The advent of targeted agents directed against vascular endothelial growth factor (VEGF) and mammalian target of rapamycin signalling pathways has led to a paradigm shift in the treatment of metastatic clear cell renal cell carcinoma (mRCC). In patients with clear cell histology and a favourable or intermediate-risk score [1,2], three different strategies have been approved for first-line treatment, either the tyrosine kinase inhibitors (TKI) sunitinib or pazopanib or the anti-VEGF antibody bevacizumab in combination with interferon-alpha (IFN-alpha).
Sunitinib, a multikinase inhibitor, has been shown to prolong progression-free survival (PFS) when compared with the former standard treatment IFN-alpha [11 months versus 5 months; hazard ratio (HR) 0.42; 95% confidence interval (CI) 0.32–0.54] . The median overall survival (OS)  was 26.4 and 21.8 months for sunitinib and IFN-alpha patients, respectively (HR 0.82; 95% CI 0.67–1.00, P=0.051). Censoring patients who crossed over from IFN-alpha to sunitinib led to a larger difference in OS 26.4 and 20.0 months for sunitinib and IFN-alpha, respectively (HR 0.81, 95% CI 0.66–0.99, nominal P = 0.036).
Two randomized trials have compared the combination of bevacizumab and IFN-alpha to IFN-alpha [5,6]. The AVOREN multicentre randomized double-blind phase III trial included 649 treatment-naive patients. Patients were randomized to receive either bevacizumab along with IFN-alpha or placebo along with IFN-alpha. The primary endpoint was OS, secondary endpoints included PFS and measures of safety. OS was not statistically significant different between treatment arms (bevacizumab + IFN-alpha 23.3 months; IFN-alpha along with placebo 21.3 months, HR 0.86; 95% CI 0.72–1.04, P = 0.13); this lack of evidence of benefit may be due to poststudy treatments offered in both arms. The median PFS was significantly longer in patients assigned to bevacizumab along with IFN-alpha than in patients assigned to placebo along with IFN-alpha (10.2 versus 5.4 months; HR 0.63, 95% CI 0.52–0.75, P = 0.0001).
A second trial, the CALGB trial 90206 , randomized 732 patients to receive bevacizumab along with IFN-alpha or IFN-alpha monotherapy. As in the AVOREN trial, OS was the primary endpoint. Again, no significant difference in OS between treatment arms was found : the median OS was 18.3 months (95% CI 16.5–22.5 months) for bevacizumab + IFN-alpha and 17.4 months (95% CI 14.4–22.5 months) for IFN-alpha alone. As in the AVOREN trial, the median PFS was significantly longer for patients assigned to bevacizumab along with IFN-alpha (8.5 months, 95% CI 7.5–9.7 months) than for patients with IFN-alpha monotherapy 95% CI 3.1–5.6 months (HR 0.71; 95% CI 0.61–0.83; P < 0.0001). Although both trials failed to demonstrate a statistically significant benefit on OS, Food and Drug Administration and European Medicines Agency approved the combined anti-VEGF-immunomodulatory strategy for the first-line treatment of mRCC because of its impact on PFS.
Pazopanib, a more selective VEGFR-TKI, was studied in a randomized, double-blind placebo-controlled phase III trial in treatment naive as well as cytokine-refractory patients . Forty-six percent of patients were cytokine refractory. The primary endpoint was PFS. Patients assigned to placebo were allowed to cross to pazopanib in an open-label fashion. The median PFS of the entire population, treatment naive and cytokine refractory, was significantly longer in patients receiving pazopanib compared with placebo (all patients: 9.2 versus 4.2 months, HR 0.46, 95% CI 0.34–0.62, P < 0.0001; treatment-naive patients: 11.1 versus 2.8 months, HR 0.40, 95% CI 0.27–0.60, P < 0.0001; cytokine-refractory patients: 7.4 versus 4.2 months, HR 0.54, 95% CI 0.35–0.84; P < 0.001). The difference in OS was not statistically significant : pazopanib 22.9 months, placebo 20.5 months HR 0.91, 95% CI 0.71–1.16, P = 0.23. The authors explained this lack of evidence of benefit by the frequent and early crossover from placebo to pazopanib.
