Banana-shaped survival curves of metastatic renal cell carcinoma treated with first-line immune-combinations, not just a matter of “palateau”

ABSTRACT

The first-line therapy of metastatic renal cell carcinoma (mRCC) has revolutionized with the approval of immune checkpoint inhibitors (ICIs) in combination with or without tyrosine kinase inhibitors (TKIs). The choice among the many different immuno-combinations (ICI-ICI or ICI-TKI) is challenging due to the lack of predictive factors. The different shapes of the Kaplan–Meier survival curves (e.g. “banana-shaped curves”) have raised many questions on the long-term survival benefit. Here, we analyzed the factors that could have impacted the different long-term survival, including the prognostic factors distribution (IMDC score), histological factors (sarcomatoid features, PD-L1 expression), and treatment characteristics (mechanism of action, duration, discontinuation rate). This overview highlights the factors that should be considered in the first-line setting for the patients’ therapeutic choice and prognostic assessment. They are also fundamental parameters to examined for head-to-head studies and real-life, large-scale studies.

KEYWORDS:

• Renal cell carcinoma
• immune checkpoint inhibitor
• immuno-combination
• immunotherapy
• survival outcomes

Introduction

In the last years, the first-line treatment of metastatic renal cell carcinoma (mRCC) has deeply changed with the approval of immune checkpoint inhibitors (ICIs) in combination with or without tyrosine kinase inhibitors (TKIs).^1,2

The first-line combinations have shown an advantage in terms of overall survival (OS) (range HR: 0.70–0.84), progression-free survival (PFS) (range HR: 0.47–0.86) and overall response rate (ORR) (range: 39.3–71.3% vs 28–40%)^1,3 compared with sunitinib.

With different immuno-combinations (ICI-ICI or ICI-TKI) available and lack of predictive factors that may guide the clinical decisions, the choice of the first-line strategy remains challenging.³ The two types of immuno-combinations differentiate, other than the toxicity profile, for the derived clinical outcomes in terms of ORR and PFS that generally favor the ICI-TKI combination.^1,4,5 However, the survival updates reported at ASCO 2023 have raised many questions on the long-term survival benefit of the main first-line immuno-combinations approved in Europe.^6–8 These mainly arise from the different shapes of the Kaplan–Meier survival curves (e.g. “banana-shaped curves”) for Keynote-426 (KN-426) and CLEAR studies compared with those observed in the CheckMate-214 (CM-214) and CM-9ER, wondering which factors are responsible for this different long-term impact.^6,7,9,10

In this context, we tried to answer these questions, and this commentary is an attempt to explain the different shapes of the survival curves of these clinical trials. In our opinion, in fact, the reasons that could explain the different survival curves lie in the different distribution of prognostic factors, histologic factors (i.e. sarcomatoid component) and PD-L1 tumor expression; the mechanism of action of the drugs with their interplay with immune system and risk of resistance; and the treatment duration and the subsequent therapeutic lines.

Prognostic factors distribution

The clinical trials of the first-line immuno-combinations are characterized by a different prognostic distribution of IMDC good- and poor-risk patients enrolled (Table 1). In fact, it is well known that IMDC good-risk patients do not benefit from a immuno-combination compared to a TKI monotherapy in terms of OS.^11,12 Conversely, IMDC poor-risk patients are the prognostic group that benefit most from immunotherapy and less from TKI.^5,13,14

Table 1. Prognostic distribution of IMDC groups in first-line mRCC clinical trials.

	(N, %) HR OS (95% CI)
	Nivolumab + Ipilimumab^Citation9 CheckMate 214	Pembrolizumab + Lenvatinib^Citation6 CLEAR	Nivolumab + Cabozantinib^Citation8 CheckMate 9ER	Pembrolizumab + Axitinib^Citation7,Citation10 Keynote 426
ITT population	N = 550	N = 355	N = 323	N = 432
	0.72 (0.62–0.85)^*	0.79 (0.63–0.99)^*	0.70 (0.56–0.87)^*	0.84 (0.71–0.99)^*
mFU (mo)	67.7	49.8	44	67.2
IMDC group
Favorable	125 (23%)	110 (31%)	74 (23%)	138 (32%)
	0.94 (0.65–1.37)	0.94 (0.58–1.52)	1.07 (0.63–1.79)	1.10 (0.79–1.54)
Intermediate	334 (61%)	210 (59%)	188 (58%)	238 (55%)
	0.74 (0.61–0.91)^*	0.85 (0.63–1.13)	0.75 (0.56–1.00)	0.63 (0.48–0.83)^*
Poor	91 (16%)	33 (9%)	61 (19%)	56 (13%)
	0.58 (0.42–0.80)^*	0.47 (0.25–0.87)^*	0.46 (0.30–0.72)^*	0.59 (0.36–0.96)^*

