Current status of antigen-specific T-cell immunotherapy for advanced renal-cell carcinoma

ABSTRACT

In renal-cell carcinoma (RCC), tumor-reactive T-cell responses can occur spontaneously or in response to systemic immunotherapy with cytokines and immune checkpoint inhibitors. Cancer vaccines and engineered T-cell therapies are designed to selectively augment tumor antigen-specific CD8⁺ T-cell responses with the goal to elicit tumor regression and avoid toxicities associated with nonspecific immunotherapies. In this review, we provide an overview of the central role of T-cell immunity in the treatment of advanced RCC. Clinical outcomes for antigen-targeted vaccines or other T-cell-engaging therapies for RCC are summarized and evaluated, and emerging new strategies to enhance the effectiveness of antigen-specific therapy for RCC are discussed.

KEYWORDS:

• Renal cell carcinoma
• tumor-infiltrating T cell
• tumor antigen
• t cell receptor
• cancer vaccine

Introduction

Renal-cell carcinoma (RCC) accounts for 90% of malignant kidney neoplasms in adults and is the eighth most common cancer in the United States.¹Although nephrectomy for localized tumors can be curative, ~30% of patients develop metastatic disease.² RCC is resistant to cytotoxic chemotherapies³ and despite the development of targeted molecular therapies in the mid 2000’s including vascular endothelial growth factor (VEGF) receptor-selective tyrosine kinase inhibitors (TKIs) and mammalian target of rapamycin (mTOR) inhibitors, primary or acquired tumor resistance is common and metastatic RCC (mRCC) is generally considered incurable.

However, among common epithelial cancers, mRCC is uniquely sensitive to systemic immunotherapy. By the early 1990’s, systemic cytokine therapies with interferon-alpha (IFN-α) or interleukin-2 (IL-2) were the standard frontline treatment options for mRCC. Although cytokines were largely replaced by molecularly targeted drugs, recent phase III studies of immune checkpoint blocking antibodies targeting programmed cell death protein-1 (PD-1) or programmed cell death ligand-1 (PD-L1) have shown a positive benefit in comparison to targeted therapy. Immunotherapy is now re-established as the most common initial treatment for advanced RCC. Nevertheless, outcomes are heterogeneous ranging from durable tumor regression to primary refractory disease. Complete radiographic responses remain uncommon, occurring in 3–9% of patients receiving frontline immune checkpoint inhibitors.^4–6

The antitumor effects produced by both cytokine and immune checkpoint blocking agents are thought to be mediated by tumor-reactive T cells. The long history of clinically applied immunotherapy for advanced RCC has encouraged the development and testing of numerous antigen-targeted T cell-mediated treatment options for mRCC with the goal of selectively augmenting anti-tumor activity and avoiding toxicity. This review provides an overview of the central role of T-cell immunity in the treatment of advanced RCC. Clinical outcomes for antigen-specific vaccines or other T-cell-engaging therapies for RCC are evaluated, and emerging new strategies to enhance the effectiveness of antigen-targeted therapy for RCC are discussed.

Early evidence for RCC-specific T-cell immunity

Since the first report in 1928,⁷ the phenomenon of spontaneous tumor regression has been more frequently associated with mRCC than most other cancer types. The frequency of spontaneous regression in RCC patients has been estimated at 1% and has been observed in both primary tumors and metastatic lesions.⁸ Host immune system activation is the leading hypothesis for the mechanism. This view is supported by the observation that spontaneous regressions are often preceded by feverish infection.⁹ Early studies therefore looked for evidence of spontaneous T-cell responses recognizing RCC.

The possibility of RCC eliciting cytotoxic immune responses was supported by the finding that 100% of RCC tumors expressed MHC class I, and 93% retained expression during tumor progression and metastasis.¹⁰ In 1991, an RCC-specific cytotoxic T-lymphocyte (CTL) line was isolated from a primary tumor and displayed lytic specificity and IFN-γ production only in co-culture with autologous tumor but not lymphoblasts or allogeneic RCC tumor.¹¹ Subsequently, a CTL culture was isolated from a human leukocyte antigen (HLA)-A*02⁺ RCC tumor with limited representation of Vα genes and demonstrated cytotoxicity towards allogeneic HLA-A*02⁺ RCC tumors, suggesting T-cell clonal expansion in the tumor and the existence of shared RCC tumor antigens.¹² These studies were followed by the discovery of multiple RCC-associated antigens that were the targets for spontaneous T-cell responses in RCC patients.^13–20 These antigens were shown to arise through point mutations, the aberrant overexpression of genes and antisense transcripts, or alternative open reading frames likely generated from frameshift mutations (Table 1).

Table 1. RCC antigens recognized by spontaeously arising tumor-reactive CD8 CTL clones

Gene	HLA restriction	Peptide sequence	Antigenic modulation	Expession on normal tissue	Shared antigenicity	T-cell origin	Reference
HLA-A*02	HLA-A*02	n/a	point mutation	Not on PBL	No	RCC TIL	13
HSP70–2	HLA-A*02	SLFEGIDIYT	point mutation	No	No	RCC TIL	14
RAGE1	HLA-B7	SPSSNRIRNT	n/a	Retina	Yes, 1.7% RCC cases	RCC TIL	15
RU1	HLA-B*51	VPYGSFKHV	alternative cleavage	Unknown	Shared cytotoxicity against melanoma	RCC PBL	16
M-CSF	HLA-B*3501	LPAVVGLSPGEQEY	alternative ORF	Kidney and liver	Yes	RCC TIL	17
iCE	HLA-B*0702	SPRWWPTCL	alternative ORF	Unknown	Yes	RCC TIL	18
RU2	HLA-B7	LPRWPPPQL	antisense transcript	Kidney, liver,testies, and bladders	Yes, 100% RCC cases	RCC PBL	19
FGF-5	HLA-A*03	NTYASPRFK	post-translational splicing	Kidney and brain	Yes	RCC TIL	20

Abbreviations: HLA – human leukocyte antigen; TIL – tumor-infiltrating lymphocyte; PBL – peripheral blood leukocyte; n/a – not applicable; ORF – open reading frame

Successful therapeutic manipulation of tumor-reactive T-cell responses was first demonstrated with systemic cytokines that came into clinical use for mRCC in the 1980s and 1990s. IL-2 activates post T-cell receptor (TCR) signaling in T-cells and promotes CD8⁺ T-cell proliferation, effector function and survival. In 1992, high-dose IL-2 was approved by the FDA as a monotherapy for mRCC. Objective responses (OR) were observed in 15% to 30% of patients receiving high-dose IL-2, with 5–8% of patients achieving durable and unmaintained complete responses (CR).^21–23 Type I interferons represent a component of an innate immune response to virus infection or neoplastic cells and serve a subsequent key role in priming the host adaptive immune response. Clinical trials of recombinant IFN-α treatment for mRCC have shown a response rate of 15%. However, CRs (< 5%) were incrementally less frequent or durable than for IL-2.^24–26 Collectively, the early evidence for spontaneous T-cell immunity to RCC antigens and clinical success in some patients receiving T cell-mediated systemic immunotherapy encouraged the search for more effective therapy strategies and focused attention on defining the mechanisms regulating tumor-reactive T-cell immunity as well as tumor escape from immune surveillance.