In first-line patients with favourable or intermediate risk, sunitinib, pazopanib and bevacizumab along with IFN have been recommended with identical evidence level I, recommendation grade A (IA) . In clinical practice, most physicians and patients are likely to favour an oral treatment (i.e., sunitinib or pazopanib). Choosing between those two TKIs, however, may prove challenging. Biomarkers studied so far focus on the individual biology of the tumour and its addiction to an inhibited pathway rather than on the activity of one agent versus another. As the main mode of action of all three treatment options is inhibition of VEGF-signalling, it is unlikely that any of these biomarkers can provide enough upfront information to allow rational selection of treatment. It has been suggested that differences in the toxicity profile of both agents may allow for treatment selection. These differences between sunitinib and pazopanib have been attributed to their different kinase selectivity. Sunitinib was shown to inhibit more kinases than pazopanib including the platelet-derived growth factor-b TK, c-KIT, Flt-3, etc. Multikinase inhibition appears to be associated with more toxicities than single-kinase inhibition, e.g. sunitinib is more active against Flt-3, thus leading to more pronounced myelotoxicity than pazopanib .
In an attempt to facilitate treatment choices between these two TKIs, two randomized head to head trials have been conducted, COMPARZ  and PISCES  (Figs. 1 and 2). We discuss below the design of these trials, their results, their potential interpretation, and their clinical implications for selecting treatment.
RECENT FINDINGS AND INTERPRETATION
The COMPARZ trial was a randomized noninferiority trial testing whether the PFS on patients with pazopanib is not inferior to that of sunitinib. The investigators defined that pazopanib would be declared noninferior to sunitinib if the upper bound of the 95% CI for the HR of pazopanib to sunitinib was less than 1.25. The final analysis included 927 patients randomized in the phase III COMPARZ trial and another 183 patients from a randomized phase II trial conducted in Asia (NCT01147822). A total of 1110 patients from these two trials received pazopanib 800 mg given once daily continuously or sunitinib at a dose of 50 mg once daily for 4 weeks followed by 2 weeks off treatment. Secondary outcomes included objective response rates and OS. Quality of life (QoL) was an additional secondary endpoint in the COMPARZ trial but not in the Asian trial. In the intent-to-treat population of the combined data from the 1110 patients from both trials, the HR for PFS was 1.05 with a 95% CI of 0.90–1.22. Thus, although the point estimate of pazopanib showed a direction of effect that was worse than in the sunitinib group, the combined data showed noninferiority of pazopanib when compared with sunitinib according to the predefined noninferiority margin. OS did not show a statistically significant difference between the two groups (HR 0.91; 95% CI 0.76–1.08). Pazopanib and sunitinib arms did have different toxicity profiles (fatigue 63% versus 55%, hand-foot-syndrome 50% versus 29%, thrombocytopenia 78% versus 41%, and increase in alanine aminotransferase 60% versus 53%, respectively); 11 of the 14 health-related QoL domains showed statistically significant difference favouring pazopanib. The authors concluded that pazopanib and sunitinib have similar efficacy but that safety and QoL favoured pazopanib.
Several aspects regarding the trial methodology, findings, and conclusions drawn from the data raise both clinical and statistical questions.
Understanding noninferiority designs and the noninferiority margin
Clinical interpretation of the results of a noninferiority trial requires understanding the meaning of noninferiority in general and what the predefined noninferiority margin signifies clinically. First, the finding of noninferiority does not mean that the treatments under study are ‘similar’, it means that the new treatment (here, pazopanib) is not unacceptably worse than the old (here, sunitinib). In this case, ‘unacceptably worse’ means that a 25% loss of PFS is acceptable. It is not clear why this degree of loss is indeed acceptable, if a new treatment had a HR of 0.75 relative to an old, the new one would be considered better. So why in the noninferiority setting did the COMPARZ investigators set a margin of 25%? Even the 22% upper bound observed in COMPARZ is a larger margin than we suspect most clinicians would accept.
