Adjuvant therapy options in renal cell carcinoma— targeting the metastatic cascade

Abstract

Localized renal cell carcinoma (RCC) is primarily managed with nephrectomy, which is performed with curative intent. However, disease recurrence occurs in ~20% of patients. Treatment with adjuvant therapies is used after surgery with the intent to cure additional patients, disrupting the establishment, maturation or survival of micrometastases, processes collectively referred to as the metastatic cascade. Immune checkpoint inhibitors and vascular endothelial growth factor receptor (VEGFR)-targeting tyrosine kinase inhibitors (TKIs) have shown efficacy in the treatment of metastatic RCC, raising mounting interest about the utility of these agents in the adjuvant setting. Pembrolizumab, an inhibitor of the immune checkpoint programmed cell death protein (PD1), is now approved by the USA Food and Drug Administration and is under review by European regulatory agencies for adjuvant treatment of patients with localized resected clear cell RCC (ccRCC) based on the results of the KEYNOTE-564 trial. However, the optimal use of immunotherapy and VEGFR-targeting TKIs for adjuvant treatment of RCC is not completely understood. These agents disrupt the metastatic cascade at multiple steps, providing biologic rationale for further investigating the applications of these therapeutics in the adjuvant setting. Clinical trials to evaluate adjuvant therapeutics in RCC are ongoing, and clinical considerations must guide the practical use of immunotherapy and TKI agents for the adjuvant treatment of localized resected RCC.

Short Summary

Adjuvant therapies in renal cell carcinoma (RCC) disrupt multiple steps of the metastatic cascade. Clinical trial experience has yielded promising results for immunotherapy, but failed to clearly demonstrate utility of VEGFR-targeting TKIs, for adjuvant treatment of resected clear cell RCC.

Introduction

Renal cell carcinoma (RCC) is considered localized when the disease is limited to the kidney and perinephric tissues or lymph nodes.¹ Localized RCC is primarily managed with nephrectomy, which is performed with curative intent.² However, disease recurrence is observed in ~20% of patients, and the majority of relapses arise at distant metastatic sites, resulting in a poor prognosis.^3–5 Biological mechanisms that drive metastasis in RCC remain poorly understood, despite advances in animal model research and tumour visualization techniques. Micrometastases are clusters of 10—200 tumour cells that produce clinically detectable metastases.⁶ The goal of adjuvant therapy is to use systemic therapies to disrupt the establishment, maturation or survival of micrometastases, processes collectively referred to as the metastatic cascade, reducing recurrence rates and resulting in cure for additional patients.

Novel immunotherapeutic and targeted approaches have revolutionized the therapeutic landscape for metastatic RCC over the past 15 years. Immunotherapy agents used in RCC inhibit receptor-ligand pairs that modulate the adaptive or innate immune system. These molecular pairs, known as immune checkpoints include programmed death 1 (PD1) and programmed death ligand 1 (PDL1), as well as cytotoxic T-lymphocyte-associated protein 4 (CTLA4) and CTLA4 multiple ligands including cluster of differentiation 80 (CD80) and CD86.⁷ PD1 and CTLA4 are expressed on a variety of immune cells, including cytotoxic T cells, and ligand binding to these receptors induces inhibition of T-cell-mediated cell death.⁸ Specific immunotherapeutic agents used in metastatic RCC include PD1 inhibitors nivolumab and pembrolizumab, PDL1 inhibitors avelumab, and CTLA4 inhibitor ipilimumab.² Targeted therapies used in metastatic RCC include, among others, tyrosine kinase inhibitors (TKIs) that target vascular endothelial growth factor receptor (VEGFR).⁹ Clear cell RCC (ccRCC), the most common subtype of RCC, is frequently characterized by the inactivation of the Von Hippel Lindau (VHL) tumour suppressor protein. VHL is a component of the E3 ubiquitin ligase complex responsible for provoking the degradation of hypoxia-inducible factor (HIF), a transcription factor which in turn causes transcription of hypoxia-inducible genes including VEGF.¹⁰ VHL protein inactivation therefore increases VEGF in the tumor microenvironment.¹¹ VEGF binds to receptors on tumor, vascular endothelial, immune, and stromal cells to drive numerous downstream effects including tumor angiogenesis, cell proliferation, migration.¹² Thus, multiple VEGFR-targeting TKIs have been developed to be used in metastatic ccRCC, including axitinib, cabozantinib, lenvatinib, pazopanib, sunitinib, and sorafenib.²

In first-line treatment of metastatic ccRCC, immune checkpoint inhibitors and VEGFR-targeting TKIs are most frequently given in combination regimens that were shown to improve overall survival (OS) compared with TKI monotherapy in randomized trials.^13–17 These combinations consist of either dual immune checkpoint blockade (ipilimumab plus nivolumab) or a single immune checkpoint inhibitor in combination with a TKI (axitinib plus avelumab, axitinib plus pembrolizumab, cabozantinib plus nivolumab, or lenvatinib plus pembrolizumab).² No direct comparison among these combination regimens is currently available; thus, the choice of first-line therapy for each patient is based on multiple individualized variables including comorbidities, disease location and burden, and psychosocial and economic factors.⁹ However, each combination regimen is considered highly effective, with objective response rates ranging from 42% to 71%.^13–17

The success of these agents in the metastatic setting led to a growing interest in assessing the adjuvant utility of these therapeutics, which resulted in clinical trials to assess the efficacy of immunotherapy and TKI agents in the adjuvant setting in ccRCC. Notably, pembrolizumab, an anti-PD1 monoclonal antibody, received USA Food and Drug Administration (FDA) regulatory approval for adjuvant use based on the results of the KEYNOTE-564 study, a randomized phase 3 clinical trial comparing adjuvant pembrolizumab versus placebo in patients with resected ccRCC¹⁸ However, long-term follow-up OS data are still pending for KEYNOTE-564, and trials investigating the adjuvant benefit of TKIs had mixed results.¹⁹ Moreover, the mechanisms of action of immunotherapy and TKI agents, including the specific oncogenic and immune pathways that they modulate, is not completely understood, hampering the optimal use of these therapeutics in the adjuvant setting.

In this Perspective, we will provide an overview of completed and ongoing clinical trials for adjuvant treatment of RCC, present biologic rationale and clinical considerations for the use of immunotherapy and TKIs in the adjuvant setting.

Evolution of micrometastases

The biologic processes through which primary tumours produce clinically detectable metastases, collectively known as the metastatic cascade (FIG. 1), are not completely understood.^20,21 However, the general consensus is that micrometastases are established when tumour cells progressively acquire a set of molecular traits that enable tumour cells to travel from the site of origin to distant sites, to establish viable metastatic tumours.^21,22 First, tumour cells acquire the ability to invade the basement membrane and intravasate into blood vessels from the primary tumour.²³ This process occurs in part through the tumour production of transforming growth factor beta (TGFβ) and the activation of downstream signalling pathways that induce the formation of actin-rich projections at the basal surface of tumour cells, enabling these cells to cross extracellular barriers.²⁴ During the transit through the circulatory system, tumour cells must acquire the ability to survive threats from mechanical stress and surveying immune cells; these steps are achieved through mechanotransduction of multiple cell-survival signalling pathways in response to the stretching of cellular membrane occurring in microcapillaries, as well as through the upregulation of immune checkpoint molecules, and interaction of tumor cells with myeloid-derived suppressor cells (MDSCs), which facilitates immune evasion.^25–27 Subsequently, tumour cells must induce pre-formation of metastasis niche sites, and then home towards these precursor niches. Tumour cells secrete a variety of pro-angiogenic and hypoxia-dependent cytokines and growth factors, which induce extracellular matrix remodelling, immunosuppression, and vascular hyperpermeability in pre-metastatic organs.²⁸ The specific processes through which tumour cells home towards these sites are poorly understood, but have been best described for bone metastases, in which stromal-cell-derived factor 1a (SDF1a) produced in bone marrow interacts with chemokine receptor 4 (CXCR4) on the cancer cell surface, facilitating cancer cell migration through bone marrow endothelial cells.²⁹ After homing, metastatic tumour cells must co-opt stroma and vasculature of the secondary organ in the new niche, which is achieved through the activation of plasminogen and matrix metalloproteases produced by fibroblasts, macrophages, and endothelial cells, and through the incorporation of existing vascular structures into tumour architecture.^30,31 Finally, after residing in the host tissue, metastatic tumour cells must enter, and eventually exit, a dormancy state that enables immune evasion. Proliferating cancer cells are characterized by constitutive activation of extracellular signal-regulated kinase (ERK), which induces cell-cycle progression and cell division.³² Exposure to intermittent hypoxia and the interaction with various components of the metastatic niche stroma (which are not completely described) induce tumour cell-cycle arrest and subsequent dormancy.³³ Entry into a dormancy state that facilitates immune evasion through downregulation of the cell machinery required for endogenous antigen presentation such as major histocompatibility complex class I (MHC I) and transporter associated with antigen presentation (TAP).³⁴ Tumour cells exit from this dormancy state through a remodelling of the niche architecture, which disrupts cell-cell interactions that help maintain the dormancy state, and through upregulation of immune checkpoint molecules that enable tumour cells to proliferate without being attacked by the host immune system.^32,33 The triggering event for this exit from dormancy is unknown but might involve increased production of oxidized lipids by neutrophils infiltrating the tumour microenvironment and pro-angiogenic factors secreted by tumors.^35,36

Figure 1: Renal cell carcinoma establishes metastases via sequential steps of the metastatic cascade.