Phenotype of tumor-infiltrating lymphocytes (TIL) in RCC

More contemporary analysis of over 7,000 tumors from The Cancer Genome Atlas (TCGA) demonstrated clear cell RCC (ccRCC), the most common subtype of RCC, has the highest CD8⁺ T-cell infiltration among 23 solid tumor types.²⁷ Within the RCC tumor microenvironment, T-cells are the most prevalent immune subset (50%) followed by tumor-associated macrophages (25%), natural killer (NK) cells (9%), B cells (4%), and other immune cells including plasma cells, dendritic cells (DC’s), and neutrophils.²⁸ Moreover, ccRCC had the highest cytolytic activity index (geometric average of GZMA and PRF1 expression) compared to 17 other human cancers.²⁹ These observations suggest RCC can elicit tumor-reactive T-cell responses. Accumulating evidence emphasizes not just the quantity but also the functional quality of T-cells in the tumor microenvironment on favorable outcomes in mRCC. Features of TIL including low expression of immune checkpoint proteins,^30,31 a T helper 1 (Th1)-type phenotype possibly mediated by chemokine recruitment of T-cells into the tumor,³² above-median CD8⁺ T-cell/regulatory T-cell (Treg) and Th17/Th2 ratios,³³ in addition to the presence of mature dendritic cells and higher adaptive immune response gene expression in the tumor microenvironment are associated with favorable prognosis.³⁴

Another important aspect of antitumor efficacy is T-cell proliferation capacity in the tumor microenvironment that may reflect a response to tumor antigen. The ability of CD8⁺ T-cells to expand in tumors marked by Ki-67 expression is an independent favorable prognostic factor in RCC.³⁵ Recurrent TCR transcripts marked by uniquely rearranged complementarity-determining region 3 (CDR3) is another powerful marker indicating antigen-driven proliferation of individual T-cell clones in tumors. TCR transcripts with the same Vβ gene can represent up to 25–30% of the tumor-infiltrating TCR repertoire in RCC.^36,37 Our research group’s ongoing single-cell analyses of RCC TIL identified clonally expanded T-cells unique to the tumor microenvironment not detected in normal adjacent renal cortex or peripheral blood that represented 8 to 24% of the total TIL repertoire. These expanded clones are enriched for an effector or effector memory CD8⁺ T-cell phenotype consistent with a tumor-reactive effector population. Moreover, IHC of the same tumor specimens revealed that Ki67⁺CD8⁺ T-cells were significantly more abundant in tumor compared to non-tumor regions, suggesting tumor antigen-driven expansion. However, a major challenge in the field is the technical difficulty in identifying the cognate antigens associated with clonally expanded TCRs.³⁸

Mechanisms that limit the antitumor activities of T-cells in RCC have been identified. CD8⁺ T-cells infiltrated into RCC tumors can be non-responsive to ex vivo stimulation, lack granule mobilization, cytolytic activity and cytokine production,³⁹ a phenotype associated with a low level of TCR-distal signaling.⁴⁰ Inhibitory co-receptors expressed on activated T-cells may contribute to T-cell anergy in the tumor microenvironment and RCC escape from immune surveillance. Cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) regulates activated T-cells by competing with CD28 to bind to CD80 and/or CD86 ligands,⁴¹ decreasing IL-2 secretion, T-cell proliferation⁴² and promoting T-cell apoptosis.⁴³ PD-1 is expressed on activated effector T-cells, binding to PD-L1 and PD-L2, and limiting the T-cell effector responses.⁴⁴ PD-L1 was shown to be expressed by 23.9% of ccRCC tumors.⁴⁵ Patients with PD-L1⁺ tumors had significantly lower 5-year survival and higher metastatic cancer progression,⁴⁵ and patients with TIL expressing PD-1 were more likely to have aggressive PD-L1⁺ RCC tumors.⁴⁶ Similarly, infiltration of PD1⁺ T-cells was shown to predict distant metastases in ccRCC.^30,47 Two other inhibitory receptors, lymphocyte activation gene 3 (LAG-3), and T-cell immunoglobulin and mucin domain-containing 3 (TIM-3) also have been found to play important roles in RCC TIL exhaustion. Analysis of TCGA data revealed that tumor LAG-3 expression was associated with poor overall survival (OS).³⁰ TIM-3 is induced by Th1 cytokines after T-cell activation in the tumor microenvironment⁴⁸ and tumor-infiltrating PD-1⁺ CD8⁺ T-cells coexpressing TIM-3 were associated with a highly aggressive RCC phenotype.⁴⁹ In addition, TIM-3 expression has been detected on intratumoral Tregs,⁴⁹ contributing to the accumulation of dysfunctional CD8⁺ T-cells.⁵⁰

Immune checkpoint blockade therapy in RCC

Given the immunotherapy responsive phenotype for RCC as well as insight into mechanisms for tumor resistance, mRCC was one of the first cancers treated with immune checkpoint blocking antibodies. In 2015, the PD-1 blocking antibody nivolumab was approved by the FDA for the treatment of advanced RCC that had failed prior VEGF pathway targeted therapy. Approval was based on the phase III CheckMate 025 trial that showed superior OS and higher response rate (25 vs 5%) for nivolumab versus the mTOR inhibitor everolimus.⁵¹ Nivolumab was subsequently combined with the CTLA-4 blocking antibody (ipilimumab) as a frontline treatment regimen and tested versus the TKI sunitinib in the CheckMate 214 trial.⁴ In the primary analysis cohort of intermediate- and poor-risk RCC patients, the nivolumab plus ipilimumab treatment arm had superior OS, progression-free survival (PFS), and overall response rate (ORR) (42 vs 27%) versus sunitinib and the combination of nivolumab plus ipilimumab received FDA approval in 2018. Immune checkpoint blockade has also been combined with antiangiogenic TKI’s for frontline therapy of advanced RCC. In the phase III KEYNOTE-426 trial the PD-1 antibody pembrolizumab plus axitinib was compared to sunitinib monotherapy⁵ and in the phase III JAVELIN Renal 101 trial, avelumab (anti-PD-L1) plus axitinib was also compared to sunitinib. Both studies showed positive clinical benefit for the immune checkpoint containing regimen leading to FDA approval for both doublets in 2019.⁶ Taken together, systemic therapy with a two-drug regimen incorporating an immune checkpoint inhibitor is now the preferred treatment for most patients with advanced RCC.

