Portal de Periódicos da CAPES

Leucocyte and eosinophil counts predict progression‐free survival in relapsed or refractory classical Hodgkin Lymphoma patients treated with PD 1 inhibition

2017; Wiley; Volume: 181; Issue: 6 Linguagem: Inglês

10.1111/bjh.14705

1365-2141

Ida Hude, Stephanie Sasse, Paul J. Bröckelmann, Bastian von Tresckow, Jesko Momotow, Andreas Engert, Sven Borchmann,

Immune Cell Function and Interaction

The programmed death-1 (PD1, also termed PDCD1) inhibitors nivolumab and pembrolizumab have recently shown remarkable efficacy in relapsed or refractory (r/r) classical Hodgkin lymhoma (cHL) (Ansell et al, 2015; Armand et al, 2016; Younes et al, 2016). Nevertheless, not all patients respond to this treatment and the risk of immune-related adverse events is not negligible. Consequently, predictive biomarkers of PD1 inhibition need to be identified. This study aimed to examine the association of differential blood count parameters at baseline with response and progression-free survival (PFS) in r/r cHL patients treated with the anti-PD1 antibody, nivolumab. We performed a retrospective analysis in r/r cHL patients treated with nivolumab at our department between September 2014 and September 2016. Nivolumab was administered at a dose of 3 mg/kg every 2 weeks. The differential blood count used to predict response was the last one performed immediately prior to the first dose of nivolumab. Response and PFS were defined according to the revised 2007 International Working Group criteria for Malignant Lymphoma (Cheson et al, 2007). Patients without disease progression were censored at the last available radiological response assessment. A Cox proportional hazards model was used to predict the risk of progression. Survival analysis was performed per the Kaplan-Meier method and survival curves were compared using the logrank test. Analyses were done with R version 3.3.1 (R Foundation for Statistical Computing, Vienna, Austria. http://www.R-project.org/). A P-value <0·05 was considered statistically significant and 95% confidence intervals are shown. A total of 30 patients with a median age of 37 years [Interquartile range: 28–44] were included. Eighteen of 30 (60%) patients were male and 12 (40%) female. All patients had received appropriate first-line treatment and all were treated with high dose chemotherapy followed by autologous stem cell transplant (ASCT) at relapse. In total, 21 (70·0%) patients had received prior brentuximab vedotin (BV); the median number of prior treatments was 5 (2–11). We tested the prognostic value of the white blood cell count (WBC, ×109/l), as well as the relative eosinophil count (REC, %), neutrophil count (RNC, %), monocyte count (RMC, %) and lymphocyte count (RLC, %) for PFS in a Cox proportional hazards model. Only the WBC (hazard ratio (HR): 1·110 [1·024; 1·202]) and the REC (HR: 0·469 [0·265; 0·828]) were prognostic for PFS in univariate analysis. In contrast, the RNC (HR: 1·034 [0·991; 1·079]), RMC (HR: 0·942 [0·850; 1·044]) and RLC (HR: 0·973 [0·908; 1·043]) were not prognostic for PFS. We subsequently evaluated if WBC and REC retained prognostic value in a multivariate model including WBC, REC, RNC, RMC and RLC. Indeed, WBC (HR: 1·158 [1·026; 1·308]) and REC (HR: 0·398 [0·175; 0·905]) retained their prognostic value, whereas RNC (HR: 1·218 [0·720; 2·062]), RMC (HR: 1·372 [0·809; 2·326]) and RLC (HR: 1·257 [0·708; 2·230]) remained non-prognostic. Next, we divided patients into those with a high or low WBC or REC. We avoided optimizing cut-off values because of our low case numbers and used the median of both variables to differentiate between a high and a low value, i.e., 7·78 × 109/l for WBC; 1·7% for REC. In the final model including the dichotomized WBC and REC, a high WBC was associated with an increased risk of progression (HR: 5·151 [1·291; 20·550]) and a high REC with a decreased risk for progression (HR: 0·137 [0·027; 0·680]). Finally, we constructed a simple score to prognosticate PFS using the tested cut-off points: WBC ≥ 7·78 × 109/l and REC < 1·7% each added 1 point to the score. Ten patients had a score of 0 (low risk) and only one