Tumour cells progressively acquire a set of molecular traits that enable cells to travel from the primary tumour site to distant sites, and establish viable tumours.^21,22 1. Tumour cells acquire the ability to invade the basement membrane and intravasate into vasculature from the primary tumour.²⁴ 2. During circulation in blood vessels, tumour cells acquire the ability to survive mechanical stress and immune cell surveillance through mechanotrasnduction of multiple cell survival signalling pathways and upregulation of immune checkpoint molecules.^25–27 3. Tumour cells induce pre-formation of metastatic niche sites and home toward these sites, through the secretion of cytokines and growth factors.²⁸ 4. Metastatic cells co-opt secondary organ stroma and vasculature to support tumour growth.^30,31 5. tumour cells in host tissue exit the dormancy state, possibly following provocation by tumour- and immune cell-secreted factors.^35,36 Exit from dormancy state occurs via remodelling of host stromal and vascular architecture and evading of immune surveillance mechanisms.^32,33

Figure 1 was modified after submission and the published version appears different visually from the version shown here. Please find the published version here: Fitzgerald KN, Motzer RJ, Lee CH. Adjuvant therapy options in renal cell carcinoma - targeting the metastatic cascade. Nat Rev Urol. 2023 Mar;20(3):179-193. doi: 10.1038/s41585-022-00666-2. Epub 2022 Nov 11. PMID: 36369389.

Primary tumour resection is an effective method for targeting the first step of the metastatic cascade (tumour cell intravasation). The goal of adjuvant therapy is to disrupt one or more of the subsequent processes, preventing the establishment of viable metastases by tumour cells that managed to disseminate into circulation or distant organs before surgery. Mounting evidence suggests that distant micrometastases are established concurrently with the primary tumour;³⁷thus, late steps in the metastatic cascade offer appealing targets for adjuvant therapy.

Targeting the metastatic cascade– adjuvant immunotherapy

Abscopal tumour regression at metastatic disease sites, including rare complete responses, have been observed following cytoreductive nephrectomy in patients with metastatic RCC.^38–41 These spontaneous regressions are widely accepted to be mediated by lymphocytes and antigen-presenting cells (APCs) infiltrate into the tumour following an unknown trigger and induce cancer-cell-directed cytotoxicity.^42–45 These observations provide clinical rationale for an immune-mediated approach to target micrometastatic sites following surgical resection of a primary tumour in clinically localized disease. Moreover, the long-term, sustained responses to immunotherapy observed in some patients with metastatic RCC makes immunotherapy an appealing tool in the adjuvant setting.⁴⁶

Biological rationale

The most used immunotherapy agents in the treatment of RCC are inhibitors of immune checkpoints PD1 and CTLA4.⁹ PD-1 is an inhibitory receptor that modulates T cell receptor (TCR) and CD28 signalling through recruitment of Src homology 2-containing phosphatase 2 (SHP2).⁴⁷ SHP2 in turn dephosphorylates CD28 and lymphocyte-specific protein tyrosine kinase (Lck); these, along with other SHP2 dephosphorylation substrates, are critical for T cell activation.⁴⁸ PD1 interaction with its ligand PDL1 therefore leads to T cell dysfunction and exhaustion, and immune tolerance.⁴⁹ PD1 blockade is thought to reinvigorate exhausted CD8+ T cells, and to reprogramme the tumour immune microenvironment to favour pro-inflammatory over suppressive myeloid cells.⁵⁰ CTLA4 is another inhibitory receptor that attenuates lymphocyte stimulation by CD28 through outcompeting CD28 for B7 ligand binding.⁵¹ CTLA4 blockade enhances co-stimulation of T cells by lowering the TCR threshold for activation.⁵² Moreover,CTLA4 inhibition is likely to suppress the activity of Treg cells, and drive the expansion of T effector cells.⁵²

Immunotherapy roles in the metastatic cascade

With regards to the metastatic cascade and formation of micrometastases, little is known about the effect of immunotherapy on tumour cell migration and intravasation into or extravasation out of systemic circulation. However, immunotherapy is known to disrupt the PD1—PD-L1 interaction that enables circulating tumour cells (CTCs) to avoid immune surveillance^53,54, and might impair colonization and maturation of micrometastases in the metastatic niche.⁵⁵

Cytotoxicity against micrometastases and circulating tumour cells

Immunotherapy-mediated tumour cell killing is a crucial feature of immunotherapy anti-metastatic function. Micrometastases are thought to (at least transiently) occupy a quiescent state, which enables these cells to resist therapeutic approaches that leverage cell cycle, metabolic demand, or vascular supply to exert antitumor effects.⁵⁶ Immune-mediated cytotoxicity is, therefore, a unique and essential tool to target disseminated dormant lesions, in addition to CTCs.

Mature tumours and the surrounding stroma have several immune suppressive features (such as Treg and MDSC infiltration, or nutrient-poor stroma), whereas CTCs and immature micrometastases lack these features and, therefore, might have increased susceptibility to immune-mediated cytotoxicity.⁵⁷ Blockade of PD1—PDL1 interaction is known to enhance anti-tumour T cell immunity, even in extratumoral sites.⁵⁸ CTC elimination by CD8+ T cell surveillance has been observed since early experiments in mouse models, in which antitumour CD8+ T cell response against early mastocytoma cells was shown to delay metastatic onset in immunocompetent mice compared with T-cell-deficient mice.⁵⁹ Subsequently, results from a study in a mouse model of metastatic melanoma showed that disseminated tumour cells occupied a dormant state, but underwent metastatic growth under depletion of CD8+ T cells.⁶⁰ FIG. 2.1 These data clearly support a role for circulating CD8+ lymphocytes in the elimination of micrometastases and CTCs, but a paucity of data is available about the effect of PD1 inhibition on T effector cells in this setting. Nevertheless, circulating immunosuppressive cells have been observed following resection of localized solid tumours, and evidence suggests a benefit of PD1 blockade in this context. For example, in patients with lung cancer, circulating MDSCs have been associated with inferior patient outcomes: in a study including 88 patients with non-small cell lung cancer, MDSCs were associated with suppression of host monocytic and dendritic cells (DCs), and patients with high numbers of MDSCs (>90^th percentile) had shorter progression-free survival (PFS) and OS than patients with low MDSCs (PFS 5.3 months versus 10.87 months, P=0.005; OS 7.1 months versus 12.9 months, P=0.008).⁶¹ In a study of matched pre- and post-treatment peripheral blood of 14 patients with resected stage III melanoma, PD1 blockade was associated with reduction in circulating Tregs compared with pre-treatment levels (18.1% versus 4.2% of total circulating lymphocytes, p=0.0001).⁶² In another study involving patients with resected localized melanoma, a reduction in Treg-suppressive capacity was observed following treatment with PD1 blockade. In this study, Treg suppressive function was assessed through allogeneic mixed lymphocyte reaction, in which flow-cytometry-sorted peripheral blood Tregs were cultured with allogeneic peripheral blood APCs and CD8+ T cells, and titrated thymidine incorporation was used as an indicator of T cell proliferation and, therefore, reduction of Treg suppressive function. Paired analyses of patient peripheral blood before and after treatment with the PD1 inhibitor nivolumab showed that treatment significantly reduced Treg suppressive capacity in patients who responded to therapy (p=0.010). Furthermore, patients who had no evidence of disease (NED) after 13 weeks of treatment with nivolumab had significantly reduced Treg suppressive capacity compared with patients who underwent tumour relapse within the same time frame (p=0.002). These findings suggest that immune checkpoint blockade augments antitumor immune response in the circulating immune system, and that this effect is essential for treatment efficacy.⁶³ FIG. 2.2

Figure 2: Adjuvant systemic therapies in renal cell carcinoma (RCC) disrupt multiple steps of the metastatic cascade.