N number of patients, IMDC International Metastatic RCC Database Consortium, mo months, HR hazard ratio, CI confidence intervals.^*HR with statistical significance

This different treatment benefit according to the IMDC score might rely on the distinct gene expression related to angiogenesis: favorable-risk patients are characterized by a higher angiogenesis gene expression, while poor-risk patients are characterized by a higher immune gene expression.^15,16

As shown in Table 1, the CLEAR and the KN-426 trials are the studies of first-line immuno-combinations with the higher percentage of IMDC favorable-risk patients (31–32%) and the lower percentage of IMDC poor-risk patients (9–13%), which may partially account for the absence of long-term survival benefit in the intention-to-treat population. The latter is clearly observed by the recently published “banana-shaped” survival curves and the greater worsening in the HRs for OS with the updating of the follow-up.^6,7 Conversely, in the CM-214 and CM-9ER trials, there is the lower percentage of favorable-risk patients (both 23%) and the higher percentage of poor-risk ones (16–19%) (Table 1).^8–10 In these trials, the survival curves remain separated in the long term, and the HRs for OS showed the least worsening.

Sarcomatoid component

Another clinical factor that may influence the different survival performance of the immuno-combinations is the varying presence of patients with sarcomatoid features (sarcomatoid patients), which is a well-known negative prognostic factor but a positive predictive factor of immunotherapy¹⁷ (Table 2). The immuno-combination Pembrolizumab + Lenvatinib presented the lowest percentage of sarcomatoid patients (8%) and the worst survival correlation in this patients’ population (HR for OS 0.91).¹⁸ Conversely, the intermediate/poor-risk subgroup of the CM-214 study has the highest percentage of sarcomatoid patients (17%), which may be considered as another factor of the long-term benefit with an impressive plateau in the curves’ tales,¹⁹ even though the highest survival benefit seems to be reported with the Nivolulamb + Cabozantib combination (HR 0.36).²⁰ It remains, however, difficult to interpret the impact of sarcomatoid patients on long-term survival in the Pembrolizumab + Axitinib combination, as the percentage is similar to that reported in the CM 9ER study, but the survival impact is similar to that observed within the CM-214 study.²¹

Table 2. Distribution of sarcomatoid features and PD-L1 expression in patients treated with first-line immune-combination in key mRCC clinical trials.

	(N, %) HR OS (95% CI)
	Nivolumab + Ipilimumab^Citation19,Citation22 CheckMate 214 (ITT population)	Nivolumab + Ipilimumab^Citation19,Citation22 CheckMate 214 (Intermediate/Poor risk)	Pembrolizumab + Lenvatinib^Citation6,Citation18 CLEAR	Nivolumab + Cabozantinib^Citation20 CheckMate 9ER	Pembrolizumab + Axitinib^Citation21 Keynote 426
ITT population	N = 550	N = 425	N = 355	N = 323	N = 432
	0.72 (0.62–0.85)	0.68 (0.58–0.81)	0.66 (0.49–0.88)	0.70 (0.55–0.90)	0.53 (0.38–0.74)
mFU (mo)	67.2	67.2	26.6	32.9	30.6
Sarcomatoid features (%)	13%	17% 0.46 (0.29–0.71)^*	8% 0.91 (0.32–2.58)	11% 0.38 (0.20–0.74)^*	12% 0.58 (0.21–1.59)
PD-L1 positive (≥1%)	23% 0.57 (0.40–0.82)^*,1,a	25% 0.45 (0.29–0.71)	30% 0.84 (0.57–1.23)^2,b	25% 0.74 (0.48–1.16)^1,a	59% 0.54 (0.35–0.84)^*,2,b

N number of patients, IMDC International Metastatic RCC Database Consortium, mo months, HR hazard ratio, CI confidence interval.

¹Dako PD-L1 IHC 28–8 pharmDx test, ²PD-L1 IHC 22C3 pharmDx assay.