progressed during treatment (median PFS not available, NA), whereas 5 out of 11 patients with a score of 1 (intermediate risk) (median PFS: 365 [129-NA] days) and 7 out of 9 patients with a score of 2 (high risk) progressed (median PFS: 197 [50-NA] days, Fig 1). The score clearly differentiated 3 risk groups for progression (P < 0·001) (Fig 1). In addition, we evaluated the best response achieved according to the initial risk score (Table 1). To our knowledge, this is the first study presenting a simple score that predicts efficacy of PD1-inhibition in cHL. WBC has been recognized as an adverse prognostic marker in newly diagnosed cHL and is included in the International Prognostic Score (Hasenclever et al, 1998). Nevertheless, its role in r/r cHL has not been established. Peripheral blood eosinophilia is frequently observed among cHL patients and preclinical data suggests that eosinophils have a role in the pathobiology of cHL (Aldinucci et al, 2010). In the immune checkpoint inhibition era, baseline peripheral eosinophil counts have already been correlated with response and survival in melanoma patients (Martens et al, 2016; Weide et al, 2016). Eosinophils can act as effectors of tumour rejection (Gatault et al, 2012), implicating a role in immunosurveillance. We hypothesize that eosinophils play an active role as co-effectors of the immune response once it is reinstated by immune checkpoint inhibition. Interestingly, the impact of activated eosinophils on a T-cell mediated anti-tumour response has recently been characterized in a mouse model (Carretero et al, 2015). Alternatively, eosinophilia could merely be a surrogate marker for a certain type of cHL microenvironment susceptible to immune checkpoint inhibition. Despite the small patient cohort, our simple prognostic model robustly discriminates three risk groups for PFS. Patients in the low risk group tend to stay in long remission whereas patients in the high-risk group tend to progress quickly despite initially often achieving a good response. The score tends to prognosticate PFS better than best response. Nevertheless, validation among r/r cHL patients treated with anti-PD1 antibodies in a separate, larger cohort of patients and determination of optimal cut-off values is warranted before clinical decisions can be based on this model. The use of readily available biomarkers might allow early recognition of non-responding patients in order to treat those with predicted poorer outcomes with alternative regimens or up-front combination treatments. Moreover, further investigation into a possible mechanistic role of eosinophils in checkpoint inhibition might pave the way for a rational design of combination approaches. I.H. received an EHA grant – Junior Short Term Collaboration Award for year 2016, to support stay and work as a guest researcher in Cologne, Germany, with the German Hodgkin Study Group. S.B. received funding from a Cologne Fortune Grant (Reference: 181/2016). Data in this study was in part collected from patients treated in a clinical trial (NCT02181738), with approval of the sponsor Bristol-Myers Sqibb. The sponsor financed the trial and thus indirectly and partly financed data collection but was not directly involved in the present study and did not financially support it. I.H., S.S. and S.B. designed the research, analysed and interpreted data and wrote the manuscript. S.S., B.v.T., P.J.B. and A.E. provided patients and collected and analysed data. J.M. collected and analysed data. All authors critically reviewed the manuscript before submission. PJB declares an advisory role to Millennium Takeda and honoraria and research funding by Bristol-Myers Squibb. BvT declares grants, personal fees, and non-financial support from Novartis; grants, personal fees, and non-financial support from Takeda; personal fees from Amgen; personal fees from Celgene; grants from Merck & Co., Inc. and non-financial support from Bristol-Myers Squibb. AE declares honoraria from Affimed, Bristol-Myers Squibb, Novartis and Takeda and research funding from Affimed, Bristol-Myers Squibb and Takeda. The remaining authors declare no conflict of interest.