Do biomarkers play a predictive role for response to novel immunotherapeutic agents in metastatic renal cell carcinoma?

KEYWORDS:

• Biomarker
• immunotherapy
• tumor mutational burden
• renal cell carcinoma
• PD-L1
• metastases

1. Introduction

Renal cell cancer (RCC) accounts for 3% of solid tumors worldwide, with an estimated incidence of around 330.000 new cases per year. RCC is diagnosed in an advanced stage in about one-third of cases, and another 30% of patients develop metastases after initial nephrectomy [1]. The prognosis of metastatic RCC (mRCC) is generally considered poor, with a 5-year survival rate lower than 20%, though it could be stratified according to several prognostic factors [2].

In the recent years, the therapeutic decision-making approach to mRCC has dramatically changed thanks to the introduction of novel drugs, based on a deeper understanding of the molecular biology of RCC, allowing to reach a remarkable improvement in progression-free survival (PFS) and overall survival (OS), as well as in patients’ quality of life. First of all, the anti-angiogenic therapies targeting the Vascular Endothelial Growth Factor and its receptor, namely bevacizumab and sunitinib, pazopanib, cabozantinib, sorafenib, axitinib, and tivozanib, respectively, deeply changed the treatment scenario of mRCC [3]. Afterward, the immunotherapy-revolution with the advent of immune-checkpoint inhibitors (ICIs) targeting the programmed death receptor 1 (PD1) and its ligand (PD-L1) played a key role in mRCC, similarly to other tumors, such as non-small cell lung cancer (NSCLC) and melanoma. In details, the anti-PD1 nivolumab proved able to confer a statistically significant and clinically relevant improvement in survival with a more favorable safety profile than everolimus in CheckMate 025 trial [4], leading to the approval from Food and Drug Administration and European Medical Association for the treatment of mRCC patients who had received a prior line with anti-angiogenic agents. Since then, several clinical trials investigating ICIs monotherapy or combination or even associated with anti-vascular agents in different treatment lines and settings have been conducted and showed their results [5–8], or are still ongoing. In details, in the first-line setting, the CheckMate 214 trial [5] showed a significant benefit in OS and overall response rate (ORR) for the combination nivolumab plus ipilimumab over sunitinib, the JAVELIN Renal 101 study [6] reported a 6.6-month increase in PFS for avelumab plus axitinib as compared to sunitinib, in the IMmotion151 trial [7] the combination atezolizumab plus bevacizumab proved to prolong PFS versus sunitinib, and, in the recently published KEYNOTE-426 study [8], a significant benefit in OS, PFS, and ORR for the pembrolizumab plus axitinib regimen versus sunitinib was demonstrated.

Altogether, these data appear extremely promising for a radical change in the clinical management of patients with mRCC in the next future. Nevertheless, only a small proportion of mRCC patients achieves a deep and durable benefit from immunotherapy; therefore, the identification of predictive biomarkers for response to the novel immunotherapeutic agents clearly represents an unmet clinical need [9].

In the next session, we will summarize the promising predictive markers and the evidence collected on this topic in the setting of mRCC so far.

