Camillo Porta, IRCCS San Matteo University Hospital Foundation, Pavia, Italy. e-mail: firstname.lastname@example.org
Study Type – Therapy (retrospective cohort)
Level of Evidence 2b
What’s known on the subject? and What does the study add?
Various targeted agents with differing mechanisms of action and toxicity profiles have now been approved for the treatment of advanced RCC. However, the optimal use of these agents remains a challenge. Since the approval of sorafenib and sunitinib, many patients have been treated with these two tyrosine kinase inhibitors (TKIs) in sequence, with current evidence suggesting that this approach is associated with continued clinical benefit, with limited or no cross-resistance between the two agents. The mammalian target of rapamycin (mTOR) inhibitor, everolimus, has been shown to be as effective after two TKIs as it is after one TKI, emphasizing the importance of understanding the optimal TKI sequence before switching to an mTOR inhibitor, to provide patients with the longest progression-free survival (PFS) benefit.
This retrospective analysis of 189 patients treated sequentially with sorafenib (800 mg/day) and sunitinib (50 mg/day; 4 weeks on 2 weeks off) showed that initial therapy with either agent was associated with similar PFS benefits (median PFS 8.4 months [sorafenib] vs 7.8 months [sunitinib]; HR 1.05, 95% CI 0.78–1.40; P = 0.758). However, patients treated with sorafenib followed by sunitinib (SoSu) appeared to derive a greater PFS benefit than those treated with SuSo (median PFS with second TKI: 7.9 months [SoSu] vs 4.2 months [SuSo]; HR 0.54, 95% CI 0.39–0.74; P < 0.001). Consistent with previous studies, these findings suggest that sequential TKI therapy is associated with continued clinical benefit and that SoSu may result in a longer PFS than SuSo.
• To conduct a retrospective, multicentre, cohort analysis to assess the sequential use of the tyrosine kinase inhibitors (TKIs) sorafenib and sunitinib.
PATIENTS AND METHODS
• Records of 189 patients with renal-cell carcinoma (RCC) who were treated with sorafenib and sunitinib sequentially between March 2004 and April 2009 at 12 Italian study centres were analysed.
• Patients were treated under European Expanded Access Programmes or, following market approval, in general clinical practice.
• Interventions were sorafenib (800 mg/day) and sunitinib (50 mg every day; 4 weeks on and 2 weeks off).
• Progression-free survival (PFS) during treatment with the first and second TKI was evaluated.
• In all, 99 patients were treated with sunitinib followed by sorafenib (SuSo) and 90 were treated with sorafenib followed by sunitinib (SoSu); 104 (55%) patients had received prior systemic therapy, mostly with cytokines.
• The median (range) PFS on the first TKI was similar between treatment groups [sorafenib 8.4 (1.1–28.9) months; sunitinib 7.8 (0.5–30.4) months; hazard ratio (HR) 1.05, 95% confidence interval (CI) 0.78–1.40, P = 0.758]. Multivariate analysis showed that good Memorial Sloan-Kettering Cancer Center status was associated with increased PFS.
• After the second TKI, patients in the SoSu group had a longer median PFS than those in the SuSo group (7.9 months vs 4.2 months, respectively; HR 0.54, 95% CI 0.39–0.74, P < 0.001).
• Multivariate analysis showed only treatment and Eastern Cooperative Oncology Group performance status (and not age, gender, study centre or previous treatment) were significantly associated with duration of PFS.
• Our findings suggest a limited cross-resistance between sorafenib and sunitinib and that the sequence SoSu may result in a longer combined PFS than SuSo. This is the largest retrospective study to date, though its findings are limited in part by the retrospective nature.