1. Programmed cell death 1—programmed cell death ligand 1 (PD1—PD-L1) axis blockade enhances cluster of differentiation 8(CD8)+ T cell antitumor immunity, promoting T-cell-mediated killing of tumour cells in transit in peripheral circulation.^57,58 2. Immunotherapy treatment has been associated with reduction of circulating immunosuppressive regulatory T cells (Tregs), increasing susceptibility of circulating tumour cells to immune-mediated killing.^62,63 3. Neutrophils infiltrate the premetastatic niche and support the maturation of the premetastatic niche, and PD1—PDL1 axis blockade might impair the formation of the premetastatic niche by disrupting the function of PDL1-positive neutrophils.^64–66,190 4. Macrophages have also been shown to promote micrometastatic colonization of the premetastatic niche.^69,71 PD1 blockade has been shown to reprogramme tumour-infiltrating macrophages from an immune-suppressive to a pro-inflammatory phenotype, and might, therefore, impair colonization and tumour formation at the premetastatic niche.⁷³ 5. Vascular endothelial growth factor (VEGF) binding to vascular endothelial growth factor receptor (VEGFR) has been shown to reduce the cytotoxic activity of T cells; thus, VEGFR inhibition might promote CD8+ T cell antitumor activity.^95,191 6. Exposure to tyrosine/kinase inhibitors (TKIs) might enhance natural killer (NK) cell-mediated tumour cell cytotoxicity.⁹⁷

Figure 2 was modified after submission and the published version appears different visually from the version shown here. Please find the published version here: Fitzgerald KN, Motzer RJ, Lee CH. Adjuvant therapy options in renal cell carcinoma - targeting the metastatic cascade. Nat Rev Urol. 2023 Mar;20(3):179-193. doi: 10.1038/s41585-022-00666-2. Epub 2022 Nov 11. PMID: 36369389.

Establishment and colonization of the premetastatic niche

Pro-tumour elements of the immune system have a prominent role in the formation of premetastatic niche and establishment of micrometastases. For example, in mouse models of human breast cancer, a neutrophil infiltration in the premetastatic niche was observed, and was shown to be driven by interleukin (IL)-17 produced by γδ T cells.⁶⁴ After infiltration into the premetastatic niche, neutrophils support the establishment of new tumours by secreting signalling factors including leukotrienes, which directly augment cancer cell proliferation and tumour angiogenesis.^65,66 This process also creates an anti-inflammatory environment that was shown to inhibit anti-tumour CD8+ T cell activity in mouse and in-vitro models of gastric and lung cancer.^67,68 In the mouse model of human gastric cancer referenced above, investigators found that a subset of tumor-infiltrating neutrophils are PDL1-positive, and inhibition of the PD1—PDL1 axis reversed the neutrophil-mediated inhibition of T cell function;⁶⁷ these findings suggest that adjuvant PD1 blockade could function in part by impairing the establishment of new metastatic sites. FIG. 2.3 Similarly, macrophages have been shown to promote metastatic seeding from primary solid tumours, as well as micrometastasis colonization of premetastatic niches. Results from a 2001 study of breast carcinoma metastasis and progression in a mouse breast cancer model showed that tumour-derived and macrophage-derived colony stimulating factor (CSF-1) is necessary for the development of metastatic tumour sites through regulating tumour-associated macrophages (TAMs) infiltration into the metastatic site.⁶⁹ Subsequent experiments from the same group showed that TAM-derived CSF-1 is required for the angiogenic switch occurring in tumour metastases; indeed, depletion of macrophage infiltration or depletion of TAM-derived CSF1 impaired the development of new tumour vasculature, although the mechanism remains unknown.⁷⁰ These findings were supported by another study in which macrophages from mice with or without tumours derived from LLC or B16 cell lines (respectively derived from murine lung cancer and melanoma) were co-cultured with lung endothelial cells; the tumour-stimulated macrophage expressed matrix metalloproteinase 9 (Mmp9) and also induced production of MMP9 by endothelial cells, while macrophages from healthy mice did not produce or induce production of MMP9.⁷¹ MMP9, in turn, was shown to be necessary for establishment of lung metastases in LLC- and B16-tumor-bearing mice: mice carrying disruptions of the Mmp9 gene had a significantly reduced number of lung-infiltrating tumour cells compared with Mmp9 wild-type mice 14 days after injection of the mice with LLC or B16 cells (p<0.01), while there was no impact on growth of the primary tumor.⁷¹ Little is known about the direct effect of PD1 blockade on the metastasis-promoting function of macrophages, but in a mouse model of colon cancer, PD1 expression levels on TAMs seemed to increase with increasing disease stage.⁷² Moreover, in a xenograft murine model of osteosarcoma, PD1 blockade was shown to reprogramme tumour-infiltrating macrophages from an immune-suppressive to a pro-inflammatory phenotype, leading to regression of lung metastases.⁷³ Together, these data suggest that PD1 blockade might have a role in suppressing macrophage-mediated metastatic niche occupation. FIG. 2.4

Preventing micrometastasis exit from the dormancy state

Micrometastases that have established a metastatic niche are known to enter a dormant or equilibrium state, in which cell expansion is paused and cells are able to survive local threats such as metabolic stress, stromal lethality, and immune surveillance.⁵⁶ Micrometastases must then exit this state in order to become viable metastases.²² According to the immune equilibrium hypothesis, quiescent micrometastases escape detection by the immune system, but upon switching to a proliferative state, the antitumour immune response prevents tumour growth.⁵⁶ Results from a study in which pulmonary metastases were assessed in a mouse model of melanoma showed that dormancy of micrometastases is mediated, at least in part, by CD8+ T cells, as outgrowth of visceral metastases was faster in mice with depleted CD8+ T cells than mice with intact CD8+ T cells.⁶⁰ In another study of pulmonary metastases in a human breast cancer mouse model, a discrete population of CD39⁺PD1⁺CD8⁺ T cells preferentially localized to dormant metastases was identified, and the presence of this population correlated positively with delayed metastatic relapse.⁷⁴ The exact effect of immunotherapy on the immune equilibrium of micrometastasis dormancy sate remains poorly understood. However, PD1 blockade is known to reinvigorate CD8+ T cells cytotoxic capabilities previously impaired by prolonged antigen exposure in tumour microenvironment;⁷⁵ this evidence suggests a possible role for immunotherapy in augmenting CD8+ T-cell-dependent maintenance of the dormancy state.

In summary, disruption of the PD1—PDL1 axis through adjuvant immunotherapy is likely to impair the formation of new tumour metastases through multiple mechanisms acting in concert, including augmentation of direct tumour cell killing by host lymphocytes, disruption of the metastatic niche formation, and impairment of micrometastasis exit from the dormancy state.

Targeting the metastatic cascade– adjuvant TKI inhibitors

The utility of VEGFR-targeting TKIs in adjuvant treatment of RCC has not been clearly shown in clinical trials to date.^76–80 However, the biological rationale supporting a possible role for these agents as adjuvant therapies for ccRCC should still be considered, as combination therapies of TKI and immunotherapy remain untested in the adjuvant setting.

Biological rationale

The primary target of VEGFR-targeting TKIs used in the treatment of RCC is VEGFR2, a tyrosine-kinase receptor that interacts with the ligand VEGF.¹¹ The downstream effectors of VEGFR2 are known to regulate tumour angiogenesis and vascular permeability, but have also less well-characterized functions in immune suppression and cancer cell survival.⁸¹ The binding of VEGF to VEGFR2 on endothelial cells induces multiple downstream signalling pathways that drive proliferation, migration, and survival of both vascular endothelial cells, leading to angiogenesis, and tumour cells, leading to tumour growth.⁸¹ Moreover, multiple immune cell subsets including effector T cells, Tregs, MDSCs, and DCs express VEGFR and undergo activity modulation following VEGF binding;⁸² key examples of this are highlighted below. Thus, VEGFR2 blockade with TKIs might prevent metastatic spread affecting multiple steps of micrometasastasis evolution.

TKI roles in the metastatic cascade

Considering the diverse locations and functions of VEGFR2 receptor, VEGFR2-targeting TKIs have the potential to disrupt multiple steps of the metastatic cascade. Specifically, VEGF—VEGFR signalling is involved in tumour cell migration,^83–85 vascular establishment of the metastatic niche,^86,87 and modulation of the immune response to cancer cells⁸².