Another way of looking at the margin is to translate the statistical definition of noninferiority into numbers related to median PFS. In this case, a 25% increase in HR translates into a roughly 1.9 month difference in median PFS; the observed 22% upper bound translates into a 1.7 month difference. In the past, approval of several agents has been based on almost similar differences in PFS. For instance, axitinib has been approved as a second-line TKI based on a superiority versus sorafenib of 2 months in median PFS (axitinib and sorafenib 6.7 and 4.7 months, respectively, HR 0.67, 95% CI 0.54–0.81, one-sided P < 0.0001). A loss of between 1.7 and 1.9 months in PFS in pazopanib-treated patients might be acceptable if pazopanib has other important advantages (e.g., a dramatically superior toxicity profile or better QoL). Below we examine whether this is the case here.
Differences in the toxicity profile and quality of life and their clinical implications
Both pazopanib and sunitinib have toxicities but their types differ. Regarding items assessed in patient-reported outcomes, the articles did not provide information on whether the scores were above the Minimally Important Difference scores [14,15], which would be considered clinically significant improvements. The Minimally Important Difference may be defined as the difference large enough for the patient to notice there is one or as the smallest change perceived by the patient as an advantage or disadvantage that could lead to changes of treatment. Providing data on the Minimally Important Differences would have been important because if the differences between pazopanib and sunitinib are below this level, the differences would be considered clinically not meaningful and should not be taken into account when making treatment decisions. In open label trials, patient-reported outcomes must be interpreted especially cautiously. Increased competition in a global market requires pharmaceutical companies to seek methods to differentiate their products from those of competitors. Sunitinib had been approved long before pazopanib as a first-in-class, in contrast, pazopanib needed to satisfy regulatory needs when entering an already crowded RCC market.
In this context, note that the assessment of QoL and toxicity on day 28 was structurally biased against the sunitinib arm. In contrast to pazopanib, which is given continuously, sunitinib is given on a schedule of 4 weeks on–2 weeks off schedule. Thus, day 28 would be expected to have the worst toxicity and QoL. A fair comparison of the two drugs would have been an average toxicity and QoL over the course of treatment, not the levels at the worst time for sunitinib.
Have the data really shown noninferiority? Intention-to-treat and per protocol population in COMPARZ
The intention-to-treat and per protocol analysis consisted of 1110 and 995 patients, respectively. In a superiority trial, the most important comparison is the intent-to-treat analysis because it provides an unbiased assessment of the treatment groups. It has the virtue of comparing like with like (i.e., the randomized groups). Unfortunately, in a noninferiority trial that uses intent-to-treat analysis, nonadherence attenuates the differences in treatment effect, leading to an increased probability of finding noninferiority. In the extreme case, a study in which no one adhered to their assigned treatment would show noninferiority. In COMPARZ, roughly 10% of the study group did not adhere to their assigned regimen, these 10% bring the two groups closer together and therefore increase the probability of finding noninferiority . That is why regulators generally recommend that studies testing noninferiority need to satisfy the noninferiority criterion for both the intention-to-treat and a well defined per protocol population [16,17]. Interpretation of open-label trials, especially those with subjective outcomes like PFS and patient-report symptoms, is problematic; the difficulties are even worse for noninferiority .
