic melanoma. J Clin Oncol 8; 26: Le Tourneau C, Servois V, Dieras V et al. Tumour growth kinetics assessment: added value to RECIST in cancer patients treated with molecularly targeted agents. Br J Cancer 12; 16: Eisenhauer EA, Therasse P, Bogaerts J et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 9; 45: Gu C, Smoothing Spline ANOVA Models: R Package gss. In Association AS (ed) Journal of Statistical Software Champiat S, Dercle L, Ammari S et al. Hyperprogressive disease (HPD) is a new pattern of progression in cancer patients treated by anti-pd-1/ PD-L1. Clin Cancer Res 17; 23: Guo X, Zhai L, Xue R et al. Mast cell tryptase contributes to pancreatic cancer growth through promoting angiogenesis via activation of angiopoietin-1. IJMS Int J Mol Sci 16; 17: E Manson G, Norwood J, Marabelle A et al. Biomarkers associated with checkpoint inhibitors. Ann Oncol 16; 27: Pollard JW. Tumour-educated macrophages promote tumour progression and metastasis. Nat Rev Cancer 4; 4: DeNardo DG, Andreu P, Coussens LM. Interactions between lymphocytes and myeloid cells regulate pro- vs anti-tumor immunity. Cancer Metastasis Rev 1; 29: Lin EY, Nguyen AV, Russell RG, Pollard JW. Colony-stimulating factor 1 promotes progression of mammary tumors to malignancy. J Exp Med 1; 193: Le Tourneau C, Paoletti X, Coquan E et al. Critical evaluation of disease stabilization as a measure of activity of systemic therapy: lessons from trials with arms in which patients do not receive active treatment. J Clin Oncol 14; 32: Volume 28 Issue 7 17 doi:1.193/annonc/mdx Downloaded from on 4 February 18