Impairment of tumour cell migration and metastasis

At primary tumour sites, the vascular endothelium undergoes structural changes that permit tumour cell invasion.²¹ VEGF has been found to mediate cell migration and invasion in multiple solid tumours, although this process has not been well studied in RCC.^84,85,88 Results from a study in which mechanisms of metastasis were assessed through gene-set expression analysis in human osteosarcoma cells showed a positive association between VEGFR2 expression and activation of the STAT3 signalling pathway, activation of immune response pathways, and cytoskeletal rearrangements.⁸⁹ Specifically, VEGFR2 signalling was shown to reduce the phosphorylation of cofilin (deactivated by phosphorylation; cofilin is an essential regulator of actin filament depolymerization and, therefore, of cytoskeletal organization and cancer cell metastasis⁹⁰) through the activation of the STAT3 signalling pathway.⁸⁹ These results suggest that VEGFR2 induces cytoskeletal rearrangements, which are the key mechanism whereby tumour cells form invasive protrusions that permit migration and metastases.⁹¹

Disruption of new vasculature recruitment at metastatic sites

The most widely accepted mechanism of VEGFR TKIs antitumour activity is the detrimental effect on tumour angiogenesis. TKI-mediated VEGF blockade has been shown to decrease the density of tumour-associated vasculature in a human tumour xenograft model for RCC, with an effect observed as early as 24 hours after treatment.⁸⁷ The exact mechanisms through which VEGF inhibition prevents microvasculature development are poorly understood, but sunitinib was found to inhibit the capillary sprouting induced by VEGF and basic fibroblast growth factor (bFGF) in a vascular endothelium spheroid model.⁸⁴ In this study, the mouse-derived, spheroid-forming microvascular endothelial cell line SMHEC4 was used as an in vitro model of angiogenesis, and treatment with sunitinib caused a significant (p<0.01) reduction in cumulative length of capillary sprouts from SMHEC4 spheroids.⁹² Together, the data described above suggest that TKIs disrupt tumor angiogenesis, perhaps by impeding capillary sprouting.

Immune-mediated induction of tumour cell death

VEGFR2 blockade can indirectly lead to RCC cell death through the modulation of antitumour immune response. Anti-CTC immune surveillance is mediated by circulating NK cells and, probably, also by CD8+ T cells.^93,94 VEGF—VEGFR binding has been shown to reduce the cytotoxic activity of T cells.⁹⁵ In a 2012 study, peripheral T cells from patients with ovarian cancer were cultured with anti-CD3 antibodies and recombinant human IL-2 to induce T cell proliferation; cultures were done with or without VEGF to study the impact of VEGF on T cell proliferation.⁹⁵ Additionally, T cells were exposed to low or high VEGF concentrations and then co-cultured with human leukaemia and lymphoma cells that were labelled with intracellular chromium-51, and the cytotoxic potential of T cells was evaluated through the measurement of chromium-51 released into the supernatant.⁹⁵ Results from these assays showed a significant reduction in the proliferative (p<0.05) and cytotoxic potential (p=0.012) of T cells that were exposed to high levels of VEGF.⁹⁵ Results from a subsequent study showed that this immunosuppressive activity of VEGF in T cells is mediated by CD47 signalling; the authors found that VEGF-mediated inhibition of proliferation and TCR signalling in T cells isolated from mice were reversed following CD47 blockade, or in Cd47-deficient mice.⁹⁶ The effects of VEGFR blockade on NK cells are less-well described than the ones on T cells; however, results from a 2019 study showed that sorafenib triggers NK-cell-mediated cytotoxicity in an orthotopic mouse model of hepatocellular carcinoma (HCC)⁹⁷. In this model, when macrophages and NK cells were co-cultured together with or without sorafenib, and then added to mouse hepatocarcinoma cells in-vitro, there was a 4-fold increase in tumour cell killing by immune cells cultured with or without sorafenib; additionally, sorafenib treatment led to increased tumour infiltration by activated macrophages.⁹⁷ Together, these data indicate that VEGFR2 blockade augments the antitumor activity of multiple immune populations, including T cells, macrophages, and NK cells. FIG. 2.5–2.6

Evidence suggests that VEGFR2 blockade might also directly induce tumour cell death, possibly through the modulation of nuclear factor kappa B (NF-κB) and RAS oncogenic pathways.^98,99 However, these mechanisms are not completely elucidated and, therefore, are still under debate.

In summary, results from clinical trials to date failed to show a clear benefit of TKI monotherapy in the adjuvant treatment of RCC; however, evidence suggests that VEGFR-targeting TKIs might disrupt multiple steps of the metastatic cascade, building a rationale for investigating these agents in combination with immunotherapy in the adjuvant setting.

Disadvantages of TKIs in the adjuvant setting

Results from studies in which VEGFR-targeted TKIs were used in the adjuvant treatment of RCC did not enable researchers to establish a clear indication for the use of these agents in the adjuvant setting. One of the theories behind the failure of these studies is the long-accepted Folkman’s mode theory, according to which primary tumours and metastases start as avascular, and angiogenesis only occurs to support the exponential tumour growth when the tumour mass is >1 mm (10⁵–10⁶ cells).^100–102 Moreover, a process known as vascular co-option has been described, in which early micrometastases migrate and grow along pre-existing vasculature.^{103–106.107,108}. These models suggest that factors supporting the early establishment of micrometastatic disease could be independent of angiogenesis.

Microvasculature has also been found to substantially differ between major organs. For example, vascular maturation in vitro was found to be preferentially suppressed by VEGFR2 inhibition in lung endothelium, and by VEGFR1 inhibition in liver endothelium.¹⁰⁹ This evidence suggests that micrometastatic angiogenesis mighty also rely on different VEGF receptors based on the tissue.¹⁰⁹ TKIs used in RCC treatment, all of which target VEGFR2 but also inhibit VEGFR1 and VEGFR3 to various extent could, therefore, have varying efficacy in preventing metastatic disease in different tissues.

Some TKI mechanisms depend on a mature tumour microenvironment. Specifically, many of the immunomodulatory functions of VEGFR-directed TKIs target suppressive immune cells in the tumour microenvironment; these functions include reduction of the suppressive capacity of MDSCs and Tregs, and promotion of DCs maturation.^110–112 Thus, TKI function might be attenuated in the immature microenvironment of CTCs or developing micrometastases.

Finally, mounting evidence suggests that primary tumours and metastases develop in parallel, based on results from numerous studies showing that tumour cells can disseminate during premalignant and preinvasive asymptomatic disease stages.³⁷ Thus, whether targeting early steps of the metastatic cascade (tumour cell migration and establishment of neovasculature) is an effective approach to curing additional patients is currently unclear. The supposed function of PD1—PD-L1 axis blockade in inducing direct tumour cell killing and, possibly, in preventing micrometastasis exit from dormancy state might, therefore, explain the difference in the clinical benefit observed with immunotherapy versus TKI monotherapy in the adjuvant setting.

Reported and ongoing trials

Multiple clinical trials to investigate VEGFR-targeting TKI and immunotherapy agents in the adjuvant setting for RCC are completed or ongoing. Notably, studies of adjuvant therapy in RCC have been almost exclusively performed in ccRCC, except for the ASSURE trial;¹¹³ thus, a paucity of data and treatment options are available for patients with RCC with non-clear cell histology. Efforts have largely focused on establishing the therapeutic utility of immunotherapy and TKI in the adjuvant setting, but in the recently completed EVEREST trial (NCT01120249), the adjuvant use of mammalian target of rapamycin (mTOR) inhibitor everolimus was investigated.¹¹⁴ mTOR is an intracellular serine/threonine kinase that modulates multiple signalling cascades involved in cell survival and proliferation, including production of hypoxia inducible factor (HIF2α). This function is of particular interest in RCC therapy development because HIF2α is a transcription factor that upregulates the expression of hypoxia-inducible genes, including VEGF and erythropoietin, and is implicated in tumour survival and cell proliferation under hypoxic conditions.¹¹⁵ HIF2α is normally degraded by an E3 ubiquitin ligase complex containing the VHL protein subunit; however, the VHL gene is lost in VHL syndrome, as well as in 90% of ccRCC tumors¹¹⁶, and HIF-1α and HIF-2α accumulation resulting from VHL inactivation is thought to be a crucial driver of RCC pathogenesis.¹¹⁷ In the EVEREST trial, 1545 patients who had undergone nephrectomy for localized ccRCC were randomized to receive adjuvant everolimus or placebo, and a no significant difference was seen in disease-free survival (DFS) among the two groups after a median follow-up of 76 months (HR 0.85, 95% CI 0.72–1.00, p1 sided=0.025; the pre-specified significance level of 0.222 was missed).