The PISCES trial, the second head-to-head comparison between sunitinib and pazopanib, was a double-blind cross-over trial that focused primarily on the patient's preference for suitinib or pazopanib . It also assessed the influence of treatment on health-related QoL and safety. Patients were randomized to receive either pazopanib or sunitinib for 10 weeks. After a 2-week break, treatment was switched to the other TKI for another 10 weeks. At the end of this period, patients were asked whether they preferred the first or the second treatment. The trial randomized 169 patients, 114 met the criteria for the preference analysis. Seventy percent of patients preferred pazopanib over sunitinib, 22% preferred sunitinib, and 8% of patients stated no preference (P < 0.001). The most common reasons for these decisions were less fatigue and better overall QoL when choosing pazopanib and less diarrhoea when preferring sunitinib. Physicians were also asked about their preference in this trial, 61% and 22% preferred pazopanib and sunitinib, respectively, whereas 17% had no particular preference. Pazopanib showed less fatigue, hand-foot syndrome, and mucositis than sunitinib.
This trial has several important methodological limitations. First, as in the COMPARZ trial, the design was structured in a way that was unfavourable to the sunitinib arm. Patients were asked about their preference at the end of week 10, which, for patients on sunitinib, was day 28 of treatment. Second, it would be important to know whether patients had been informed prior to treatment that several side-effects were shown to be predictive of efficacy. Their preferences may have been influenced by what they knew of the expected side-effects. Finally, 33% of patients were not evaluable for the primary endpoint for various reasons.
The first head-to-head comparisons between pazopanib and sunitinib confront the clinical with several pitfalls; thus, clinical considerations and physicans’ experience, in addition to the reported results of these trials, may continue to influence treatment choices. Differences in the toxicity profile should be considered when selecting treatment. Pazopanib treatment has been shown to be associated with less off-target toxicity (i.e., toxicities arising from inhibition of other targets than VEGFR) . Examples are myelotoxicity, hand-foot-syndrome, etc. Whenever these toxicities present a treatment-limiting condition, a switch from sunitinib to pazopanib may be reasonable. Importantly, sunitinib was not used optimally in either COMPARZ or PISCES. Since publication of these trials, several authors have published their experience with alternative sunitinib schedules, mostly 2 weeks on–1 week off (schedule 2/1); they have reported a significant reduction of incidence and grade of side-effects. Miyake et al. reported that the proportion of patients experiencing grade 3 or more toxicities was significantly lower on schedule 2/1 than on schedule 4/2. This resulted in a significant improvement in QoL as assessed by short form 36 health survey questionnaire (SF-36). A small study in Asia found similar results . Najjar et al. retrospectively reviewed the records of 30 patients who were changed from schedule 4/2 to schedule 2/1, grade 3/4 toxicities being the reason in all but one of these patients. On the 2/1 schedule, significantly fewer grade 3 and 4 toxicities were observed (P = 0.0001); the reduction was particularly pronounced for fatigue (P = 0.0003) and hand-foot syndrome (P = 0.0004). Neri B et al. presented data from 31 patients receiving sunitinib on a 2/1 schedule. The treatment duration was 16 months (range 2.0–36.0+ months), 42% of patients achieved an objective response. The authors concluded that sunitinib 50 mg given in a 2/1 schedule provides high efficacy and reduced toxicity. Buti et al. studied a different schedule. Upon development of treatment-limiting toxicities, patients were switched to sunitinib on days 1–5 in weeks 1–5 and sunitinib 50 mg on days 1, 3, and 5 in week 6. The authors reported that the reduction of toxicities was associated with a decrease in delays and dose reductions from 50% to 25% and from 37% to 12%, respectively. Finally, Bjarnason et al.[24▪▪] conducted a retrospective review on 172 patients treated with sunitinib either at the standard schedule (4/2) or at four different dose/schedule modifications. The authors reported on a significant improvement in PFS (10.9–11.9 versus 5.3 months, respectively, P < 0.001) and OS (23.4–24.5 versus 14.4 months, respectively, P = 0.003) in patients treated at individual schedules when compared with the standard schedule (Table 1).