REnal Cell cancer treatment with Oral, RAD001 given Daily (trial)
mammalian target of rapamycin
vascular endothelial growth factor
sorafenib followed by sunitinib
sunitinib followed by sorafenib
Response Evaluation Criteria In Solid Tumors
Eastern Cooperative Oncology Group performance status
Memorial Sloan-Kettering Cancer Center
Istituto Nazionale Tumori
With the arrival of targeted therapies (sorafenib, sunitinib, temsirolimus, bevacizumab plus interferon-α, everolimus and pazopanib), patients with advanced RCC now have an expanded range of treatment options, all of which have shown clinical benefit and manageable toxicity in phase III trials [1–5]. However, their different mechanisms of action, distinct and sometimes overlapping toxicity profiles, and the possibility of patient resistance to a particular therapy have raised the question of how to use these agents optimally. As there is currently no long-term cure for advanced RCC, the goal of therapy should be to extend progression-free survival (PFS) while maintaining a patient’s quality of life. This may be achieved by using targeted therapies in combination or in sequence.
Since the approval of sorafenib and sunitinib for the treatment of advanced RCC, many patients have been empirically treated with these two tyrosine kinase inhibitors (TKIs) sequentially to prolong stabilization of disease. This strategy has been supported by results from three small prospective studies [6–8] and several small retrospective studies [9–15], all of which indicate that there is limited or no cross-resistance between sorafenib and sunitinib. Interestingly, most of these studies suggest that the sequence of sorafenib followed by sunitinib may provide longer PFS than sunitinib followed by sorafenib.
More recently, data from the phase III RECORD (REnal Cell cancer treatment with Oral RAD001 given Daily) trial (N = 416) showed that the mammalian target of rapamycin (mTOR) inhibitor, everolimus, was associated with clinical benefit in patients who had been previously treated with TKIs, and resulted in a PFS of 5.9 months and 3.9 months post-sorafenib and post-sunitinib therapy, respectively [5,16]. That study also showed that everolimus was as effective after two TKIs as after one TKI (hazard ratio [HR] for disease progression 0.32 after sorafenib and sunitinib vs 0.25 after sorafenib and 0.34 after sunitinib; P < 0.001) [5,16]. Taken together, data from these prospective and retrospective studies [7,9–15] suggest that a switch in mode of action may not be necessary after disease progression with the first vascular endothelial growth factor (VEGF)-targeted agent. Furthermore, as there is an absence of data supporting the efficacy of any treatment after failure on an mTOR inhibitor, a better understanding of the potential for sequencing TKIs is mandatory if we are to provide patients with the longest possible PFS.
The aim of the present study was to evaluate the benefits of sequential treatment with sorafenib followed by sunitinib (SoSu) or sunitinib followed by sorafenib (SuSo) through the retrospective analysis of data from a large, multicentre cohort. The benefit of each sequence was evaluated in terms of the combined PFS for SoSu compared with SuSo, to assess which sequence resulted in the longest stabilization of disease.
PATIENTS AND METHODS
A retrospective analysis was performed using data from patients with RCC who were treated with SoSu or SuSo between March 2004 and April 2009 at 12 centres across Italy. Patients received sorafenib at a dose of 800 mg/day and sunitinib at a dose of 50 mg every day on a 4 weeks on and 2 weeks off treatment schedule. Patients were treated under European Expanded Access Programmes or, following market approval of the TKIs, in general clinical practice. In Italy, sorafenib is approved for the treatment of patients with advanced RCC who have failed prior interferon-α- or interleukin-2-based therapy or who are considered unsuitable for such therapy, while sunitinib is approved for the treatment of advanced/metastatic RCC.
The status of disease progression during the first and second TKI was determined by radiological assessment (Response Evaluation Criteria In Solid Tumors [RECIST])  approximately every 12 weeks. Patients who were treated with any other agent during the treatment gap between sorafenib and sunitinib therapy were excluded from the present study.
The primary objective of this analysis was to determine PFS on the first and second TKI. Data analysed in this study were obtained from the medical charts of each individual patient. Baseline characteristics, treatment start dates, dates of disease progression and time between first and second TKI were recorded.
Time to progression for the initial treatment period was calculated as the time from the start of the first TKI to the time of disease progression on the first TKI. PFS for the second treatment period was calculated as the time from the start of the second TKI to the time of disease progression or death on the second TKI.