RCC-antigen targets associated with immune checkpoint blockade

Tumor neoantigens created by tumor-specific mutations were first associated with response to immune checkpoint blockade for melanoma and NSCLC, tumors with a high single nucleotide variant mutation burden [10 to 400 mutations/megabase (Mb)].^52–54 Tumor mutation burden (TMB) has been evaluated as a more generalizable biomarker for response to immune checkpoint blockade showing a strong correlation for most tumor types (Figure 1) and high TBM [≥10 mutations/Mb)] is a validated tumor-agnostic biomarker for selecting treatment with pembrolizumab.^55,56 However, despite a relatively low number of coding somatic mutations (1.8 mutations/Mb) RCC has a disproportionately high response rate to PD-1 inhibition (Figure 1), a discrepancy further exacerbated by a higher response rate (38%) observed in more recent front-line data with PD-1 inhibition.⁵⁷ Of note, in a study of 19 cancer types, RCC was found to have the highest number and proportion of insertion and deletion (indel) mutations representing an alternate mechanism for neoantigen formation.⁵⁸ However, biomarker discovery efforts with tumor samples from RCC patients enrolled in studies of nivolumab monotherapy,⁵⁹ or front-line regimens with nivolumab plus ipilimumab,⁶⁰ avelumab plus axitinib⁶¹ or atezolizumab plus bevacizumab⁶² are consistent for failing to show an association of TMB or neoantigen density with clinical benefit for patients receiving immune checkpoint blockade. These data suggest other mechanisms separate from neoantigen expression may play an important role contributing to RCC immunogenicity.

Figure 1. RCC has a high objective response rate to immune checkpoint blockade despite low TMB. Median number of coding somatic mutations per megabase (MB) of DNA for 27 tumor types is plotted in log scale versus objective response rates for patients who received PD-1 or PD-L1 inhibitors as described in published studies. RCC is within the top 25% of objective response rates while being within the bottom 33% of median number of coding somatic mutations per MB. MMRd denotes mismatch repair-deficient, MMRp mismatch repair proficient, and NSCLC non-small cell lung cancer. Reproduced from Yarchoan, Hopkins & Jaffee⁵⁵ with copyright permission

Accumulating evidence suggests that human endogenous retrovirus (hERV)-derived antigens contribute to immune checkpoint inhibitor-associated responses. These germline-encoded elements have integrated into the genome and are identified by unique sequences derived from 5′ and 3′ long terminal repeats (LTRs) and retroviral genes. These normally quiescent sequences can be translationally re-expressed due to epigenetic dysregulation preferentially in RCC versus other tumors.^63–67 Mechanistically, the up-regulation of the HIF2α transcription factor in ccRCC was found to target a response element in the proviral 5′LTR, turning on hERV-E expression in tumors.⁶⁸ HERV-encoded proteins can be immunogenic and T-cells specific for epitopes from the hERV-E 5′LTR and the hERV-E envelope gene have been identified.^69–71 HERVs may harbor immunogenic hotspots with multiple epitopes per retroelement, and the conserved retroviral epitopes are widely shared between patients.⁷¹ In small studies, expression of select hERV elements has been associated with better clinical outcomes for ccRCC treated by immune checkpoint blockade including higher response rates and longer PFS further encouraging the study of hERV as biomarkers or as therapeutic targets.^71–73

Inducing tumor-specific T-cell immunity against RCC-associated antigens

The longstanding recognition of RCC as an immune responsive tumor has encouraged the development and clinical testing of numerous antigen-targeted vaccine and T-cell-engaging therapies. RCC tumor antigens defined by spontaneously arising T-cells were either patient-specific, expressed in normal tissues, or had low-frequency expression by tumors and were not suitable candidates for vaccine development. Thus, selection of RCC antigen targets has focused on inducing immunity against proteins that demonstrate tumor-specific expression and are broadly shared between tumors.

Single antigen vaccines and T-cell engaging therapies (Table 2)

Mutation or inactivation of the Von Hippel-Lindau (VHL) tumor suppressor gene occurs in most ccRCC tumors,⁵³ and results in constitutive activation of HIF transcription factors. Overexpressed proteins encoded by hypoxia-inducible genes represent widely shared target antigens for ccRCC tumors including carbonic anhydrase IX (CAIX), vascular endothelial growth factor receptor 1 (VEGFR1), and hypoxia-inducible protein 2 (HIG2). Mutant versions of the VHL protein itself have also been evaluated as a vaccine target.

Table 2. Clinical testing of single antigen-specific vaccination or T-cell engaging therapy for advanced RCC