In the ARISER study, girentuximab, an anti-carbonic anhydrase IX (CAIX) monoclonal antibody, was also assessed as a potential adjuvant therapy in RCC. A high CAIX expression, which is regulated by HIF-1α, is observed on the cell surface of most RCC cells making CAIX a useful diagnostic marker and an appealing therapeutic target in RCC.¹¹⁸ However, in the ARISER study, 864 patients who had undergone nephrectomy for localized ccRCC were randomized to receive adjuvant girentuximab or placebo, and no significant differences were reported between the two groups in terms of median disease-free survival (DFS, 71.4 months and not reached for girentuximab and placebo, respectively; hazard ratio (HR) 0.97, CI 0.79–1.18) or OS (not reached for both groups; HR 0.99; 95% CI, 0.74–1.32). Exploratory analysis of this trial showed that in a subgroup of patients with CAIX-high disease (CAXI score ≥200; score=intensity of staining (1–3) x % of positive cells (0–100%)), treatment with girentuximab was associated with improved DFS (HR, 0.75; 95% CI, 0.55–1.04), although statistical significance was not reached (P=0.08).¹¹⁹

Immunotherapy

Over the past decade, immunotherapy regimens including single-agent PD1 and combined CTLA4 and PD1 inhibition approaches have shown efficacy in the treatment metastatic ccRCC.^120–122 Thus, a mounting interest developed around the application of immunotherapies in the adjuvant setting, and several clinical trials have been conducted or are currently ongoing. This work has led to the approval¹²³ in the USA by the Food and Drug Administration (FDA) of single-agent pembrolizumab for adjuvant treatment of patients with resected ccRCC with intermediate-high risk or high risk of recurrence, based on the results of the KEYNOTE-564 trial.¹⁸ This regimen has been recommended for approval by the European Medicines Agency (EMA) and is awaiting review by the European Commission (EC); evaluation by the Medicines and Healthcare products Regulatory Agency (MHRA) in the United Kingdom is also pending.

KEYNOTE-564 is a randomized phase III clinical trial in which 994 patients with intermediate-high risk or high risk locoregional ccRCC were randomized to receive pembrolizumab (200mg intravenously every 3 weeks) or placebo for up to 1 year. Protocol-defined criteria for intermediate-high risk or high risk of recurrence included tumour stage =2 with nuclear grade 4 or sarcomatoid differentiation, tumour stage ≥3, regional lymph-node metastasis, or stage M1 tumours with no evidence of disease (NED) in which metastases were resected within 1 year from nephrectomy. The primary outcome was DFS from time of randomization to disease recurrence or death, and the secondary end point was OS. At the pre-specified time of interim analysis, median time from randomization to data cut-off date was 24.1 months; 61.1% of patients in the pembrolizumab group had completed the full 17 cycles of treatment (21.3% and 10.5% of patients discontinued treatment owing to adverse events and disease recurrence, respectively). At the time of interim analysis, DFS was significantly prolonged in patients from the pembrolizumab arm (HR 0.68; 95% CI: 0.53–0.87; p=0.0010). In subgroup analyses, a DFS benefit was observed in patients with M0 tumours (HR 0.74; 95% CI: 0.57–0.96) and patients with M1 tumours with NED (HR 0.29; 95% CI: 0.12–0.69), and DFS benefit from pembrolizumab was greater in patients with a PDL1 combined positive score (CPS) ≥ 1 (HR 0.67 95% CI: 0.51; 0.88) compared with patients with a PDL1 CPS < 1 (HR 0.83 95% CI: 0.45–1.51); CPS is calculated as the percentage of PD-L1-expressing tumour and infiltrating immune cells relative to the total number of tumour cells. The particularly robust HR observed for DFS in patients with M1 tumour with NED is not surprising, as these patients with pre-existing metastatic disease probably harbour additional CTCs and micrometastases and are, therefore, likely to benefit from therapeutic modalities that target downstream steps of the metastatic cascade. At 24-month follow-up time, 77.3% of patients in the pembrolizumab group (95% CI 72.8–81.1) and 68.1% in the placebo group (95% CI 63.5–72.2) were alive and recurrence-free. Death was reported in 5% of patients in the overall population (HR 0.54; 95% CI, 0.30–0.96); median OS was not reached in either group (96.6% and 93.5% of patients alive at 24 months in the pembrolizumab and placebo groups, respectively). Grade 3–5 adverse events occurred in 32.4% of patients in the pembrolizumab group and 17.7% in the placebo group; most common adverse events included fatigue (29.7% versus 24.2%), diarrhoea (25.4% versus 22.4%), pruritus (22.7% versus 13.1%), and arthralgia (22.1% versus 18.8%), and additional common adverse effects seen in the pembrolizumab group included hypothyroidism (21.1%), hyperthyroidism (11.9%), and rash (20.1%). These findings are compelling and might lead to a practice-changing development in RCC management.¹⁸ However, these findings should be interpreted in the context of forthcoming median OS data, which was not reached in either group at the time of this report.

The KEYNOTE-564 data should also be interpreted in the context of several additional large, randomized phase III clinical trials to investigate RCC treatment with immunotherapy in the adjuvant setting that have been recently completed with pending publication of data or are ongoing. (TABLE 1) These include the PROSPER (NCT03055013)¹²⁴, IMMOTION 010 (NCT03024996)¹²⁵, CheckMate 914 (NCT03138512)¹²⁶, and RAMPART (NCT03288532)¹²⁷ trials. Complete data are unavailable for the above studies, but limited data were presented for PROSPER, IMMOTION 010, and CheckMate 914 at the European Society for Medical Oncology (ESMO) Congress on September 10^th, 2022; these are summarized below.

Table 1.

Clinical trials investigating immunotherapy in the adjuvant setting in RCC

Trial identifier	Therapeutic agent	Inclusion criteria (tumour stage and grade)	Histology	Primary endpoint(s) and results, if available	Estimated completion†	Ref
KEYNOTE 564 NCT03142334	Pembrolizumab 200mg IV q3 weeks for up to 17 cycles	pT2 N0 (G4 only), pT3a N0 (G3–4), pT3b–T4 N0, pTx N1, M1 NED	ccRCC, might include sarcomatoid features	DFS for treatment vs placebo: HR 0.68; 95% CI: 0.53–0.87; p=0.0010	Dec 2025	160
IMmotion 010 NCT03024996	Atezolizumab 1200mg q21 days for 1 year	pT2 N0 (G4 only), pT3a N0 (G3–4), pT3b–T4 N0, pTx N1	RCC containing clear cell or sarcomatoid component	DFS (assessed through IRF) for atezolizumab vs surgery alone: HR: 0.97; 95% CI 0.74–1.28; p=0.43	May 2022	130
RAMPART NCT03288532	Part A: Durvalumab 1500mg q28 days for 1year Part B: Durvalumab 1500mg q28 days x 1 year + tremelimumab 75mg on weeks 1 and 4	Leibovich score 3–11	All histological subtypes included (except pure oncocytoma, collecting duct, medullary and transitional cell cancer)	DFS, OS (results pending)	July 2024	127
CHECKMATE 914 NCT03138512	Part A: Nivolumab 240mg IV q2 weeks, up to 12 doses, ipilimumab 1mg/kg IV, up to 4 doses given on cycle 1, 4, 7, and 10 Part B: Nivolumab 240mg IV q2 weeks, up to 12 doses	pT2a N0 (G3–4), pT2b–4 N0, pT(any) N1	ccRCC, might include sarcomatoid features	DFS (assessed through BICR) for nivolumab/ipilimumab vs placebo (part A): HR 0.92; 95% CI 0.71–1.19; p=0.5347 Part B results pending.	July 2024	131
PROSPER NCT03055013	Nivolumab 480mg IV, 1 dose given prior to surgery, up to 9 doses given q28 days following surgery	pT2–4 N0, pT(any)N1	ccRCC, might include sarcomatoid features	EFS for nivolumab (HR: 0.97; 95% CI 0.74–1.28; p=0.43)	Nov 2023	129

^†The date on which the last participant in the clinical study was examined or received an intervention to collect final data for the primary outcome measure.¹⁷⁷

q: every; ccRCC: clear cell renal cell carcinoma, DFS: disease-free survival, IRF: independent review facility, OS: overall survival, BICR: blinded independent central review, NED: no evidence of disease, EFS: event-free survival