The median PFS was estimated using the Kaplan–Meier method. Patients who remained on the second TKI without disease progression at the end of the study period were censored from the analysis. The median combined PFS was defined as the sum of the first PFS and second PFS (excluding any elapsed time between the two treatment periods).
The potential influence of TKI treatment sequence and baseline characteristics on the median PFS was assessed using a log- rank test (Mantel–Cox, univariate analysis). A stepwise forward Cox proportional hazard model analysis of time to progression was used to assess baseline predictors of PFS and treatment effect simultaneously. The significance of each variable in the model was assessed using a chi-square test.
Kaplan–Meier and Cox univariate analyses were performed using Prism v 5.02 (GraphPad software Inc, USA) and MedCalc v 10.1.6.0 (MedCalc Software, Belgium). A P < 0.05 was considered to indicate statistical significance.
In all, 189 patients were identified for inclusion in this retrospective analysis; 99 were treated with SuSo and 90 with SoSu. Baseline patient characteristics were similar between the treatment groups for age, sex, Eastern Cooperative Oncology Group performance status (ECOG PS) and histology (Table 1). Most of the patients were male (75%), with clear-cell histology (86%) and an ECOG PS of 0 (71%). There was an imbalance between the groups in terms of Memorial Sloan-Kettering Cancer Center (MSKCC) risk status, with more poor-risk patients in the SuSo group than in the SoSu group (32 patients [32%] vs 9 patients [10%], respectively). There was also a difference in the proportion of patients in each group at different study centres: Milan Istituto Nazionale Tumori (INT) only included patients treated with SoSu, whereas Naples mainly included patients treated with SuSo. In addition, more patients in the SuSo group (71%) had received systemic therapy prior to starting treatment with a TKI compared with patients in the SoSu group (38%). Previous treatments included cytokines, chemotherapy and bevacizumab plus interferon-α. In most cases, prior therapy was with cytokines only (71% and 58% for the SoSu and SuSo groups, respectively). The median (range) time between treatment with the first and second TKI was 0.9 (0–13.8) months in the SoSu group and 1.3 (0–15.1) months in the SuSo group; 78% of patients had a treatment gap of <3 months (82% in the SoSu group and 75% in the SuSo group).
Table 1. Baseline demographics and clinical characteristics
Expressed as a percentage of patients who previously received systemic therapy.
Treatments included cytokines (interferon and/or interleukin-2), chemotherapeutic agents (vinblastine, 5-fluorouracil, capecitabine, gemcitabine, irinotecan, cisplatin), progestins (medroxyprogesterone acetate or megestrol acetate), as well as thalidomide.
Median (range) time between first and second TKI, months
Number with gap ≥3 months, n (%)
Centre, n (%)
TIME TO PROGRESSION AFTER INITIAL TKI
At the end of the initial treatment period, the median (range) PFS was similar between treatment groups: 8.4 (1.1–28.9) months for sorafenib and 7.8 (0.5–30.4) months for sunitinib; HR 1.05, 95% CI 0.78–1.40, P = 0.758 (Table 2, Fig. 1A). The similarity in the median PFS was maintained regardless of whether patients had received a previous systemic therapy or not (Table 2, Fig. 1B). Treatment-naïve patients in the SoSu group had a median PFS of 7.0 months vs 7.4 months for patients in the SuSo group. The median PFS of pretreated patients in the SoSu and SuSo groups was 10.6 and 8.2 months, respectively. However, in both treatment-naïve and pretreated patient groups, the differences in PFS were statistically non-significant (P = 0.532 and P = 0.487, respectively).
Table 2. Influence of baseline characteristics on PFS during the first and second TKI treatment periods (univariate analysis)
* P value associated with the chi-square for the log-rank test in the comparison between Kaplan–Meier curves. †As there were only five patients with an ECOG status of 2 this group of patients was not included in the analysis. ‡For one patient there were no data available for histology or MSKCC score. §Other = Rome (11 patients), Legnano (two), Livorno (six), Pordenone (12), Modena (11), Aviano CRO (14), Turin (nine), Udine (six), Milan Niguarda (four).All other centre comparisons were statistically non-significant.
