Axitinib beyond first-line therapy of Metastatic Renal Cell Carcinoma: Real World Data from the STAR-TOR registry

1 INTRODUCTION

Treatment of advanced or metastatic renal cell carcinoma (a/mRCC) is continuously evolving, with checkpoint inhibitors (CPIs) being now part of all first line treatment regimens for suitable patients [1–3]. To simultaneously target various tumor pathways, CPIs are preferably administered as combination therapy [1–3]. The approval of combined Pembrolizumab or Avelumab and Axitinib for all a/mRCC risk groups in 2019 has lately highlighted the clinical relevance of the anti-VEGF tyrosine kinase inhibitor (TKI) Axitinib [4, 5]. While now considered a first-line treatment in combination with checkpoint inhibitors, Axitinib monotherapy had originally been approved in 2012 as second-line treatment after Sunitinib or cytokines according to the pivotal AXIS trial data [6]. The first phase III trial in which Axitinib was evaluated as monotherapy in second-line after cytokine or Sunitinib in mRCC patients demonstrated a longer median progression-free survival (PFS) of 6.7 months for Axitinib versus 4.7 months for Sorafenib (p < 0.0001) [6].

Although combinations of CPI and targeted therapy have demonstrated superior outcomes to previous regimens, few of the targeted therapies have been comprehensively examined for their efficacy and adverse effect profile in a single-agent setting [7–9]. This complicates the assessment of the contribution to efficacy and toxicity of each individual agent in immunotherapy combinations.

In addition, Axitinib is one of the standard second and third line treatment options after CPI-treatment failure [10]. As demonstrated by a large-scale national study in Germany, 56% of fist-line patients progress to second-line, 29% to third-line, and up to 13% to subsequent lines [11]. This context highlights the clinical relevance of Axitinib monotherapy, even in the CPI era. A recent meta-analysis by Wenzel et al. suggests a better second-line median PFS for Pazopanib after IO-failure, when compared to Axitinib [12]. Yet, the authors themself admit their analyses to be based on retrospective studies with small sample size and heterogeneous patient populations. Nevertheless, the study demonstrates that Axitinib plays an important role as treatment option for second and later lines.

Still, real-world data from large a/mRCC patient cohorts treated with Axitinib is limited [13–18].

Therefore, this study aims to evaluate Axitinib as a single-agent in the real-world setting. The study is one of the largest patient cohorts with clinical routine data that has been published so far. The here presented details bear several yet unknown but important details that can aid in understanding the effects seen in combination therapies.

2 MATERIAL AND METHODS

This study evaluates data originating from the prospective STAR-TOR registry maintained by Pfizer Pharma GmbH, Berlin, Germany, and was conducted in accordance with the Declaration of Helsinki in its most recent version. The STAR-TOR registry received prior approval by the ethics committee of Münster University Hospital in 2007 (Nr.2007-484-f-S) and was registered in the US library of medicine database (NCT00700258). The registry is intended to evaluate the effectiveness, tolerability, and safety of patients with advanced RCC, recurrent or refractory mantle cell lymphoma and gastrointestinal stromal tumors treated with a) Axitinib, b) Sunitinib, or c) Temsirolimus in a routine clinical setting. Additional details regarding the STAR-TOR registry have been previously published [19–23].

3 RESULTS

3.3 Axitinib treatment response

207 patients were analyzed for best overall response. The clinical benefit rate of Axitinib was 50.7% : 3 patients (1.4%) experienced a complete response (CR), 35 patients (16.9%) had partial remission (PR), and 67 patients (32.4%) experienced a stable disease (SD). Progressive disease (PD) was observed in 56 patients (27.1%). 46 patients were not assessable. For 3 patients, response data were missing. Table A in the supplementary material shows the response rates by therapeutic line. We did not detect marked differences even though the results have to be interpreted cautiously as patient numbers decreased in later lines.

The overall median PFS was 5.6 months and median OS was 18.3 months. Median PFS and OS for 108 second-line patients were 4.8 and 16.1 months. Patients in later lines (n = 102) demonstrated numerically longer PFS (6.3 months) and OS (18.4 months), respectively (p = 0.68 and 0.75).

For patients older than 65 years compared to patients≤65 years of age, PFS was noticeably longer with 6.5 vs. 4.3 months (p = 0.005), whereas the OS difference was only numerically longer with 18.4 vs. 15.6 months (p = 0.31). Detailed survival data are shown in Table 3.

OPEN IN VIEWER

Table 3

Progression-free and overall survival according to patient subgroups

	Patient number n	Progression-free survival Months	Overall survival Months
All patients	210	5.6	18.3
Second line patients	108	4.8	16.1
≥Third line patients	102	6.3	18.4
Favourable risk patients *†	17	10.5	37.6
Intermediate risk patients *†	38	5.7	25.8
Poor risk patients *†	58	3.5	9.3
Patients with LDH≤300 U/l †	131	6.2	19.0
Patients with LDH > 300 U/l †	32	5.1	8.2
Patients with clear cell histology	170	5.6	18.3
Patients with other histologies	40	5.8	22.2
Patients with bone metastases	60	5.2	12.7

* MSKCC score for pretreated patients. ^† exploratorative analysis proved statistical significance with p < 0.05.