^aTumor cells; ^bCombined positive score^*HR with statistical significance

PD-L1 expression

As per the sarcomatoid features, PD-L1 expression is also a negative prognostic and potential positive predictive factor for immuno-combinations, although it is not currently used in clinical practice.²³ According to the sub-analyses on the PD-L1 expression of the phase III trials, PD-L1-positive patients derived greater benefit from Nivolumab + Ipilimumab, particularly the intermediate/poor risk patients,^19,22 while no significant benefit was observed with Pembrolizumab + Lenvatinib and Nivolulamb + Cabozantinib combinations despite the similar percentage of PD-L1-positive patients (23–30%) (Table 2).^6,18,20 Differently from the other two ICI-TKI combinations, Pembrolizumab + Axitinib was associated with a survival benefit over sunitinib in PD-L1-positive patients, who were highly represented (59%)²¹ (Table 2).

Treatment mechanism of action

According to the treatment characteristics, the long-term benefit (i.e. the tails of the curves) seen in the CM-214 study might be explained by the way ipilimumab triggers the immune system through the induction of immunological memory, especially the memory T cell populations.²⁴ Ipilimumab is a monoclonal antibody that blocks the cytotoxic T lymphocyte-associated antigen 4 (CTLA), which is a negative regulator of T-cell-mediated immune responses, including memory and effector T cells.

Also TKIs, such as Cabozantinib and Axitinib, play an immunomodulatory role in the tumor microenvironment, particularly when combined with immunotherapy.²⁵ However, considering the different long-term impact, TKIs may have a different impact on the induction of the immunological memory compared to ipilimumab. The latter may be one of the possible biological rationales, together with a high percentage of poor-risk patients, to explain the benefit from immunotherapy.

Nivolumab and Pembrolizumab are both monoclonal antibodies that block the Programmed Death-1 (PD-1), which is a cell surface receptor on T cells and B cells playing a role in down-regulating the immune response.²⁶ However, they differ in terms of mechanisms of action and pharmacokinetic properties, and, among them, the different antigen-binding epitopes and the binding affinity (higher with nivolumab than with pembrolizumab) may play a role, although their clinical impact is unknown.²⁶

Treatment duration and discontinuation rate

The different immunotherapy durations, and subsequently the immune system stimulation, may also influence the different prognostic performance of the immuno-combinations (Table 3). The CM-214 is the only study in which immunotherapy treatment was administered until progression of disease or clinical benefit (i.e. also after 2 y), while the other studies allowed a maximum immunotherapy duration of about 2 y⁹ (Table 3). However, a recent meta-analysis and sub-analyses conducted in patients who have completed immunotherapy treatment showed that these patients derive long-term benefit despite immunotherapy discontinuation (KN-426, CM-9ER) (Table 3).^7,8,27,28 Therefore, it is not completely understood how immunotherapy discontinuation due to treatment completion could impact on long-term survival.

Table 3. Treatment characteristics of first-line mRCC clinical trials.

	Nivolumab + Ipilimumab^Citation9 CheckMate 214 (ITT population)	Nivolumab + Ipilimumab^Citation9 CheckMate 214 (Intermediate/ Poor risk)	Pembrolizumab + Lenvatinib^Citation6,Citation18,Citation29 CLEAR	Nivolumab + Cabozantinib^Citation8,Citation20 CheckMate 9ER	Pembrolizumab + Axitinib^Citation7 Keynote 426
N	550	425	355	323	432
mFU (mo)	67.7	67.7	49.8	44.0	67.2
HR OS unadjusted (95% CI)	0.72 (0.62–0.85)	0.68 (0.58–0.81)	0.79 (0.63–0.99)	0.70 (0.56–0.87)	0.84 (0.71–0.99)
Treatment characteristic
IT duration	2 y (optional)	2 y (optional)	35 cycles (maximum)	2 y (maximum)	35 cycles (maximum)
Immunotherapy completion (N, %) (median, 95% CI)	NA	NA	101 (28%)^b 3y–OS 94.5%^b	115 (36%) mTST 20.6 (7.9–NE)	120 (28%) mOS NR (70.6–NR)
Duration of therapy (median, range)	7.9 (2.1–21.8)	NA	17.0 (0.1–39.1)	21.8 (8.8–34.0)	10.4 (0.03–21.2)^c
Treatment discontinuation (%)	93%	NA	67%	82%	89%
Treatment reduction (%)	–	–	71%^a	61%	20%^c
Progressive disease as best response (%)	17.6%	19.3%	5%	6.5%	11.3%
Subsequent systemic therapy (%)	55%	53%	51%	22%	62%
1^st subsequent systemic therapy	Sunitinib	Sunitinib	Cabozantinib	Axitinib	Cabozantinib
HR OS adjusted by subsequent therapy (95% CI)	–	–	0.55 (0.44–0.69)	–	0.67 (0.52–0.84)
Control arm characteristics
mOS (months)	38.4	26.6	54.3	34.3	40.8
Subsequent immunotherapy (%)	45.6%	40.5%	54.6%	35.1%	80.0%