2. Biomarkers

2.1. Pd-L1

The first and most studied potential biomarker for response to anti-PD1/PD-L1 agents is the immunohistochemical (IHC) expression of PD-L1. However, even though the negative prognostic role of PD-L1 was shown in the setting of mRCC, its potential predictive value is still controversial [10]. First of all, in CheckMate 025 trial, the expression of PD-L1 on tumor cells, assessed with Dako IHC staining, was associated with poor survival, but not with a response to nivolumab. In details, in both treatment arms an inferior OS was observed in patients with PD-L1 ≥1% as compared to those with PD-L1 <1%, with a similar trend for the 5% cutoff, while nivolumab proved its benefit over everolimus in both PD-L1 subgroups (median OS 21.8 versus 18.8 months in PD-L1 ≥1% and 27.4 versus 21.2 months in PD-L1 <1%, respectively) [4]. Afterward, in CheckMate 214 trial, PD-L1 expression, assessed with the same method, seemed to correlate with a longer PFS but not with an improvement in OS and ORR [5]. Moreover, in the JAVELIN Renal 101 trial, where PD-L1 positivity was assessed with Ventana PD-L1 (SP263) assay on immune cells infiltrating the tumor with a 1% cutoff, the benefit in PFS and ORR obtained in the experimental arm was substantially comparable in the overall population and in the PD-L1 positive subgroup (ORR 51.4% versus 25.7% and 55.2% versus 25.5%; median PFS 13.8 versus 8.4 months and 13.8 versus 7.2 months, in avelumab plus axitinib versus sunitinib arm) [6]. Furthermore, in the phase 2 IMmotion150 trial, randomizing patients to a first-line therapy with the anti-PD-L1 atezolizumab alone or combined to bevacizumab versus sunitinib, a consistent trend in progressively increasing treatment efficacy with higher PD-L1 expression (defined as any intensity staining in immune cells covering absent/<1 (IC0), ≥1 to <5% (IC1), ≥5 to <10% (IC2) or ≥10% (IC3) of tumor area assessed by SP142 assay) was observed, overall in the atezolizumab/bevacizumab arm [12], recently confirmed in the larger casuistic of the IMmotion151 trial [7]. Finally, in the KEYNOTE-426 trial, the benefit of pembrolizumab plus axitinib over sunitinib was observed regardless PD-L1 expression (assessed with IHC 22C3 pharmDx assay according to the combined score, including tumor cells, lymphocytes and macrophages), even though the hazard ratios for PFS and OS seemed to highlight an increased benefit in PD-L1 ≥1 versus <1 subgroups (0.62 (95%CI 0.47–0.80) versus 0.87 (95%CI 0.62–1.23) for PFS and 0.54 (95%CI 0.35–0.84) versus 0.59 (95%CI 0.34–1.03) for OS) [8]. The critical points to remark are: first, it is unclear whether the best biomarker would be PD-L1 or PD1, and, additionally, the role of PD-L2, the other ligand of PD1, has not been fully defined yet. Second, a considerable heterogeneity among the different determination methods is available, since each drug has its own companion antibody. Third, it is unclear whether only tumor or combined tumor and immune infiltrate expression should be assessed and which should be the optimal positivity threshold. Finally, PD-L1 expression may dynamically vary under the treatment pressure or according to the tumor evolution, and it could show a high intratumoral heterogeneity or even remarkable differences between primary tumor and metastases [13]. Therefore, ad-hoc biomarker-driven studies are warranted in order to derive more robust conclusions on the potential role of PD1/PD-L1 expression in predicting response to ICIs.

2.2. Tumor mutational burden (TMB)

TMB is defined as the total number of mutations per coding area of tumor genome, and, recently, it emerged as a potential biomarker of enhanced response to immunotherapy, hypothesizing that the higher the mutational load, the more increased could be the production of neoantigens and therefore the stimulation of the anti-tumoral immune system response. Non-conclusive results have been obtained in other tumor settings, such as NSCLC, considering the heterogeneity in TMB definition and assessment methods.

Remarkably, mRCC represents the tumor endowed with the highest proportion of insertion/deletion alterations, potentially leading to the production of neoantigens [14]. However, while TMB deeply varies according to the different histological RCC subtypes, with chromophobe displaying a very low TMB versus a comparable value of TMB in clear cell and papillary tumors, no correlation was found between TMB and clinically defined prognostic groups according to IMDC and MSKCC [15,16]. Additionally, the exploratory analysis of IMmotion150 trial did not show an association between TMB or neoantigen burden and clinical benefit from immunotherapy [12]. Therefore, prospective trials aimed at assessing the potential predictive role of TMB in mRCC are awaited, and the exploratory analyses results of the ongoing NIVES study (NCT03469713), combining nivolumab and stereotactic radiotherapy in pretreated mRCC patients, could provide evidence on this topic.