In the univariate analysis, baseline ECOG PS of 0 was associated with a longer PFS than a baseline ECOG PS of 1 (P = 0.013) and a good MSKCC risk status was associated with a longer PFS than an intermediate or poor MSKCC status (P = 0.002 and P = 0.006, respectively; Fig. 2A,B). Age, gender, histology, centre and previous treatment were not statistically significantly associated with duration of PFS. In the multivariate analysis, only baseline MSKCC was an independent predictor of PFS. Patients with a poor or intermediate MSKCC risk status were at greater relative risk of disease progression than those with a good MSKCC risk status (Table 3).
Table 3. Baseline characteristics associated with risk of progression on first and second TKI in a multivariate analysis (stepwise regression)
ECOG PS = 1
PFS AFTER SEQUENTIAL TREATMENT
Among patients treated with SoSu and SuSo, 16 of 90 (18%) and 14 of 99 patients (14%), respectively, remained on the second treatment without disease progression.
The median (range) PFS after the second TKI was longer for those patients who received SoSu, at 7.9 (0.8–26.9) months, than for those patients who received SuSo, at 4.2 (0.1–34.7) months, HR 0.54, 95% CI 0.39–0.74, P < 0.001 (Table 2, Fig. 1C). Consequently, the median combined PFS was longer for the SoSu group (16.3 months) than for the SuSo group (12.0 months). In addition, pretreated patients in the SoSu group had a significantly longer PFS vs pretreated patients in the SuSo group (8.3 vs 4.4 months, P = 0.006; Fig. 1D). Treatment-naïve patients in the SoSu group also had a longer PFS vs treatment-naïve patients in the SuSo group, although this difference failed to reach statistical significance (7.6 vs 3.5 months respectively, P = 0.054).
In the univariate analysis, treatment, baseline ECOG PS (0 vs 1), baseline MSKCC (good vs poor and poor vs intermediate) and centre (Milan vs Naples and Milan vs other) all showed some association with duration of PFS on the second treatment (Table 2). However, only treatment and ECOG PS retained an association with duration of PFS in the multivariate model. Patients in the SoSu group had a lower relative risk of disease progression on the second treatment vs patients in the SuSo group (Table 3; relative risk 0.54, 95% CI 0.40–0.75, P < 0.001).
The results of this study suggest a limited cross-resistance between sorafenib and sunitinib and indicate that patients who have disease progression on sorafenib or sunitinib can derive further clinical benefit by switching to the other agent in a ‘real-life’ clinical setting. The data presented here also suggest that the use of SoSu results in a longer combined PFS than SuSo.
To our knowledge this is the largest, multicentre, retrospective study performed to date to report comparative results of the sequential use of sorafenib and sunitinib in a clinical setting. The retrospective nature of the present study does impose some limitations. For example, the lack of standardized assessment of response between centres meant that it was not possible to evaluate clinical benefit rates, even though the timelines for response assessment were followed by all centres. However, response rate was not the primary endpoint in this evaluation.
There were some imbalances between treatment groups for the distribution of patients across the study centres, the percentages of patients with intermediate or poor MSKCC risk scores (e.g. 10% of patients were poor risk in the SoSu treatment group compared with 32% in the SuSo group) and the percentage of patients who had received a systemic therapy prior to the first TKI. Nevertheless, multivariate analyses showed that neither the study centre nor previous treatment status were independent predictors of the duration of PFS in either treatment period. Furthermore, although MSKCC risk score was an independent predictor of increased PFS on the initial study TKI, it was not an independent predictor of increased PFS on the second study TKI.
Another factor to consider is that some patients had a gap of more than one treatment cycle (3 months) between the two targeted agents, and disease progression during this period may have influenced the response to subsequent treatment. This potential bias may be counterbalanced by the fact that the two treatment groups were matched in terms of interval between treatments.