Immune Target	Antigen Formulation	Adjuvant	HLA restriction	Combination therapy	Phase	Enrollment		Cellular Immune Response	Clinical Outcome	Reference
						Key Criteria	N
CAIX	3 HLA class I synthetic peptides	Montanide ISA-51	HLA-A*24	None	I	cytokine-refractory ccRCC	N = 23	Peptide reactive CTL response in 16/21 (76%) patients	3/23 PR (13%); lung metastases	74
	1 HLA class I plus 1 HLA class II synthetic peptide loaded on DC	keyhole limpet hemocyanin (KLH)	HLA-A*02 and HLA-DR	None	I	cytokine-refractory ccRCC	N = 8	No CTL response detected	No objective responses	75
	CAR T	None	not restricted	SQ IL-2	I/II	TKI or cytokine-refractory ccRCC	N = 12	Post infusion persistence < 5 weeks	No objective responses	76
VEGF-R1	1 HLA class I synthetic peptide per patient; matched to HLA expression	Montanide ISA-51VG	HLA-A*02:01 or *24:02	None	I	TKI or cytokine-refractory ccRCC	N = 18	Peptide reactive CTL response in 15/18 (83%) patients	2/18 PR (11%); lung and bone metastases	77
Hypoxia-inducible protein 2	1 HLA class I synthetic peptide	Montanide ISA-51VG	HLA-A*0201 or A*0206	None	I	TKI or cytokine-refractory RCC	N = 9	Peptide reactive CTL response in 8/9 (89%) patients	No objective responses	78
mutant VHL	1 HLA class I patient-specific synthetic peptide	Montanide ISA-51	mutant peptide selected based on autologous HLA motif	None	I	ccRCC expressing non-synonymous mutated VHL gene	N = 6	Peptide reactive CTL response in 4/5 (80%) patients	No objective responses	79
5T4 (trophoblast glycoprotein, TPBG)	Recombinant MVA expressing 5T4 (MVA-5T4; TroVax)	None	Not restricted	Low-dose IL-2	II	metastatic clear cell or papillary RCC	N = 25	8/18 (45%) 5T4-specific ELISPOT;	2 CR, 1 PR (12%)	80
				High-dose IL-2	II	metastatic clear cell or papillary RCC	N = 25	13/23 (56%) 5T4-specific CD8 + T-cell response	3 tumor regressions (12%) rendered disease free with surgery	81
				IFN-α	I/II	metastatic clear cell or papillary RCC; no prior systemic therapy	N = 11	5/11 (45%) 5T4-specific ELISPOT;	No objective responses	82
				None or IFN-α	II	treatment refractory metastatic clear cell or papillary RCC	N = 28	7/21 (33%) 5T4-specific ELISPOT;	1 PR (4%)	83
				Sunitinib, IL-2, or IFN-α	III (TRIST)	metastatic ccRCC; no prior systemic therapy	MVA-5T4 (n = 365) or placebo (n = 368)	Not evaluated	Primary endpoint OS was not increased; Median OS 20.1 (MVA-5T4) vs 19.2 mo (placebo); HR 1.07 (P = 0.55)	84
	5T4-specific Fab fused to the SEA/E-120 superantigen (Naptumomab estafenatox (nap))	None	Not restricted	IFN-α	II/III	metastatic clear cell or papillary RCC	Nap + IFN-α (N = 253) or IFN-α (N = 260)	Not applicable	Primary endpoint OS was not increased; Median OS 17.1 (Nap + IFN) vs 17.5 mo (IFN); HR 1.08 (P = 0.56)	105
MUC1	2 HLA class I peptides loaded on DC	Pan-DR class II peptide (PADRE)	HLA-A*0201	Low-dose IL-2	I/II	ccRCC with MUC1 expression	N = 20	Peptide reactive CTL responses in 12/20 (60%) patients	1 CR, 2 PR (15%)	85
	Recombinant MVA expressing MUC1 and IL-2 (TG4010)	None	Not restricted	None, IL-2, or IFN-α	I/II	mRCC	N = 37	6/23 (26%) CD8+ responses; 5/28 (18%) CD4+ responses	No objective responses	86
WT-1	1 HLA class I synthetic peptide	Montanide ISA-51	HLA-A*2402	IL2 or IL2 and IFN-α	I	RCC	N = 3	Peptide reactive CTL responses in 2/2 evaluated patients	No objective responses	87
	Autologous DCs loaded with 1 HLA class I synthetic peptide per patient; matched to HLA expression	OK-432	HLA-A*02:01, *02:06 or *24:02	None	I/II	Treatment refractory ccRCC or BC with WT1 expression	N = 5 RCC (plus 5 bladder cancer)	Peptide reactive CTL responses in 5/5 patients by tetramer staining, ELISPOT or cytoplasmic IFN-γ assays	No objective responses	88

Abbreviations: PR: partial responses, SD: stable disease, CR: complete response, PD: progressive disease, NED: no evidence of disease; OS: overall survival, SAE:serious adverse events, PADRE: pan-HLA-class II binding peptide

Table 3. Clinical testing of multiple antigen vaccines for advanced RCC

Immune Targets	Antigen Formulation	Adjuvant	HLA restriction	Combination therapy	Phase	Enrollment		Cellular Immune Response	Clinical Outcome	Reference
						Key Criteria	N
personalized peptide vaccine^1	4 HLA class I synthetic peptides per patient from a pool of 25 HLA-A24 and 23 HLA-A2 peptides	Montanide ISA-51VG	HLA-A*02 or *24	None	I	Cytokine-refractory mRCC	N = 10	Peptide reactive CTL responses in 2/10 (20%) patients	No objective responses	89
telomerase and survivin	8 telomerase + 11 survivin synthetic peptides (or tumor lysate for non HLA-A2 cohort) loaded on DC	Pan-DR class II peptide (PADRE)	HLA-A*02+ for peptide vaccine	Low-dose IL-2	I/II	Progressive mRCC	A2^+ peptide pulsed DC (n = 13) or A2- tumor lysate pulsed DC (n = 14)	CTL against 1 or more survivin/telomerase peptides in 6/6 A2^+ patients	No objective responses	90
multiple tumor- associated peptides^Citation2	9 HLA class I synthetic peptides plus 1 class II peptide. (Phase I included class I marker HBV peptide)	GM-CSF	HLA-A*02	None	I	mRCC	N = 28	20/27 (74%) patients developed CTL response to one or more peptide; 8/27 (22%) responded to more than one peptide; 14/27 (52%) responded to HBV	1 PR (4%)	91
	9 HLA class I synthetic peptides plus 1 class II peptide (IMA901)	GM-CSF	HLA-A*02	None or low-dose cyclophosphamide (Cy)	II	TKI or cytokine-refractory metastatic RCC	Cy + IMA901 (n = 33) or IMA901 (n = 35)	64% patients developed CTL response for one or more peptide; 26% patients responded to more than one antigen	1 CR, 2 PR (4%); investigator assessed	91
				Sunitinib	III (IMPRINT)	metastatic ccRCC; no prior systemic therapy	Sunitinib + IMA901 (n = 204) or sunitinib monotherapy (n = 135)	3-fold decrease of tumor-peptide specific CD8+ response compared with the phase I/II	Primary endpoint OS was not increased; Median OS 33.2 mo (IMA901+ Su) versus NR (Su); HR 1.34 (P = 0.087)	92
autologous tumor RNA-endcoded antigens	Autologous DCs co-electroporatedwith amplified tumor RNA and syntheticCD40L RNA (AGS-003, Rocapuldencel-T)	syntheticCD40L RNA	Not restricted	Sunitinib	II	metastatic ccRCC; no prior systemic therapy	N = 21	10 of 14 (71%) patients displayed an increasein CD28^+/CD45RA^− CTLsover baseline	9/21 PR (43%)	93
			Not restricted	Sunitinib	III (ADAPT)	metastatic ccRCC; no prior systemic therapy	AGS-003+ sunitinib (n = 307) or sunitinib (n = 155)	Immune responders (increased CD28+/CD45RA- CTL) after Rocapuldencel-T administration was 72% to 82%	Primary endpoint OS was not increased; Median OS 27.7 mo (AGS003+ Su) versus 32.4 (Su); HR 1.10	94