As depicted in Fig. 1, survival also differed relevantly by MSKCC prognosis group: median PFS for patients with favorable, intermediate, and poor risk was 10.5, 5.7, and 3.5 months (p = 0.01). For the respective prognosis groups, OS also differed with a median of 37.6, 25.8 and 9.3 months (p = 0.004).

OPEN IN VIEWER

Fig.1

Progression-Free and Overall Survival according to LDH level and MSKCC risk score.

Clear cell RCC patients and those with non-clear cell histologies had similar outcomes in terms of PFS (5.6 and 5.8 months, p = 0.19) and OS (18.3 and 22.2 months, p = 0.28). Due to small numbers, non-clear cell histologies were not differentiated further.

Patients with high baseline LDH levels showed a worse OS and PFS than patients with LDH levels≤300 U/l (OS: 8.2 vs. 19.0 months, p = 0.008; PFS: 5.1 vs 6.2 months, p = 0.15). The corresponding Kaplan-Meier plots are also shown in Fig. 1.

In the ever-evolving landscape of systemic treatment for a/mRCC, current guidelines advocate for combined targeting of various tumor pathways, including CPIs and TKIs [1–3]. Still, the real-world evidence on the effectiveness and adverse event profile on the TKI Axitinib as a single-agent is limited, in particular in a prospective multicenter setting in a European-based patient cohort.

Here, the results of the multicenter STAR-TOR Axitinib cohort are reported, which is one of the largest real-world Axitinib datasets. In this analysis, Axitinib yielded a median PFS of 5.6 months and OS of 18.3 months in a/mRCC patients in second-line treatment and beyond.

High serum LDH levels were identified as a potential prognostic marker of unfavorable outcome: LDH levels > 300U/l were associated with a clinically relevant shorter OS than LDH levels < 300U/l (8.2 vs. 19.0 months, p = 0.008).

The STAR-TOR real-world data compares well to the phase III AXIS trial evaluating Axitinib versus Sorafenib as second-line treatment in mRCC [6, 28]. The AXIS trial reported a median PFS of 4.8 months for patients pretreated with Sunitinib and median OS of 15.2 months, including 33% of MSKCC poor-risk patients.

To provide a comprehensive overview of the literature, Table 5 summarizes outcomes from studies evaluating Axitinib for a/mRCC. Compared to other studies, the STAR-TOR cohort included the highest proportion of MSKCC poor-risk patients (51.3% vs.15.5% and 37% in the studies by Facchini et al. and Rini et al.) [6, 9, 13–17].

Interestingly, despite the inclusion of a remarkable number of poor-risk patients, PFS and OS rates in the STAR-TOR registry are comparable to those reported in other studies. For example, Rini et al. described a median PFS of 6.7 months and OS of 20.1 months for their total study population and 4.8 and 15.2 months for the patients pretreated with Sunitinib, respectively. This compares well to 5.6 and 18.3 months in the STAR-TOR cohort, respectively [6, 28]. These observations could lead to the interpretation that even in poor-risk a/mRCC patients Axitinib yields a high effectiveness. Still, longer PFS and OS rates than reported in Caucasian patients were described in Japanese populations by Miyake et al. and Osawa et al., although the retrospective design and exclusion of roughly 10% of patients with treatment duration of less than 1 month in the latter trial may have distorted the results [16, 18].

While the overall benefit rate of Axitinib in the STAR-TOR registry with 50.7% was lower than that reported in other studies, these differences might partly result from missing data in almost 23% of the patients owing to the STAR-TOR registry's design. This assumption is further supported by a disease progression rate of 30.5%, which is comparable to the most recent Axitinib studies by Facchini et al. and Osawa et al. [17, 18]. We did not observe marked differences in response rates when comparing the therapeutic lines. Similarly, Osawa et al. did not identify treatment line as a significant predictor of OS [18]. Yet, Matias et al. reported significantly longer PFS and OS for patients in second-line treatment versus third or later line [15] A publication by Graham et al. investigated PFS and ORR for patients treated with VEGFR-TKIs (Axitinib, Sunitinib, Cabozantinib, Pazopanib, Bevacizumab, and Sorafenib). The authors reported for second, third, fourth and later lines increasing PFS with 3.8, 5.7, and 6.1 months but decreasing ORR of 23%, 22%, and 10% [29]. We suppose that a strong selection bias may lead to the observed effects.