PROSPER is a phase III, randomized, open-label trial which assessed the DFS benefit of another PD-1 inhibitor, nivolumab, given perioperatively (1 dose prior to surgery followed by 9 adjuvant doses) in patients undergoing nephrectomy for clinical tumour stage ≥2 or any node-positive ccRCC.¹²⁸ Eight hundred and nineteen patients meeting the above-mentioned criteria were enrolled and randomized to undergo surgery with perioperative nivolumab or surgery alone. RFS was similar between the arms (HR: 0.97; 95% CI 0.74–1.28; p=0.43); median RFS was not reached, and grade ≥3 adverse effects were reported in 20% of patients receiving nivolumab and 6% of patients receiving placebo. The trial was stopped early for futility.¹²⁹ IMMOTION 010 is a phase III, randomized, placebo-controlled trial evaluating PD-L1 inhibitor atezolizumab as adjuvant therapy in patients with RCC containing a clear cell or sarcomatoid component with high risk of recurrence, where the definition of high-risk was the same as that used in KEYNOTE-564. Seven hundred and seventy-eight patients meeting the above criteria were randomized to receive one year of adjuvant atezolizumab or placebo. After 44.7 months of median follow-up, investigator-assessed (INV) DFS was similar between the arms (HR: 0.93; 95% CI 0.75–1.15; p=0.495) and median INV-DFS was 57.2 months (95% CI 44.6, NE) for atezolizumab and 49.5 months (95% I 47.4, NE) for placebo; grade ≥3 adverse effects were reported in 27% versus 21%, respectively.¹³⁰ CheckMate 914 is a phase III, randomized, 2-part trial evaluating ipilimumab plus nivolumab versus placebo (part A) or nivolumab monotherapy versus ipilimumab plus nivolumab versus placebo (part B) as adjuvant therapy in patients with ccRCC with high risk of recurrence; high risk patients were defined by the same criteria as KENYOTE 564 and IMMOTION 010, except that patients with tumour stage 2, histologic grade 3 disease were also included. The results of part A were reported at ESMO Congress 2022: 816 patients were randomized to ipilimumab plus nivolumab versus placebo. With 37 months of median follow-up, DFS was similar between the two arms (HR 0.92; 95% CI 0.71–1.19; p=0.5347). Median DFS was not reached with ipilimumab plus nivolumab and 50.7 months with placebo; grade ≥3 adverse effects were reported in 28.5% versus 2.0%, respectively.¹³¹ For the above respective studies, investigators concluded that perioperative nivolumab, adjuvant atezolizumab, and ipilimumab plus nivolumab did not improve DFS compared with placebo in patients with resected ccRCC. RAMPART is evaluating PD-1 inhibitor durvalumab with CTLA-4 inhibitor tremelimumab; the study is ongoing at the time of this report.¹²⁷

In light of the DFS benefit seen with pembrolizumab in the Keynote 564 study, the above results are surprising. The clinical and biologic rationale for why postoperative DFS benefit was seen with adjuvant pembrolizumab but not perioperative nivolumab, adjuvant ipilimumab plus nivolumab, or adjuvant atezolizumab is unknown, but there are multiple possible contributing factors. Notably, PROSPER enrolled and treated patients prior to nephrectomy; surgical outcomes and toxicities could therefore play a role in study outcomes. Furthermore, preclinical studies have suggested that removal of a primary tumour may modulate the systemic immune response to metastases;^132,133 it is unknown whether the persistent presence of an RCC primary tumour could impact patient response to immunotherapy. For each of the above-mentioned studies, differences in inclusion criteria could impact disease relapse risk for enrolled patients, subsequently impacting the number of disease progression events observed in treatment and placebo arms. Furthermore, differences in treatment-limiting toxicity rates might impact the amount of treatment patients received, possibly playing a role in DFS outcomes. Random chance could also contribute to, though is unlikely to entirely explain, the difference in DFS benefit seen across these trials. Finally, and perhaps most importantly, it remains unknown whether there are clinically-relevant differences in the mechanisms whereby PD-L1 versus PD-1 inhibitors, or different of PD-1 inhibitors, modulate the metastatic cascade, or whether underlying molecular disease or host immune features dictate whether an individual patient will respond to these therapeutic mechanisms.

Tyrosine kinase inhibitors

VEGFR-targeting TKIs have been investigated in the adjuvant treatment of RCC in several large, randomized phase III clinical trials(TABLE 2). However, overall results from these studies were conflicting and did not show an improvement in OS; thus, the use of TKIs in the adjuvant setting remains controversial.

Table 2.

Clinical trials investigating TKIs in the adjuvant setting in RCC

Trial identifier	Therapeutic agent	Inclusion criteria (tumour stage and grade)	Histology	Results for primary endpoint	Ref
ASSURE NCT00326898	-Sunitinib 50mg* daily for 4 weeks on and 2 weeks off for 9 cycles (1 year) *Amended to 37.5mg daily with dose escalation -Sorafenib 400mg BID* for 54 weeks *Amended to 400mg daily with dose escalation	pT1b N0 (G3–4), pT2–4 N0, pT(any) N1	All subtypes (except duct-Bellini subtype)	DFS, sunitinib vs placebo: HR 1.02, 95% CI 0.85–1.23, P=0.80) DFS, sorafenib vs placebo: HR 0.97, 95% CI 0.80–1.17, P=0.71	76
S-TRAC NCT00375674	Sunitinib 50mg daily (4 weeks on and 2 weeks off)	pT3 N0 (G2–4), pT4 N0, pT(any) N1	ccRCC	DFS, sunitinib vs placebo: HR 0.76, 95% CI 0.59–0.98, P=0.03	178
PROTECT NCT04321148	Pazopanib 800mg daily* *amended to 600mg daily with dose escalation	pT2–4 N0 (G3–4), pT3–4 N0, pT(any) N1	ccRCC	DFS, pazopanib 600mg starting dose vs placebo: HR 0.86, 95% CI 0.70–1.06, P=0.165	79
ATLAS NCT01599754	Axitinib 5mg BID for 3 years	pT2–4 N0, pT(any) N1	>50% ccRCC	DFS, axitinib vs placebo: HR 0.87, 95% CI 0.660–1.147, P=0.321	77
SORCE NCT004922	Sorafenib 400mg daily* for 1 or 3 years *Amended to 400mg daily with dose escalation	pT1a N0 (G4 only) pT1b N0 (G3–4), pT2–4 N0, pT1b–pT4 N1	All subtypes	DFS, sorafenib vs placebo: HR 1.01, 95% CI 0.83–1.23, P=0.99	140

p, pathologic; T, tumour stage; N, nodal stage; G, tumour grade; ccRCC, clear cell renal cell carcinoma; DFS, disease-free survival; HR, hazard ratio; CI, confidence interval

In the three-arm ASSURE study, 1,943 patients with resected high-grade RCC of any histology received sunitinib, sorafenib or placebo. Patient stratification into recurrence risk categories (intermediate-high, high, or very high) was based on modified UCLA International Staging Criteria and pathologic grading.^134,135 Results from a pre-planned interim analysis showed no differences in DFS between the groups [HR 0.97, P=0.8038 and HR 1.02, P=0.7184 for sorafenib and sunitinib versus placebo, respectively]; furthermore, no differences in DFS or OS were observed in a secondary analyses of patients stratified by risk category or treatment dose.^76,136,137 Conversely, in the S-TRAC study including 615 patients with resected ccRCC that was tumor stage 3 or higher, and/or with regional lymph node metastases, an improvement in DFS (HR 0.76, 95% CI 0.59 – 0.98, p= 0.03) was observed in patients treated with sunitinib compared with patients receiving placebo. However, this DFS benefit did not translate to an OS benefit; median OS was not reached in either arm (HR 0.92; 95% CI 0.66–1.28; p=0.6).^78,80 In the PROTECT study, the efficacy of adjuvant treatment with pazopanib was investigated in 1,538 patients with resected high-risk RCC, defined as any tumour >7cm (T2 according to the AJCC 7^th edition staging guidelines)¹³⁸ with histologic grade 3–4, any tumour extending into major veins or perinephric structures (T3), or any tumour with lymph node involvement⁷⁹. No difference in DFS was seen between patients treated with adjuvant pazopanib and placebo (HR: 0.86, 95% CI 0.70 – 1.06, p=0.16). According to the initial study design, pazopanib was given 800 mg once daily; however, the initial dose was subsequently reduced to 600 mg once daily because the treatment discontinuation rate was higher than expected based on blinded aggregate safety review. Results from a secondary analysis of DFS in patients who started pazopanib treatment at 800mg showed a DFS benefit in patients treated with pazopanib compared with placebo (n=403, HR 0.69, 95% CI 0.51–0.94), whereas no difference in DFS were observed in patients treated with 600mg of pazopanib (n=1,135, HR: 0.86, 95% CI 0.70 – 1.06, p=0.16).⁷⁹ These results suggest that although a higher dose (800mg) of pazopanib has a potential for DFS benefit, this regimen is poorly tolerable, and the more tolerable pazopanib dose of 600mg did not demonstrate a clinical benefit in adjuvant DFS.

Adjuvant treatment with axitinib was investigated in the ATLAS trial, a phase III trial including 724 patients who had undergone nephrectomy for ccRCC that was tumour stage ≥2 and/or node-positive were randomized to receive axitinib or placebo⁷⁷. Results from a pre-planned subgroup interim analysis of DFS in patients with high-risk disease showed DFS improvement in this population (HR 0.641, 95% CI 0.468 – 0.879, P=0.0051); however, DFS was not improved in the intention-to-treat population (HR 0.87, 95% CI 0.660 – 1.147, P=0.3211).⁷⁷ Thus, the trial was stopped for futility.