¹Antigen targets: SART1, SART2, SART3, MRP3, EZH2, HER2/neu, PTHrP, PSA and PAP, CEA, UBE and Lck. ²Antigen targets: (class I) PLIN2, APOL1, CCND1, GUCY1A3, PRUNE2, MET, MUC1, RGS5; (class II) MMP7. Abbreviations: PR: partial responses, SD: stable disease, CR: complete response, PD: progressive disease, NED: no evidence of disease; OS: overall survival, SAE: serious adverse events, PADRE: pan-HLA-class II binding peptide

CAIX

CAIX is one of the first characterized RCC-associated antigens ⁹⁵ and high-level expression is observed in ccRCC tumors at the earliest stage of the disease.^96,97 IHC staining revealed CAIX expression in over 85% of tumors but not normal kidney tissue.^95,98 CAIX epitopes have been identified to induce cytolytic T-cell responses from bulk T-cell populations restricted by HLA-A*24, HLA-A*02, and HLA-DR. In a phase I study, synthetic peptides corresponding to three putative HLA-A*24-restricted CAIX T-cell epitopes were used to vaccinate 23 mRCC patients refractory to cytokine therapy.⁷⁴ After 6–9 intradermal doses of the vaccine, cytolytic T-cell reactivity specific for one or more peptides was observed in 76% of the patients. Three patients had a partial response with a regression of pulmonary metastases. In a separate phase I study, synthetic peptides corresponding to naturally processed CAIX epitopes restricted by HLA-A*02⁹⁹ and HLA-DR¹⁰⁰ were used to develop a DC-based vaccine. The two CAIX-derived peptides and keyhole limpet hemocyanin (KLH) adjuvant were loaded on autologous DC’s from five patients with cytokine-refractory mRCC. After five intradermal vaccinations, all patients developed humoral responses against KLH, however, none mounted detectable CAIX-peptide-specific cellular immunity, and no clinical responses were observed.⁷⁵

CAIX has also been targeted with a first-generation chimeric antigen receptor (CAR) vector retrovirally transduced into autologous peripheral blood T-cells from 12 RCC patients refractory to prior systemic therapy with cytokines or TKIs.⁷⁶ Four of the first eight patients experienced liver toxicity associated with CAIX expression on bile duct epithelium and CAR-T-cell infiltration. In four additional patients, a CAIX-blocking antibody administered before CAR-T-cell infusions attenuated the liver toxicity. However, no clinical responses were observed.¹⁰¹

VEGR-R1

VEGFR1 is an important factor associated with RCC tumor angiogenesis. In a phase I vaccine trial, 18 patients were subcutaneously administered an HLA-A*0201- or HLA-A*2402-restricted VEGFR1-derived peptide weekly for 5 weeks, and then every 2 weeks.⁷⁷ Peptide-reactive CTL responses were observed in 15/18 patients and two patients showed a partial response.

HIG2

HIG2 is expressed in 86% of RCC tumors at an early stage of tumor development but not in normal kidney and functions as an autocrine factor enhancing tumor growth.¹⁰² High HIG2 expression was found to correlate with disease stage and is a poor prognostic marker for RCC patient survival. A phase I dose-escalation study deployed an HLA-A*0201/0206-restricted HIG2–9–4 peptide to vaccinate nine patients with refractory mRCC after cytokine and/or TKI therapies.⁷⁸ The vaccine was administered subcutaneously weekly for 4 weeks in each cycle and vaccination cycles continued until disease progression. HIG2–9–4-specific CTL responses were detected in eight of nine patients; however, there were no objective responses.

VHL

A pilot study identified patient-specific VHL mutation-spanning peptides with computationally predicted high binding affinity to an autologous class I HLA molecule.⁷⁹ Among six metastatic ccRCC patients, two had frameshift mutations, creating completely new 12-mer or 13-mer sequences; others harbored centered point mutations with 8-mer peptides flanking each side. These patients were treated by subcutaneous vaccination with their personalized synthetic peptide. Peptide-specific CD8⁺ T-cell responses were detected in 4/5 evaluable patients; however, no responses were observed in patients with measurable disease at study enrollment.

5T4/Trophoblast glycoprotein

Cancer-testis antigens have highly restricted expression in normal adults limited to male germ cells in the testis, and to ovary and trophoblast T-cells in females, but can be aberrantly expressed by many tumor types. However, the most studied immunogenic cancer-testis antigens NY-ESO-1, MAGE-A4 and SAGE were shown to have limited expression in RCC.¹⁰³ 5T4 is a glycosylated cancer-testis antigen that is a highly expressed in human trophoblast T-cells and is overexpressed in RCC and a wide range of other solid tumors including prostate, pancreatic, ovarian, breast, cervical, gastric, and non-small cell lung cancer but not normal adult tissue.^104–106 IHC staining has revealed over 95% of RCC tumors express at least focal 5T4 protein, and 75% of tumors have strong staining of cell surface 5T4. Expression is retained in metastatic tumors.¹⁰⁴ The observation of circulating preexisting CD8⁺ and CD4⁺ T-cell 5T4 responses in RCC patients^107,108 has encouraged the development of 5T4 targeted therapies. By IFN-γ ELISpot assays, to date 4 MHC class I-restricted 5T4 epitopes have been associated with common HLA alleles including HLA-A*0201, HLA-A*0101, and HLA-Cw7.^107,109,110