In the context of currently available data on the efficacy and tolerability of sunitinib when given on a 2/1 schedule and because of the lack of clinically meaningful differences in QoL in COMPARZ, it appears unlikely that a potential advantage in terms of either tolerability or QoL, or both, of pazopanib over sunitinib will be maintained. Several randomized phase II trials are studying an alternative sunitinib schedule (NCT 02398552, NCT00570882, NCT02060370). More recently, novel immunotherapies using T-cell checkpoint inhibitors have raised enormous interest in mRCC. These agents are currently being explored compared to, or in combination with, sunitinib. The fact that all these trials use sunitinib on a 4/2 schedule is a frequently expressed concern. Choosing the optimal primary endpoint is also a difficult task for future clinical trials. Although OS is an attractive endpoint, it is heavily biased by subsequent treatment choices and treatment availabilities. Differences in OS may become visible only when studying novel agents that are expected to dramatically improve OS. This might not be the case yet with TKIs, which appears to have reached a plateau regarding efficacy. In the context of comparisons between agents with similar mode of action, such as sunitinib and pazopanib, PFS remains an ideal endpoint; however, the ideal head-to-head comparison of pazopanib versus sunitinib is yet to be performed .
The authors thank Ursula Vogl, MD, for her assistance with reference management.
Financial support and sponsorship
Conflicts of interest
MS has received honoraria from Novartis, Pfizer, Roche, Bayer, Bristol-Myers Squibb, GlaxoSmithKline, AVEO. and Astellas. Statistics Collaborative, the company for which Dr Wittes works, has contracts with Pfizer and GSK. There are no conflicts of interest.
REFERENCES AND RECOMMENDED READING
Papers of particular interest, published within the annual period of review, have been highlighted as:
- ▪ of special interest
- ▪▪ of outstanding interest
1. Heng D. A unified prognostic model for first- and second-line targeted therapy in metastatic renal cell carcinoma
(mRCC): Results from a large international study. Proc Am Soc Clin Oncol 2010; 28 (Suppl 15): abstr 4523. 2010..
2. Motzer RJ, Bacik J, Murphy BA, et al. Interferon-alfa as a comparative treatment for clinical trials of new therapies against advanced renal cell carcinoma
. J Clin Oncol 2002; 20:289–296.
3. Motzer RJ, Hutson TE, Tomczak P, et al. Sunitinib versus interferon alfa in metastatic renal-cell carcinoma. N Engl J Med 2007; 356:115–124.
4. Motzer RJ, Hutson TE, Tomczak P, et al. Overall survival and updated results for sunitinib compared with interferon alfa in patients with metastatic renal cell carcinoma
. J Clin Oncol 2009; 27:3584–3590.
5. Escudier B, Bellmunt J, Negrier S, et al. Phase III trial of bevacizumab plus interferon alfa-2a in patients with metastatic renal cell carcinoma
(AVOREN): final analysis of overall survival. J Clin Oncol 2010; 28:2144–2150.
6. Rini BI, Halabi S, Rosenberg JE, et al. Phase III trial of bevacizumab plus interferon alfa versus interferon alfa monotherapy in patients with metastatic renal cell carcinoma
: final results of CALGB 90206. J Clin Oncol 2010; 28:2137–2143.
7. Rini BI, Halabi S, Rosenberg JE, et al. Bevacizumab plus interferon alfa compared with interferon alfa monotherapy in patients with metastatic renal cell carcinoma
: CALGB 90206. J Clin Oncol 2008; 26:5422–5428.
8. Sternberg CN, Davis ID, Mardiak J, et al. Pazopanib in locally advanced or metastatic renal cell carcinoma
: results of a randomized phase III trial. J Clin Oncol 2010; 28:1061–1068.
9. Sternberg CN, Hawkins RE, Wagstaff J, et al. A randomised, double-blind phase III study of pazopanib in patients with advanced and/or metastatic renal cell carcinoma
: final overall survival results and safety update. Eur J Cancer 2013; 49:1287–1296.
10. Escudier B, Porta C, Schmidinger M, et al. Renal cell carcinoma
: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 2014; 25 (Suppl 3):iii49–iii56.