N number of patients, IMDC International Metastatic RCC Database Consortium, mo months, HR hazard ratio, CI confidence interval, mTST median time to subsequent therapy, mOS median overall survival, NR not reached, NE not evaluable.

a:³⁰, b:²⁷, c:²¹

In this context, the treatment duration, discontinuation and reduction are important factors to be considered (Table 3). However, while Nivolumab + Ipilimumab were administered with no reductions, they were associated with the shortest treatment duration and the highest percentage of treatment discontinuation compared to ICI-TKI combinations.

Performance of the control arm

Looking at the control arm is also fundamental to interpret the survival benefit of an experimental treatment. Interestingly, the banana-shaped survival curves of the pembrolizumab-based combinations could also be explained by the higher survival performance of the sunitinib arm compared to the other trials (40.8–54.3 months vs 26.6–38.4 months) (Table 3). In addition, it is also necessary to see how many patients received a subsequent immunotherapy line and, as expected, the higher survival performance in the sunitinib arm was also due to the higher percentage of patients who received subsequent PD-(L)1 inhibitor therapy (Table 3).

Impact of nephrectomy

The absence of the primary tumor is a well-known prognostic factor in mRCC patients,³¹ but, in the context of the first-line setting, it is not simple to differently assess the prognostic impact of prior nephrectomy of the localized disease and that one of the cytoreductive nephrectomy of the advanced disease. All the first-line clinical trials have reported a survival benefit of the immuno-combination over sunitinib in both patients who received and not received nephrectomy, also in advanced patients who received cytoreductive nephrectomy prior to a ICI-based combination.³² No statistical comparison is available on the survival impact of the nephrectomy within each treatment arm, although the mOS is numerically higher in patients with prior nephrectomy in both treatment arms (Table 4). This is less evident in the CheckMate 9ER where the percentage of patients treated with the experimental arm and prior nephrectomy is lower than that one reported in the other clinical trials (68.7% vs 73.8–83.1%) (Table 4). Therefore, currently, the role of nephrectomy on the different survival curve shape of the first-line trials is unknown.

Table 4. Impact of nephrectomy in first-line mRCC clinical trials.

	Nivolumab + Ipilimumab^Citation9 CheckMate 214 (ITT population)	Nivolumab + Ipilimumab^Citation9 CheckMate 214 (Intermediate/ Poor risk)	Pembrolizumab + Lenvatinib^Citation18 CLEAR	Nivolumab + Cabozantinib^Citation20 CheckMate 9ER	Pembrolizumab + Axitinib^Citation33 Keynote 426
mFU (mo)	67.7	–	26.6	32.9	30.6
Immuno–combination arm vs Sunitinb arm
N	550 vs 546	–	355 vs 357	323 vs 328	432 vs 429
mOS (mo)	55.7 vs 38.4	–	NE vs NE	37.7 vs 34.3	NR vs 35.7
Immuno–combination arm
Prior nephrectomy
N (%)	455 (82.7%)	–	262 (73.8%)	222 (68.7%)	359 (83.1%)
mOS (mo)	59.7	–	NE	39.4	NR
No prior nephrectomy
N (%)	95 (17.3%)	–	93 (26.2%)	101 (31.3%)	73 (16.9%)
mOS (mo)	31.9	–	33.1	31.1	30.6
Sunitinib arm
Prior nephrectomy
N (%)	439 (80.4%)	–	275 (77%)	233 (71%)	359 (83.7%)
mOS (mo)	45	–	NE	36.2	NR
No prior nephrectomy
N (%)	107 (19.6%)	–	82 (23%)	95 (29%)	70 (16.3%)
mOS (mo)	15.5	–	24.0	27.1	19.3

N number of patients, IMDC International Metastatic RCC Database Consortium, mo months, mOS median overall survival, NE not evaluable, NR not reached.