A Modified Vaccinia Ankara (MVA) virus was engineered to express the full-length 5T4 protein and shown to elicit 5T4-specific cellular or humoral responses in vaccinated RCC patients that correlated with clinical benefit in four phase I/II trials.^80–83,111 In the phase III TRIST trial, patients were randomized to MVA-5T4 (n = 365) or placebo (n = 368) in combination with sunitinib, IL-2, or IFN-α as first-line mRCC therapy.⁸⁴ Although well-tolerated, MVA-5T4 failed to show enhanced OS of vaccinated patients versus placebo (HR 1.07; P = 0.55). In post hoc analysis, there was an association between enhanced patient survival and 5T4 antibody response (56% of treated patients) but not MVA antibody responses (96% of treated patients) linking induced 5T4-specific immunity with better clinical outcomes. However, during early phase clinical development of MVA-5T4, a cellular immune response against 5T4 measured by IFN-γ ELIspot was detectable in <50% of patients. It is therefore likely that in the TRIST trial, only a minority of patients vaccinated with MVA-5T4 mounted 5T4-specific T-cell responses.^112,113

Naptumomab estafenatox (anti-5T4-SEA/E-120) is a fusion protein conjugating a bacterial superantigen variant to the Fab binding domain of a 5T4 monoclonal antibody¹¹⁴ in order to activate both CD4⁺ and CD8⁺ T-cells in proximity to 5T4-expressing tumor. In a phase II/III study, 512 patients with RCC were randomized to receive naptumomab estafenatox with IFN-α or IFN-α alone.^115,116 However, this study failed to meet its primary endpoint of improved survival for naptumomab estafenatox treated patients (HR 1.08; P = 0.56).

Mucin 1 (MUC1)

MUC1 is transmembrane glycoprotein restricted to the luminal surface of epithelial cells.¹¹⁷ In ccRCC, MUC1 is overexpressed, aberrantly glycosylated and diffused from the luminal surface, promoting cancer cell differentiation and metastasis.¹¹⁸ High MUC1 expression is an independent prognostic factor for advanced disease and metastasis in ccRCC.^117,119,120 MUC1 has a variable number tandem repeat (VNTR) of 20-amino-acids each with five potential sites of O-glycosylation.¹²¹ The hypoglycosylation of VNTR sequences in malignant T-cells can produce tumor-specific glycopeptide antigens. HLA-A*02-restricted T-cell epitopes from the VNTR core were discovered,^122,123 and MUC1-specific CTL clones from HLA-A*02⁺ healthy donors were isolated that recognized a variety of cancer cell lines including renal tumor lines.^122,124,125 In addition, a non-MHC-restricted recognition of tumoral MUC1 epitopes by CD8⁺ T-cells was documented in different tumor types.¹²⁶

In a phase I trial, 20 HLA-A2⁺ metastatic RCC patients with MUC1 expressing tumors were vaccinated with autologous DCs pulsed with HLA-A2 binding MUC1 peptides plus the pan-DR helper peptide PADRE.⁸⁵ Twelve of 18 evaluable patients developed detectable MUC1 peptide CTL responses and 10/18 had PADRE-specific responses. Three objective responses were observed including one complete response.

TG-4010 is an MVA-based vaccine that is designed to express both IL-2 and MUC1 protein. The safety profile was confirmed in several phase I studies. MUC1-specific T-cell and antibody responses were seen in some pancreatic cancer patients,¹²⁷ while in another phase I study, only T-cell responses to MUC1 were observed with various-advanced cancers.¹²⁸ The best clinical responses were observed in non-small cell lung cancer.¹²⁹ A phase II study enrolled 37 mRCC patients and reported 5/28 evaluable patients developed a MUC1-specific CD4⁺ T-cell response during therapy and 6/23 had MUC1-specific CD8⁺ T-cells detected before or during therapy. However, there were no objective responses.⁸⁶

Wilms’ tumor 1 (WT-1)

WT1 is expressed in normal gonad, uterus, kidney, mesothelium, and hematopoietic progenitor cells and is frequently overexpressed in RCC.¹³⁰ In a 2007 phase I study, two patients with RCC were treated with an HLA-A*2402-binding WT1 peptide with a modified anchor residue.⁸⁷ Peptide-specific T-cells were detected from PBMCs in both patients; however, neither experienced an objective response. In another phase I study, five patients with metastatic or relapsed RCC expressing WT1 were treated with HLA-matched WT1 peptide-loaded DCs and OK-432 adjuvant,⁸⁸ a toll-like receptor 4 ligand shown to facilitate the maturation of DCs and stimulate Th1-type cytokine secretion.¹³¹ Vaccinations were given in combination with a TKI or mTOR inhibitor. Peptide-reactive CTL responses were detected in all five patients by tetramer staining, ELISPOT, or cytoplasmic IFN-γ assays. However, no objective responses were observed.

Multi-antigen vaccines (Table 3)

In contrast to vaccines targeting single antigens, multi-antigen vaccine platforms are anticipated to increase the likelihood for antigen priming and vaccine-induced T-cell responses. A multi-valent immune response may also reduce the potential for tumor escape from immune surveillance.

Personalized peptide vaccines

In 2007, an RCC vaccination protocol enrolled 10 patients with refractory metastatic disease in which four HLA class I restricted synthetic peptides per patient were selected from a pool of 25 HLA-A*24-restricted and 23 HLA-A*02 peptides, based on the presence of preexisting peptide-specific CTLs in PBMC and IgG in the plasma of RCC patients.⁸⁹ Peptide-reactive CTL responses were detected in only 2/10 patients post-treatment, peptide-specific IFN-γ production in post-vaccination PBMC was only minimally increased, and no objective responses were observed.

Survivin and telomerase

Survivin and telomerase are overexpressed in the majority of RCC tumors. Survivin is an oncofetal protein that inhibits apoptosis and regulates cell division, while telomerase regulates telomere shortening during DNA replication and is activated in a majority of human cancers. Histological analysis of 634 ccRCC tumors demonstrated that survivin was expressed in all tumors, with 31.2% tumors having high expression (> 15 cells/mm²).¹³² A meta-analysis of 12 independent studies showed that increased survivin expression predicted poor prognosis in RCC patients.¹³³ In a phase I/II multi-peptide vaccination trial, eight telomerase- and 11 survivin-derived HLA-A*02-restricted peptides (or tumor lysate for the non-HLA-A*02 cohort) were loaded on autologous DCs to vaccinate 13 HLA-A*02⁺ patients.⁹⁰ In 14 non-HLA-A*02⁺ patients, autologous DCs were pulsed with tumor lysate prior to administration, in hope to present tumor antigen to CTLs. Cytotoxic T-cell responses against one or more survivin/telomerase peptides were found in 6/6 evaluated HLA-A*02⁺ patients. No objective responses were observed.