The first phase of the present study showed that PFS for the initial TKI treatment period was similar for sorafenib and sunitinib therapy (8.4 months and 7.8 months, respectively). This similarity in PFS was maintained regardless of whether patients had received a previous systemic therapy or not. The PFS for sorafenib reported here was higher than that obtained in the phase III Treatment Approaches in Renal Cancer Global Evaluation Trial (TARGET) (5.5 months) . However, in that study all patients had been pre-treated, mainly with cytokines, whereas in the present study, 62% of the group receiving sorafenib in the initial treatment period had received no previous systemic treatment. Conversely, the PFS reported here for sunitinib was lower than that reported in the phase III sunitinib study (11.6 months)  but again there are differences between the study populations; the phase III study was carried out in treatment-naïve patients  whereas 71% of the group receiving sunitinib in the initial treatment period in the present study had received previous systemic therapy. Indeed, PFS for sunitinib-treated patients in the two phase II studies leading to registration of the drug (performed in pre-treated patients) was 8.7 months and 8.3 months, respectively [18,19]. Other differences between study populations, such as number and location of metastatic sites, may also contribute to the differences in PFS noted above (comparative data are not available to examine this in detail). It is worth noting that data from other retrospective studies also suggest little difference between first-line sorafenib and sunitinib for PFS [9–14], with four of these studies reporting a PFS of ≈8 months for both agents [11–14]. Randomized trials and a non-randomized Expanded Access Programme have also reported PFS values for sorafenib in first-line treatment that are similar to the 8.4 months reported here (7.4–9.2 months) [20–23].
We found that treatment with a second TKI provided some clinical benefit in both treatment groups. However, multivariate analysis showed that the relative risk of disease progression on the second treatment was lower in the SoSu group compared with the SuSo group. These results are comparable with those reported previously in studies that included patients treated with sorafenib and sunitinib [9,10,14]. For example, in the sorafenib European Expanded Access study, patients previously treated with sunitinib achieved a median PFS of 4.1 months on subsequent sorafenib treatment , which is comparable with the PFS of 4.2 months for sorafenib post-sunitinib reported in the present study.
The combined PFS was longer for the SoSu sequence (16.3 months) than for the SuSo treatment sequence (12.0 months). This confirms the trend observed in other retrospective studies [9,10,13–15,25], where SoSu was reported to result in a combined PFS of between 11.8  and 21 months , whereas the SuSo sequence resulted in a combined PFS of between 8.6  and 17.4 months  (Fig. 3) [7–15,26]. The reasons for the shorter PFS for patients treated with the SuSo sequence vs SoSu are unclear. One explanation could be that patients receiving sunitinib first may have more severe adverse events than patients receiving sorafenib as the initial treatment . These patients may be ‘weakened’ by the toxicity they experience on their first TKI, resulting in inferior efficacy and/or duration of treatment on their second TKI compared with those patients treated with sorafenib first. Another hypothesis could be that initial sunitinib treatment leads to the accelerated development of resistance. Indeed, a study carried out in a murine cancer model found that potent anti-angiogenic treatment resulted in adaptive and invasive tumour progression . As there are currently no clinical data to support these theories, further studies are needed to confirm these hypotheses.
These data represent only part of the treatment sequencing options. The mTOR inhibitors, everolimus and temsirolimus, have also been investigated after treatment with TKIs, but are associated with very different patient outcomes. In the phase III RECORD study, everolimus resulted in a PFS of 4.9 months following multiple previous systemic therapies (including sorafenib and/or sunitinib) . In a subgroup analysis of that study, the PFS for everolimus post-sunitinib was similar to that for sorafenib post-sunitinib in the present study (3.9 months vs 4.2 months, respectively) . Conversely, temsirolimus (currently approved for advanced, high-risk, treatment-naïve RCC) has provided limited benefit (PFS between 1.4 and 1.9 months ) in the second-line setting after treatment with sunitinib or sorafenib .