11. Kumar R, Crouthamel MC, Rominger DH, et al. Myelosuppression and kinase selectivity of multikinase angiogenesis inhibitors. Br J Cancer 2009; 101:1717–1723.
12. Motzer RJ, Hutson TE, Cella D, et al. Pazopanib versus sunitinib in metastatic renal-cell carcinoma. N Engl J Med 2013; 369:722–731.
13. Escudier B, Porta C, Bono P, et al. Randomized, controlled, double-blind, cross-over trial assessing treatment preference for pazopanib versus sunitinib in patients with metastatic renal cell carcinoma
Study. J Clin Oncol 2014; 32:1412–1418.
14. Cella D, Yount S, Sorensen M, et al. Validation of the Functional Assessment of Chronic Illness Therapy Fatigue Scale relative to other instrumentation in patients with rheumatoid arthritis. J Rheumatol 2005; 32:811–819.
15. Cella D, Yount S, Brucker PS, et al. Development and validation of a scale to measure disease-related symptoms of kidney cancer. Value Health 2007; 10:285–293.
16. Schumi J, Wittes JT. Through the looking glass: understanding noninferiority. Trials 2011; 12:106.
17. United States Food and Drug Administration: Guidance for Industry Non-Inferiority Clinical Trials. 2010.
18. Rothmann M, Li N, Chen G, et al. Design and analysis of noninferiority mortality trials in oncology. Stat Med 2003; 22:239–264.
19. Miyake H, Harada K, Miyazaki A, Fujisawa M. Improved health-related quality of life of patients with metastatic renal cell carcinoma
treated with a 2 weeks on and 1 week off schedule of sunitinib. Med Oncol 2015; 32:78.
20. Makino K, Yoda K, Tomoishi J, Kume H. Efficacy and tolerability of a low-dose, 2-week administration of sunitinib followed by a week rest (2/1 schedule) for metastatic renal cell carcinoma
: a single center experience of six cases. BMC Res Notes 2014; 7:872.
21. Najjar YG, Mittal K, Elson P, et al. A 2 weeks on and 1 week off schedule of sunitinib is associated with decreased toxicity in metastatic renal cell carcinoma
. Eur J Cancer 2014; 50:1084–1089.
22. Neri B, Vannini A, Brugia M, et al. Biweekly sunitinib regimen reduces toxicity and retains efficacy in metastatic renal cell carcinoma
: a single-center experience with 31 patients. Int J Urol 2013; 20:478–483.
23. Buti S, Donini M, Lazzarelli S, Passalacqua R. A new modified schedule of sunitinib for metastatic renal cell carcinoma
: a retrospective analysis. Acta Biomed 2012; 83:88–94.
24▪▪. Bjarnason GA, Khalil B, Hudson JM, et al. Outcomes in patients with metastatic renal cell cancer treated with individualized sunitinib therapy: correlation with dynamic microbubble ultrasound data and review of the literature. Urol Oncol 2014; 32:480–487.
The publication by authors is the largest patient series on sunitinib given at individualized schedules. The authors report on a significant improvement in PFS and OS for patients being treated at an individualized schedule when compared with the standard schedule. In addition, the authors provide pharmacokinetic data from 23 patients included in this trial.
25. Escudier B, Pluzanska A, Koralewski P, et al. Bevacizumab plus interferon alfa-2a for treatment of metastatic renal cell carcinoma
: a randomised, double-blind phase III trial. Lancet 2007; 370:2103–2111.
26. Motzer RJ, Barrios CH, Kim TM, et al. Phase II randomized trial comparing sequential first-line everolimus and second-line sunitinib versus first-line sunitinib and second-line everolimus in patients with metastatic renal cell carcinoma
. J Clin Oncol 2014; 32:2765–2772.
27. Wittes J. Active-control trials: a linguistic problem. Int Chin Stat Assoc Bull 2001; 39–40.