Finally, adjuvant sorafenib was investigated in the SORCE trial, in which 1,711 patients who had undergone nephrectomy for RCC and had intermediate or high risk of recurrence based on the Leibovich risk model¹³⁹ (described in Table 3) were randomized to one of three arms: sorafenib for 1 year followed by placebo for 2 years, sorafenib for 3 years, or placebo for 3 years¹⁴⁰. The starting dose was 400mg twice daily, but was amended to 400mg daily after the observation of a higher-than-expected treatment discontinuation rate. The initial primary endpoint was DFS after 1 year of sorafenib versus placebo, but was revised to DFS after 3 years of sorafenib versus placebo according to the results from the ASSURE and S-TRAC trials (which were reported after SORCE enrolment was completed but before data were reviewed). No significant difference was observed in DFS, with median not reached in patients treated with sorafenib or placebo for 3 years (HR 1.01; 95% CI 0.83–1.23; p=0.95). Moreover, despite the starting dose modification, more than half of participants stopped treatment before 12 months, and 24% of patients who received sorafenib reported grade 3 hand-foot skin reaction.¹⁴⁰

Table 3.

Prognostic models for risk of recurrence in patients with resected RCC

Model	Type	Histology	Inclusion criteria (pathologic stage)	Variables	Outcome
Cindolo150,179	Formula	Clear cell, papillary, chromophobe	T1–3 N0 M0	Tumour size, local symptoms	RFS
Karakiewicz180	Nomogram	Clear cell, papillary, chromophobe	T1–3 N0–2 M0–1	TNM*, tumour size, Fuhrman grade, histologic subtype, local symptoms, age, sex	CSS
Kattan181	Nomogram	Clear cell, papillary, chromophobe	T1–3 N0 M0	TNM*, tumour size, histologic subtype, local symptoms	RFS
Leibovich139,149	Algorithm	ccRCC	Tany N0–2 M0	TNM*, tumour size, positive node(s) Fuhrman grade, necrosis	MFS
MSKCC182	Nomogram	ccRCC	T1–3b N0 M0	TNM*, tumour size, Fuhrman grade, necrosis, local symptoms	RFS
PRELANE183	Algorithm	Any	Tany N0–1 M0	TNM*, tumour size, positive node(s), histologic subtype, Fuhrman grade, lymphovascular invasion, age, sex	RFS after 5 years
SSIGN184	Algorithm	ccRCC	Tany N0–2 M0–1	TNM*, tumour size, positive node(s), presence of metastases, Fuhrman grade, necrosis	CSS
UISS185	Kaplan-Meier analysis	Clear cell, papillary, chromophobe	Tany N0–2 M0–1	TNM*, Fuhrman grade, ECOG	OS
Yaycioglu186	Formula	Clear cell, papillary, chromophobe	T1–3N0 M0	Tumour size, local symptoms	RFS

^*AJCC Staging Classification 5^th edition (1997)¹⁸⁷ used for all models except MSKCC and Karakiewicz models, for which the 6^th edition (2002)¹⁸⁸ was used, and the PRELANE score, for which the 7^th edition (2009)¹³⁸ was used. ccRCC=clear cell RCC; CSS=cancer-specific survival; ECOG=Eastern Cooperative Oncology Group performance status; MFS=metastasis-free survival; MSKCC=Memorial Sloan Kettering Cancer Center score; OS=overall survival; RFS=relapse-free survival; SSIGN=Stage, Size, Grade and Necrosis score; TNM=tumour stage, nodal stage, metastasis stage; UISS=University of California Los Angeles Integrated Staging System

Based on the results of the available studies, sunitinib is currently the only TKI to have received FDA approval¹⁴¹ for use in the adjuvant setting of RCC in the USA, and no TKIs are recommended for use by the EMA or MHRA. A meta-analysis of phase III randomized clinical trials to assess adjuvant treatment with TKIs in ccRCC showed a pooled HR of 0.89 for OS (95% CI 0.76–1.04) and 0.84 for DFS (95% CI 0.76–0.93). Patients were stratified by high or low risk of disease recurrence; patients at a high risk of recurrence showed one or more tumour features among positive nodes, T4 tumours (extending beyond Gerota’s fascia), and/or T3 tumours with high histologic grade(3–4). After patient stratification, pooled DFS HR in the low- and high-recurrence risk populations was respectively 0.98 (95% CI 0.82–1.17) and 0.85 (95% CI 0.75–0.97). Thus, results from this meta-analysis showed no OS benefit after TKI treatment, but a DFS benefit might exist in overall and high-risk patient populations.¹⁹

Study failures are frequently interpreted as drug inactivity, but alternative possible explanations might exist. These alternative reasons include poor tolerability of specific TKIs, patient selection design with inclusion of patients with low-risk disease leading to low disease progression event rates in placebo arms, and lack of standardization of relapse-risk assessment across trials. Drug tolerability is a particularly concerning issue, as highlighted by the dosage amendment and results of the PROTECT and SORCE studies.^79,140 Data from ATLAS, PROTECT, and S-TRAC trials suggest that only patients treated with a starting full dose of VEGFR TKI would achieve meaningful benefit. However, in real-world settings, dose reductions are common, and maintaining high dosage is challenging owing to toxic effects. Therapeutic dosage in adjuvant trials should generally be in the range of real-world tolerability, and results from clinical trials to date might provide rationale for exploring adjuvant treatment with TKIs at a high starting dose with possibility of dose reduction, similarly to what was done in the S-TRAC trial.⁷⁸ This scenario is supported by initiatives such as the FDA Project Optimus, which is aimed at reforming dose selection in oncology drug development with the aim to educate and collaborate with drug developers to optimize dosing paradigms.¹⁴²

Clinical considerations for adjuvant therapy in RCC

Adjuvant treatment with TKIs and immunotherapy in RCC is promising, but cautious patient selection and risk-benefit assessment is crucial to avoid therapy overuse. Clinical decision-making in the adjuvant setting is ideally centred on three important factors: likelihood that micrometastatic disease is present and will lead to overt relapse; likelihood of response to the proposed therapy; and risk of morbidity or mortality owing to possible toxic effects of the proposed therapy. Response rates can be extrapolated from the results of ongoing clinical trials in RCC (when the results will be available) or from clinical trials conducted with the same agents in metastatic RCC; rates of toxic effects of TKIs or immunotherapy are also available from clinical trial results. However, no well-established clinical model or biomarker to identify patients who will respond to therapy or experience severe adverse effects is currently available. Similarly, predicting disease relapse is challenging, although several methods have been proposed to overcome these issues.

Potential toxic effects

Each VEGFR-targeting TKI used in RCC has distinct adverse effects owing to unique multi-kinase targeting profiles; however, common toxic effects among all VEGFR TKIs exist and include skin rash, hypertension, hepatotoxicity, gastrointestinal toxicity (anorexia, nausea, vomiting, diarrhoea, and/or constipation), and mucocutaneous toxicity (hand-foot syndrome, oral ulcers, anogenital ulcers, rectal fistulae).^143–145 In the reported phase III clinical trials of adjuvant TKI monotherapy, grade 3–4 adverse events occurred in 60.5–72% of patients.^76–78,80 Importantly, TKI toxic effects are often reversible with a dose reduction or withdrawal of the therapeutic agent,^143–145 which is an appealing quality for therapies used in the adjuvant setting, when the risk of permanent adverse effects might have a considerable importance in clinical decision making.

Most commonly observed immune-related adverse events of immunotherapy in RCC include dermatitis and thyroiditis,¹⁴⁶ which are rarely life-threatening; however, other immune related adverse events including colitis, hepatitis, pancreatitis, inflammatory endocrinopathies (such as adrenalitis and hypophysitis), pneumonitis, myocarditis, and encephalitis can all lead to permanent morbidity or mortality.¹⁴⁶ In the KEYNOTE-564 trial, grade ≥3 adverse events were reported in 32.4% of patients receiving pembrolizumab; no deaths were observed. The CheckMate-914, in which the combination of ipilimumab plus nivolumab is being tested in the adjuvant setting for RCC is still ongoing; however, results from the CheckMate-214 trial, in which patients with metastatic RCC were treated with ipilimumab plus nivolumab, showed a grade 3–4 adverse event rate of 46% and eight treatment-related deaths.^18,89,121

Relapse risk assessment

Risk assessment is crucial in selecting patients who will benefit the most from adjuvant therapy. Generally, inclusion criteria for clinical trials of adjuvant TKI and immunotherapy in RCC (Table 1 and 2) were derived from several existing clinical prediction models, which were proposed based on the findings of several large retrospective, single-institution studies in which RCC recurrence rates after nephrectomy were assessed.^147,148 Overall, eight of these models are currently used in clinical practice to predict risk of RCC relapse(TABLE 3). However, the validity of these models remains uncertain. The Leibovich¹⁴⁹ and Cindolo¹⁵⁰ models were validated in independent studies; however, in a prospective validation study in which the eight models were validated using real-world patient outcomes, the predictive ability of all models showed high variability over time.¹⁴⁷

In the last ten years, promising attempts have been made to incorporate histologic, genetic, and immunologic characteristics into RCC relapse prediction models, although validation in large prospective cohorts is still missing. These efforts were based on transcriptomic profiling,^151–154 tumour immunoprofiling,¹⁵⁵ and tumour heterogeneity profiling methods in ccRCC.¹⁵⁶ These methods each aim to designate molecularly-profiled patient subgroups who have high risk of disease relapse. Additionally, ongoing efforts are aimed at developing imaging methods and theranostic (concurrently therapeutic and diagnostic) agents utilizing CAIX as an imaging and anti-tumour therapeutic target.^157,158 However, to date, none of these methods has received regulatory approval or achieved widespread use.