Impact of second-line treatments

Most importantly, the topic of progression and subsequent treatment lines is one of the most discussed factors that may have influenced long-term survival, especially for those studies in which a clear advantage is maintained in PFS but not in OS (KN-426 and CLEAR studies). In this context, it is worth considering how many patients experienced progressive disease, how many patients received a subsequent treatment line, the therapeutic resistance and subsequent choices.

From a biological point of view, the progression on an ICI-TKI combination arises from resistance from two mechanisms of action, subsequently reducing the therapeutic possibilities and responses in the further lines. It is clear that after progression on an ICI-ICI combination, the possibility of inhibiting the kinase pathways increases the possibilities of treatment choice and response. This has been shown by the different activity of Cabozantinib in mRCC patients progressing on first-line ICI-based therapies (CaboPoint trial – NCT03945773), showing a slightly higher overall response rate in patients treated with ICI-ICI combination compared to those treated with ICI-TKI combination (Pembrolizumab + Axitinib, Avelumab + Axitinib).³⁴ In the retrospective study conducted by Marteau et al. (NCT04353765) on mRCC patients treated with ICI, Cabozantinib was associated with a significantly higher ORR and time to treatment discontinuation compared to other TKIs (Axitinib, Lenvatinib, Pazopanib, Sorafenib, Sunitinib).³⁵ The activity of Cabozantinib after ICI-based therapy was also observed in the retrospective analysis conducted by McGregor et al.³⁶

Therefore, in the context of mechanisms of resistance, also the spectrum of target activity of TKIs used in the first-line treatment is fundamental to choose subsequent lines. In particular, Cabozantinib and Lenvatinib are multi-kinase inhibitors, and, particularly, the TKIs with the wider range of target inhibition, including VEGFR1–3, PDGFR, MET, RET and AXL, while Axitinib is a potent and selective VEGFR inhibitor (VEGFR1–3).³⁷

Actually, patients receiving first-line treatment not containing Cabozantinib have the chance to receive Cabozantinib as a further treatment line, which is one of the most effective TKI.³⁸ On the other hand, patients treated in the CM-9ER trial mainly received Axitinib and Sunitinib as subsequent treatment lines, which have a narrower spectrum of kinase inhibition.

However, this biological and practical view should be considered in the context of how many patients actually progressed and received subsequent systemic therapies.

The CM-214 study had the highest percentage of progressive disease as the best response with a possible higher impact of subsequent therapies on OS compared to ICI-TKI-treated patients.⁹ On the other hand, the CLEAR and CM-9ER trials had a lower percentage of patients with ‘disease progression’ as the best response.

In consideration of the important impact of subsequent therapies, survival adjusted by subsequent therapy has been performed in the KN-426 and CLEAR studies reporting a higher theoretical (or “artificial”) long-term survival benefit of the immuno-combination over TKI (KN-426: HR 0.67 vs 0.84; CLEAR: HR 0.55 vs 0.79), showing a different shape of the Kapla–Meier curves.^6,7

Conclusions

From this overview of possible patient and treatment factors that may have influenced the different long-term survival performance of first-line immuno-combinations, it can be inferred that many factors should be considered when choosing first and subsequent lines, and many questions are still unanswered.

On this basis, head-to-head studies on first-line immuno-combinations are highly warranted. In their nascence, real-world studies on a large scale, such as the Meet-URO 33, ARON-1 and CARE-1 RLT studies,^38–40 could be of paramount importance in answering many of these clinically relevant questions.

Author contributions

All authors have made a substantial contribution to the manuscript and have read and approved the final manuscript.

Acknowledgments

The authors would like to thank the Italian Network for Research in Urologic-Oncology (Meet-URO group) for its support on genito-urinary tumors research.

Disclosure statement

Dr Rebuzzi received honoraria as speaker at scientific events and travel accommodation from BMS, Amgen, GSK, Ipsen, Astellas, Janssen, MSD. Dr Fornarini services advisory boards for Astellas, Janssen, Pfizer, Bayer, MSD, Merck and received travel accomodation from Astellas, Janssen, Bayer. Dr Buti received honoraria as speaker at scientific events and advisory role by BMS, Pfizer, MSD, Ipsen, Roche, Eli Lilly, AstraZeneca, Pierre-Fabre, Novartis, Merck, Gentili, Astellas. Dr Banna reports personal fees from AstraZeneca and Astellas for speaker bureau. Dr Rescigno services advisory boards for MSD, AstraZeneca and Janssen. Prof. Signori has no conflicts of interest to disclose.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.