Due to the differences in study designs and patient populations, the results of these studies investigating sequential targeted agents cannot be directly compared and it is not possible to fully define the optimal sequence of treatments for advanced RCC. In all cases, the choice of targeted therapies should always be individualized according to patient and disease characteristics as well as treatment history . However, as the current evidence suggests that there is limited cross-resistance between TKIs and that the use of two TKIs does not compromise the efficacy of everolimus, in the absence of data to support the use of any targeted agent after failure of mTOR treatment, there appears to be a need to explore treatment options with VEGF-targeted agents before switching to mTOR inhibitors.
Although phase III data represent the highest level of evidence and warrant the strongest grade of recommendation, it is important to consider all levels of evidence when making treatment decisions. With the paucity of data from large, prospective, multicentre studies, examining the optimal sequencing approach, retrospective studies can provide valuable clinical data. Indeed, Professor Sir Michael Rawlins (Chairman of the UK National Institute for Health and Clinical Excellence) recently called for decision makers to: ‘assess and appraise all the available evidence irrespective as to whether it has been derived from randomized clinical trials (RCTs) or observational studies’. This is because: ‘RCTs are generally undertaken in selected patient populations for a finite, usually relatively brief, period of time’. In contrast, retrospective data describe the efficacy and tolerability of targeted agents across patient subpopulations and include patients who are often under-represented in phase III clinical trials  (e.g. patients with a poor performance status and elderly patients).
In conclusion, the results of the present study add to the evidence base suggesting that there is limited cross-resistance between sorafenib and sunitinib. Furthermore, the relative efficacies of the TKI sequences support the theoretical importance of continuous inhibition of the VEGF pathway in RCC. However, there is still a clear need for comparative prospective data to confirm the optimal sequence of available (or soon to be available) targeted agents. It is anticipated that ongoing studies such as the large, phase III SWITCH trial (sorafenib/sunitinib cross-over, NCT00732914), RECORD 3 (sunitinib/everolimus cross-over, Novartis, NCT00903175), 404 (sorafenib vs temsirolimus after failure on sunitinib, Pfizer/Wyeth, NCT00474786), and AXIS (sorafenib vs axitinib after failure on one prior systemic therapy, Pfizer, NCT00678392) will answer some of these important questions.
The authors are grateful to 7.4 Limited for providing editorial support on this manuscript, with financial support from Bayer Schering Pharma.
CONFLICT OF INTEREST
Camillo Porta received research grants from Bayer Schering Pharma and Novartis Pharma and acted as a paid consultant or speaker for Bayer Schering Pharma, Pfizer Oncology, Hoffmann-La Roche, Wyeth Pharmaceuticals, Novartis Pharma and GlaxoSmithKline.
Giuseppe Procopio has acted as a paid consultant or speaker for Bayer Schering Pharma, Pfizer Oncology, Hoffmann-La Roche, Wyeth Pharmaceuticals and Sanofi-aventis.
Giacomo Cartenì has acted as a paid consultant or speaker for Pfizer Oncology, Hoffmann-La Roche and Wyeth Pharmaceuticals.
Roberto Sabbatini acted as a paid consultant or speaker for Bayer Schering Pharma, Pfizer Oncology, Hoffmann-La Roche, Wyeth Pharmaceuticals and Novartis Pharma.
Isabella Chiappino has acted as a paid speaker for Pfizer Oncology, Hoffmann-La Roche and Wyeth Pharmaceuticals.
Enzo Maria Ruggeri has acted as a speaker or paid consultant for Bayer Schering Pharma, Hoffmann-La Roche, Pfizer Oncology and Wyeth Pharmaceuticals.
Ilaria Imarisio has acted as a paid speaker for Bayer Schering Pharma.
Elena Verzoni has acted as a paid speaker for Pfizer Oncology.
Chiara Paglino has acted as a paid speaker for Bayer Schering Pharma and Hoffmann-La Roche.
Alessandra Bearz, Giovanni Lo Re, Riccardo Ricotta, Fable Zustovich, Lorenza Landi, Anna Calcagno, Mimma Rizzo, Valentina Guadalupi and Emilio Bajetta have nothing to disclose.