IMA-901

IMA-901 is a vaccine developed from multiple tumor-associated peptides that are naturally presented in human RCC tissue.¹³⁴ The discovery workflow included mass spectrometry sequencing of HLA-associated peptides eluted from primary RCC tumors, mRNA expression profiling of the genes encoding HLA-bound peptides to select those preferentially expressed in tumor versus healthy tissues, and assessment of peptide immunogenicity by in vitro priming of peptide-specific T-cells in PBMC of healthy donors. Nine HLA-A*02-presented peptides and one HLA-DR-restricted peptide from highly overexpressed genes in tumor (PLIN2, APOL1, CCND1, GUCY1A3, PRUNE2, MET, MUC1, RGS5, MMP7) were included in the vaccine, and granulocyte macrophage-colony stimulating factor (GM-CSF) was used as a local immune adjuvant.⁹¹ In a phase I trial, 20/27 patients developed a T-cell response to at least one peptide antigen, and 8/27 patients responded to more than one peptide.¹³⁵ A phase II trial in 68 patients demonstrated similar immune response efficacy where 64% of patients treated with IMA901 plus GM-CSF with or without a single infusion of cyclophosphamide-developed T-cell responses, and 26% responded to more than one peptide.⁹¹ Better OS was associated with T-cell responses against multiple peptides. In the subsequent phase III IMPRINT study of HLA-A*02⁺ patients randomized to IMA-901 (n = 204) plus sunitinib versus sunitinib monotherapy (n = 135),⁹² no improvement in OS was observed for the vaccinated patients (HR 1.34, P = 0.087). Notably, the magnitude of T-cell responses was threefold lower compared with the preceding phase II study, and there was no clear association between T-cell responses and clinical outcomes.

AGS-003

In a phase I study, renal tumor RNA-transfected DCs were administered to 10 patients. The vaccine successfully induced T-cell responses against a broad set of tumor antigens including hTERT, CAIX, and OFA, but not against antigens expressed by autologous normal renal tissue.¹³⁶ AGS-003 (Rocapuldencel-T) is a DC-based vaccine in which autologous mature DCs are electroporated with tumor lysate-derived mRNA and synthetic CD40L RNA in order to present patient-specific tumor antigens (mutated and non-mutated; class I and class II), and to activate co-stimulatory signals in T-cells.¹³⁷ Following a promising result in a phase II study of 21 patients (median OS of 30.1 months),⁹³ the phase III ADAPT study was performed in which 462 mRCC patients undergoing cytoreductive nephrectomy were randomly assigned to AGS-003 plus sunitinib versus sunitinib monotherapy.⁹⁴ By intent to treat analysis, there was no significant OS benefit for combination therapy (HR 1.10). Of interest, for the 70% of the vaccine-treated patients with a positive T-cell activation biomarker assay, the magnitude of the T-cell response positively correlated with OS.

Discussion and future directions

Extensive clinical development of vaccine and other antigen targeting strategies for RCC has culminated with four completed phase III clinical trials. However, a successful primary endpoint for better survival in patients who received the investigational therapies has not yet been demonstrated, and there are currently no FDA-approved vaccine or other tumor antigen-specific compounds for advanced RCC.

Multiple early peptide vaccine studies showed encouraging rates for T-cell priming against the target antigen. However, conventional immune monitoring post-peptide-based vaccination may overestimate the frequency of true tumor-reactive T-cells in tumor by detecting a T-cell pool with a wide range of antigen affinity/avidity in periphery. T-cells with low avidity respond to peptide-loaded targets but may not recognize lower antigen density on tumor.⁸⁵ Multi-peptide vaccines therefore were developed to increase the frequency of T-cell responses and also decrease the potential for tumor escape from immune surveillance. Despite deploying a panel of ten synthetic tumor antigen peptides, the phase III IMPRINT study failed to show improved survival in the vaccine cohort associated with immune response data significantly inferior to prior phase I and II data with the same compound.⁹²

Compared to synthetic peptides, recombinant vaccinia vaccines that require T-cell recognition of naturally processed antigen may stimulate T-cells with higher avidity and more robust anti-tumor activity. Such constructs also are anticipated to prime both antigen-specific CD8⁺ and CD4⁺ T-cells and therefore may elicit better helper function than synthetic peptide vaccines stimulating only CD8⁺ T-cell responses. Nevertheless, MVA-based vaccines incorporating 5T4 or MUC1 achieved detectable antigen-specific T-cell priming in < 50% of vaccinated patients. Rapid development of serologic immunity to the MVA vector likely limited the ability of repeat vaccination to drive sustained T-cell responses ⁸⁴ and contributed to the failure of MVA-5T4 to show a survival benefit in the phase III TRIST trial. A prime-boost protocol with sequential heterologous recombinant 5T4 expression vectors (ChAdOx1-MVA-5T4) has yielded considerably higher specific T-cells responses¹³⁸ in recent phase I testing in prostate cancer.¹³⁹

The naptumomab estafenatox fusion protein in principle could overcome the non-responder phenotype observed in vaccine trials by delivering antigen-directed T-cell engagement to all treated patients. However, the bacterial derived superantigen component is immunogenic, and like the MVA reagents, serologic immunity to the bacterial protein also appeared to diminish the activity.^115,116 While the primary analysis of the phase II/III trial with naptumomab estafenatox was negative for improved survival versus the IFN-γ control arm, the patient subgroup having below median of baseline anti-SEA/E-120 showed a trend toward improved OS and PFS.

The tumor RNA-transfected DC platform for the AGS-003 vaccine addressed several limitations inherent in the other vaccine and naptumomab estafenatox therapies. Antigen targets expressed in AGS-003 are naturally processed proteins that are expected to prime high avidity T-cell responses. Antigens are personalized based on autologous tumor RNA and can include both mutated and nonmutated antigens without requiring the laborious effort needed to identify patient-specific antigen targets. AGS-003 is also fully autologous without foreign elements and should be suitable for serial administration and T-cell boosting without eliciting vector-specific immunity. Despite these advantages, the phase III ADAPT trial failed to show a survival benefit for patients receiving AGS-003. One limitation of individualized antigen priming is the inability to conduct quantitative antigen-specific immune monitoring, leaving substantial uncertainty about the magnitude and durability of T-cell priming against tumor antigens with this approach.