Regarding Axitinib-related AEs, there is heterogeneous data from the STAR-TOR cohort. On one hand, the rate of gastrointestinal AEs of 37.6% (including 29% diarrhea) is consistent with data obtained from other real-world datasets: Facchini et al. described gastrointestinal disorders in 36.5% of their Axitinib patients and Osawa et al. reported 22.5% of the patients from a Japanese to suffer from diarrhea [17, 18]. On the other hand, the rate of vascular disorders in the STAR-TOR registry (mainly arterial hypertension) with 7.6% was unexpectedly low: Miyake et al. reported hypertension in 58.9% of patients, while hypertension was evident in 40% of patients in the AXIS trial [6, 16]. One reason for these discrepancies might be a higher prevalence of pretreated and thus compensated hypertension among real-world patients enrolled in the STAR-TOR registry compared to the randomized controlled AXIS trial. Also, varying comedications and comorbidities might contribute to the observed differences. Finally, worsening, or new-onset hypertension might be clinically underreported in the real-world setting, as elevated blood pressure could be deemed clinically less relevant than other treatment-related AEs, such as gastrointestinal disorders.

Patients’ serum LDH levels were identified as a prognostic marker for survival in this study, where patients with lower LDH levels had a larger clinical benefit from Axitinib treatment in terms of longer OS than those with high LDH levels (median OS 19 vs. 8.2 months, p = 0.008). While the prognostic potential of LDH levels has so far been not described for Axitinib, other authors have reported similar results with favorable outcomes in mRCC patients with low LDH levels treated with Temsirolimus or interferon alpha [30]. Furthermore, a high serum LDH level is well known to be associated with poor survival in other solid tumors [31].

Interestingly, similar results were reported for patients with clear-cell and non-clear-cell RCC treated with Axitinib in this study. Still, the results need to be interpreted carefully given the small number of non-clear-cell RCC patients in our analysis. In particular, since the authors of other publications reported worse treatment outcomes for non-clear-cell histologies [32, 33].

The here presented study is not devoid of limitations, which are mainly rooted in its design: First, there were missing data across several variables, which is inherent in the nature of a non-interventional trial. While this data unavailability could have distorted our findings, the monitoring of participating STAR-TOR centers reduced the risk of relevant bias. Nevertheless, missing MSKCC score data in almost 50% of cases may have impacted the power of our statistical comparisons of patient survival by risk group.

The design as a register study may yield a certain degree of imprecision regarding the outcome measures. Although all STAR-TOR sites were exhorted to use the RECIST 1.1. criteria, this was not mandatory and radiological assessment of treatment response may lack precision. Likewise, monitoring of adverse events may lack precision as not all physicians used CTCAE.

Second, this study does not provide information about the treatment effectiveness of Axitinib with regard to the type of first-line treatment received since these data were not captured due to the design of the STAR-TOR registry. Furthermore, our study does not include the effect of Axitinib in immunotherapy combinations. Given the recent approval of Axitinib in combination with Pembrolizumab or Avelumab as first-line treatment for mRCC, these data have not been generated in clinical routine registers yet, as patient accrual must increase to provide adequately powered analyses.

Third, our results have to be compared carefully to studies using other methods of risk stratification, such as the IMDC score, even though we consider the MSKCC score used in this study for pretreated patients as an adequate proxy.

Finally, one cannot discount some reporting bias in a phase IV observational, multicenter design like the STAR-TOR registry even though participating centers were randomly monitored. This restriction, however, applies for any data acquisition in the everyday clinical routine setting.

Nevertheless, our study provides important evidence contributing to the existing knowledge concerning Axitinib beyond first line. Notwithstanding the many publications involving treatment strategies with Axitinib, data on real-world experience is sparse:

Our literature search yielded only five other real-world studies [13, 15–18]. Yet, the patient cohorts of Matias et al., Miyake et al., and Facchini et al. were of only about half the size than our study cohort with 210 individuals [15–17]. In addition, the largest publication by Mac Lean et al. with 659 patients did not report on adverse events [13]. Therapeutic lines later than second line were analyzed only by three studies with the publication by Osawa et al. presenting no data beyond third line [13, 15, 18]. Our study presents multicenter data what also holds true only for three of the other the publications [13, 17, 18]. As the studies by Miyake et al. and Osawa et al. report exclusively data on Japanese patients there seems to be limited comparable evidence for European patient populations [16, 18]. Another important point is that except our study only the publication by Matias et al. reports on prospectively collected data [15].

This real-world study confirms the clinical relevance of Axitinib for a/mRCC treatment. Its effectiveness is not only of value in the immunotherapy combinations but also as single-agent in the second-line and beyond setting. Although a high proportion of patients with unfavorable MSKCC poor-risk group were included, OS and PFS were in concordance with pivotal trials, while demonstrating a favorable safety profile. A high LDH serum level seems to be a negative prognostic marker for Axitinib effectiveness, which can aid in clinical decision making.