Efforts towards predictive biomarkers

The precise underlying biologic characteristics that drive DFS benefit with adjuvant immunotherapy or VEGFR-targeting TKIs are unknown. In a prospectively designed exploratory analysis of tissue biomarkers in the S-TRAC trial to identify predictors of treatment benefit, the expression of PDL1, CD4, CD8, and CD68 was assessed in tumour samples through immunohistochemistry (IHC). In patients treated with sunitinib (n=101), DFS was longer in patients with CD8⁺ T-cell density ≥ median density than patients with CD8⁺ T-cell density < median density (median not reached (6.83–not reached) versus 3.47 years (1.73–not reached); HR 0.40 (95% CI, 0.20–0.81); P = 0.009], but the same effect was not observed in patients treated with placebo (n = 90).¹⁵⁹ These findings suggest that disruption of the VEGF—VEGFR axis could augment the antitumor activity of cytotoxic T lymphocytes, and provide additional rationale for the exploration of adjuvant combinations of TKI with immunotherapy. A prospective evaluation of biomarkers associated with response to adjuvant pembrolizumab in the KEYNOTE-564 study is planned but has not yet been reported.¹⁶⁰ In future studies, evaluating the quantity and activity of immune cell populations in peripheral circulation and in potential metastatic niche organs would offer increased insight into patient-specific factors that might facilitate or disrupt the metastatic cascade.

Clinical trial end points

In the design and evaluation of clinical trials in the adjuvant setting, defining optimal clinical trial end points is essential. To date, DFS has been used as a primary end point in completed and ongoing clinical trials of adjuvant therapies in RCC. DFS is defined as the time from randomization (or registration, in non-randomized trials) to objective disease recurrence or death from any cause, whichever occurs first.¹⁶¹

DFS is accepted by the FDA and EMA as a basis for drug approval^161,162, but the validity of using DFS benefit as a surrogate for OS benefit is uncertain. OS remains the gold standard outcome for measuring clinical benefit of a therapeutic agent, but the real-world utility of OS as a clinical trial primary end point in the adjuvant setting is limited by the population size and follow-up time that would be required to measure OS.¹⁶³ The ideal adjuvant trial end point would be a surrogate for OS measurable within a reasonable timeframe and sample size. It is unknown whether DFS benefit from a therapeutic intervention leads to OS benefit; therefore the validity of DFS as a substitute for OS as a gold-standard clinical trial endpoint remains unclear and should be prospectively evaluated.¹⁶⁴ Thus, although the continued use of DFS as a surrogate of OS is supported by regulatory agencies, additional clinical end points should be explored. In the metastatic setting, progression-free survival after second line of therapy (PFS2) is endorsed by the EMA as an oncologic clinical trial end point.¹⁶¹ PFS2 is superior to PFS in terms of surrogacy for OS,¹⁶⁵ as well as for the ability to describe outcomes from sequencing multiple therapies, as it takes into account two consecutive lines of therapy. As a result, PFS2 is a particularly helpful endpoint in metastatic ccRCC, where two distinct therapeutic classes (immunotherapy and TKIs) have demonstrated clinical benefit, but optimal sequencing of these approaches is unknown. Likewise, in adjuvant trials for ccRCC, OS benefit has yet to be demonstrated for any intervention, so it is unknown whether postoperative surveillance followed by systemic therapy at the time of disease progression could yield the same survival benefit as systemic therapy started after surgery. Therefore, prospective measurement of time-to-relapse and progression on next line of therapy (TTRP), another endpoint that would take into account two consecutive lines of therapy, would be beneficial for describing therapeutic benefit in adjuvant trials.

Future perspectives

Immunotherapy treatment in the adjuvant setting is promising, but the risk of disease relapse remains high; in the KEYNOTE-564 trial, relapse was observed in 22.7% of patients in the intervention group.¹⁸ Further development of adjuvant immunotherapies is crucial, but efforts must be focused in alternating and augmenting adjuvant options, also considering that novel therapeutic classes emerged in the RCC therapeutic landscape. RCC has a low tumour mutational burden compared with other immunotherapy-responsive solid tumours and, therefore, the mechanism whereby an immune response is provoked by RCC is unclear.¹⁶⁶ Thus, findings about adjuvant immunotherapy in other solid tumour types cannot be translated to RCC, and new prospective studies are needed to address outstanding issues, including the role of multimodality regimens combining immunotherapy and targeted therapy in adjuvant RCC management, and the optimal sequencing of surgery and systemic therapy.

The oral HIF2α inhibitor belzutifan was assessed in a phase II study including patients with VHL-disease-associated RCC; overall response rate, the primary end point of the study, was 49% (95% CI 36–62).¹⁶⁷ Based on this result, belzutifan was approved for treatment of metastatic VHL-disease-associated RCC by the FDA and the EMA,^168,169 and is under accelerated review by the NHRA. Of note, belzutifan has shown overall tolerability: most frequent adverse effects of any grade were anaemia (90%), fatigue (66%), headache (41%), and dizziness (39%), and grade ≥3 adverse events occurred in 33% of patients. Overall, this safety profile is tolerable compared with the one of immunotherapy and TKI agents, and makes belzutifan a promising agent for evaluation in the adjuvant setting, when the risk of adverse effects has a higher importance in the risk-benefit assessment than in the metastatic setting.

Combination regimens of TKI with immunotherapy also remain to be investigated in the adjuvant treatment of RCC. Axitinib with pembrolizumab, cabozantinib with nivolumab, and lenvatinib with pembrolizumab have all shown efficacy in metastatic ccRCC, and are likely to augment immunotherapy function also in the adjuvant setting, considering the known immune-modulatory effects of TKIs.^13,15,16 Novel immunotherapy-based approaches including dual-affinity antibodies and CAR T cells are currently being investigated in the metastatic setting in RCC^170,171 and could offer additional adjuvant treatment options in the future.

The potential applications of immunotherapy, TKIs, and novel therapeutic modalities in nccRCC are poorly understood, and clinical trials are needed to address this current gap in RCC treatment. Response rates to immunotherapy-based regimens in metastatic nccRCC are lower than metastatic ccRCC^172–174; thus, these trials mighty require large patient populations to detect drug activity in the adjuvant setting, and should be approached with multicentre and cooperative group efforts.

Lastly, the optimal sequencing of surgical and systemic interventions to maximize cure rates is unknown. Investigation of systemic agents in the neoadjuvant and perioperative setting is limited. The role of TKI¹⁷⁵, immunotherapy¹²⁸, and the combination of the two regimens¹⁷⁶ in the neoadjuvant setting have been investigated in clinical trials(Table S1). Considering the limited trial experience to date, whether persistence of the primary tumour might affect attempts to modulating the metastatic cascade, or whether different immunotherapy agents might have different degrees of efficacy in the adjuvant or neoadjuvant setting, remains unclear.

Conclusions

Data from large-scale clinical trials of immunotherapy and TKIs in the adjuvant setting for RCC remain largely conflicting in terms of results for DFS benefit of either treatment modality, or are still or pending, but extensive biological rationale exists for the application of both TKI and immunotherapy agents in the adjuvant setting. This rationale consists of a large body of evidence, summarized in this review, that immunotherapy and VEGFR-targeting TKIs disrupt the metastatic cascade at multiple steps, including vascular intravasation, survival in peripheral circulation, extravasation from vasculature into metastatic tissue sites, co-option of host organ vasculature at metastatic sites, and exit from dormancy state. TKI and immunotherapy disrupt multiple steps in the metastatic cascade, and anti-PD1 monotherapy is currently approved for use in the adjuvant setting in the USA and Europe based on the results of the KEYNOTE-564 trial. However, despite this progress, questions remain regarding the optimal adjuvant regimen in RCC and the appropriate method through which patients should be stratified to undergo treatment or observation in the adjuvant setting. The adjuvant utility of additional novel therapies including HIF2α blockade or TKI and immunotherapy combinations remains to be explored; however, these therapies might be promising to optimize the adjuvant management of RCC.