It is noteworthy that all of the phase III vaccine studies were conducted during the cytokine or targeted therapy era of RCC, and it has been suggested that the combining partner for the tested vaccines may have contributed to the lack of efficacy. For example, reduced monocyte counts in the IMPRINT study has been associated with sunitinib and suggested to contribute to the poor immune response outcomes noted.⁹² Therapeutic combinations of tumor vaccine and immune checkpoint blockade may produce better synergy to activate and maintain effective anti-tumor T-cell immunity.¹³⁸ Currently active studies include the personalized neoantigen peptide vaccine NeoMax in combination with ipilimumab at the vaccine injection site to direct anti-CTLA4 activity to the vaccine-draining lymph nodes (NCT02950766), the WT1 multi-peptide vaccine DSP-7888 administered in combination with anti-PD-1 antibodies nivolumab or pembrolizumab (NCT03311334), and an mRNA-based personalized neoantigen multi-epitope vaccine in combination with the anti-PD-L1 antibody atezolizumab (NCT03289962).

The impressive success of engineered T-cells expressing chimeric antigen receptors (CARs) to achieve complete remissions of refractory acute lymphocytic leukemia and non-Hodgkin lymphoma^140,141 has created intense interest to extend engineered, re-directed T-cells as a therapeutic modality to treat other cancers including solid tumors. Promising early results with T-cell receptor (TCR) engineered T-cell therapy targeting NY-ESO-1, MAGE-A4 and human papillomavirus proteins for melanomas, sarcomas, and cervical cancers encourage further development of tumor antigen-specific TCR in addition to CAR vectors.^142–144 Adoptive T-cell therapy appears capable of generating far higher numbers of anti-tumor effector T-cells in treated patients than can be generated with available vaccine technologies and provides an anti-tumor product to all treated patients in contrast to vaccines that suffer from high frequencies of non-responders. However, the initial experience in RCC patients with CAR-T-cells targeting CAIX associated with off-tumor toxicity without tumor response reinforces the need for both careful target selection and also strategies to augment the anti-tumor potency.⁷⁶ Cellular therapies lend themselves to further optimizations including the targeting receptor gene vector (TCR affinity and expression enhancement,¹⁴⁵ CAR signaling domain optimization¹⁴⁶), selecting the phenotype and effector potential of transduced T-cells (CD4⁺, CD8^+,147 central memory T-cell¹⁴⁸), and further genetic manipulations of the transduced cell, (CRISPR/cas9 targeted disruption of the native TCR locus,¹⁴⁹ or inhibitory co-receptors¹⁵⁰). Safe and effective cellular products may subsequently be combined with synergistic non-specific immunotherapies including checkpoint blocking antibodies or T-cell agonist cytokines. Emerging phase I and II trials enrolling RCC patients are actively testing engineered T-cells with CAR vectors targeting CD70, or c-MET in addition to TCR’s specific for personalized neoantigens or an HLA-A*11 presented hERV-E epitope (Table 4).

Table 4. Active trials with engineered T-cell therapy enrolling metastatic RCC patients

Antigen target	Phase	Specific RCC Setting	Engineered T-cells	Combination	Clinicaltrials.gov identifier
CD70	I/II	CD70+ ccRCC	autologous CAR-T	fludarabine, cyclophosphamide, high-dose IL-2	NCT02830724
CD70	I	ccRCC	allogeneic CAR-T-cells with CRISPR-Cas9 targeting of MHC I and endogenous TCR (CTX130)	lymphodepleting chemotherapy	NCT04438083
c-MET	I/II	c-MET+ RCC	autologous CAR-T	fludarabine, cyclophosphamide	NCT03638206
personalized neoantigen	I	individual neoantigen target(s)	autologous TCR-transduced	fludarabine, cyclophosphamide, high-dose IL-2	NCT04596033
hERV-E	I	ccRCC, HLA-A*11:01+	autologous TCR-transduced CD8+/CD34+	fludarabine, cyclophosphamide, moderate-dose IL-2	NCT03354390

Abbreviations: RCC – renal cell carcinoma; IL-2 – interleukin 2; CAR – chimeric antigen receptor; HLA – human leulocyte antigen; TCR – T-cell receptor

5T4 is the most extensively studied RCC antigen and has been targeted in phase III trials by vaccination with MVA-5T4 and the antibody-superantigen conjugate naptumomab estafenatox without detection of on-target, off-tumor toxicities. Our group has sequenced the TCRs from seven high avidity CTL clones recognizing a defined HLA-A2 presented 5T4 epitope.¹⁵¹ CD8⁺ T-cells from healthy donors transduced with these 5T4-specific TCRs recognized 5T4-expressing tumor lines and primary RCC tumors, but not normal renal tubular epithelial cells. Clinical testing of TCR engineered T-cells redirected to target 5T4 would closely parallel the emerging success with engineered T-cells targeting single cancer-testis class antigens NY-ESO-1 and MAGE-A4 in other solid tumors.

Lastly, discovery of novel tumor antigen targets in RCC is a research priority, with highest interest in the antigen targets associated with sustained and deep tumor regression seen in some patients receiving checkpoint blocking immunotherapies. However, a major technical challenge is the identification of the cognate antigen for defined TCR sequences. T-cell antigen discovery may be accelerated by the recent developments in high-throughput methods for identifying TCR-epitope pairs.^152–154 Our group and others are applying whole-genome CRISPR-Cas9 screening to facilitate the discovery of tumor antigens targeted by RCC TIL.^155,156 Emerging computational approaches that consider information such as TCR sequence similarity, structural information, V gene usage bias, CDR3 length, and HLA types are also in development to predict peptide antigen sequences for select TCRs. Success in these efforts may serve to prioritize among known RCC tumor antigens those with the highest clinical value for therapeutic targeting.

Disclosure of potential conflicts of interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Additional information

Funding

This work was supported by the Cancer Research Institute, Clinic and Laboratory Integration Program, the Cancer Therapeutics Endowment, and philanthropic gifts made to the Fred Hutchinson Cancer Research Center.