2.3. Immune-related adverse events (irAEs)

ICIs induce a peculiar spectrum of toxicities, derived from an enhanced activity of the immune system: the ‘irAEs’, displaying a high variability for what concerns severity, time of occurrence and organs involved, mainly skin, gastrointestinal tract, endocrine glands, lung, kidney and systemic [17]. Non-conclusive evidence has been collected upon a possible association between the occurrence of irAEs and increased survival benefit from ICIs, both anti-CTLA4, and anti-PD1, in the setting of melanoma and NSCLC. Recently, it was shown that, in the Italian Early Access Program for nivolumab in mRCC, patients reporting irAEs had a significantly longer OS (median not reached versus 16.8 months, p = 0.002). No specific criteria to define immune-related toxicity have been established yet, however, it was observed that a selection of AEs potentially determined by the nivolumab-related immune system activation led to observe a stronger association with an improved survival, thus warranting further investigation in this setting [18]. Moreover, a secondary analysis of the CheckMate 214 trial showed that the 24-month OS in patients who discontinued nivolumab plus ipilimumab for treatment-related AEs was 74% (versus 71% in all patients treated with nivolumab plus ipilimumab) and 42% of them were alive and free from second-line therapy at 24 months [19].

2.4. Gene expression signatures

In other tumor settings, specific gene expression profiles suggestive of immune system activation and anti-tumoral response, mainly based on the interferon gamma signaling pathway, have been developed and explored as potential predictive biomarkers of response to immunotherapy [13]. Initial evidence has been collected in mRCC, where the results of the exploratory analyses of IMmotion150 trial showed that the T-effector immune gene signature was associated with PD-L1 expression and CD8 T-cell infiltration, as well as with an enhanced benefit from atezolizumab treatment, in terms of ORR (49% vs 16%) and PFS (HR 0.50; 95%CI 0.30–0.86) [12]. Further studies on this topic are awaited, given the complexity of the molecular and biological cancer scenario in mRCC.

3. Expert opinion

The therapeutic decision-making approach and the clinical management of mRCC patients have deeply changed in the last decade and are still undergoing an evolution, on the basis of the novel agents approved or under study. Specifically, immunotherapeutic drugs have entered the therapeutic scenario of mRCC in the everyday clinical practice, however, several efforts have been made to move them to an earlier disease phase, either in first-line or even as adjuvant treatment after nephrectomy, and studies are ongoing in this setting.

Nevertheless, the proportion of mRCC patients achieving a clinically meaningful and long-lasting benefit from immunotherapy is still low and there is an urgent medical need to improve the selection of patients for ICIs, as well as to maximize the outcomes of treatment. In this light, we underline that the identification of predictive biomarkers for response to immunotherapy is fundamental and that the initial data collected so far on this issue are promising, yet not conclusive.

We want to remind that, from the biologic and metabolic point of view, RCC is a very heterogeneous disease and its clinical course could be largely different according to several prognostic factors related to both patients and pathology. In this light, the lack of predictive factors remains a relevant issue for our decision-making in the clinical practice.

Additionally, all randomized phase 3 trials were performed in well-selected populations and less evidence is now available in the real-world setting.

• Big data may enable the discovery of more nuanced relationships between prognostic/predictive factors and clinical outcomes and could be considered as hypothesis-generating studies.

• Real-world and big data may permit to translate the molecular features of the disease into a personalized approach for every patient and would allow to focus on the research outcomes that matter for the patients.

Therefore, large, prospective, biomarker-driven studies are warranted to identify new biomarkers with high sensitivity and specificity in the prediction of response to ICIs.

Declaration of interest

G Procopio declares receiving honoraria for advisory board from Bayer, Bristol Myers Squibb, Ipsen, Merck, Novartis, and Pfizer. E Verzoni declares receiving honoraria for advisory board from Pfizer, Bristol Myers Squibb, Ipsen, and EUSA pharm. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer Disclosures

One of the reviewers on this paper has received research funding from and consulted for Pfizer, Merck, BMS, and Roche. Other peer reviewers on this manuscript have no relevant financial relationships or otherwise to disclose.

Additional information

Funding

No funding was received for this manuscript.