Articles

Axitinib versus sorafenib as second-line treatment for

advanced renal cell carcinoma: overall survival analysis and
updated results from a randomised phase 3 trial
Robert J Motzer, Bernard Escudier, Piotr Tomczak, Thomas E Hutson, M Dror Michaelson, Sylvie Negrier, Stephane Oudard, Martin E Gore,
Jamal Tarazi, Subramanian Hariharan, Connie Chen, Brad Rosbrook, Sinil Kim, Brian I Rini

Summary
Lancet Oncol 2013; 14: 552–62 Background In a phase 3 trial comparing the efficacy and safety of axitinib versus sorafenib as second-line treatment
Published Online for metastatic renal cell carcinoma, patients given axitinib had a longer progression-free survival (PFS). Here, we
April 16, 2013 report overall survival and updated efficacy, quality of life, and safety results.
http://dx.doi.org/10.1016/
S1470-2045(13)70093-7
Methods Eligible patients had clear cell metastatic renal cell carcinoma, progressive disease after one approved
This online publication has
been corrected. The corrected systemic treatment, and an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0–1. 723 patients
version first appeared at were stratified by ECOG PS and previous treatment and randomly allocated (1:1) to receive axitinib (5 mg twice daily;
thelancet.com/oncology on n=361) or sorafenib (400 mg twice daily; n=362). The primary endpoint was PFS assessed by a masked, independent
May 28, 2013
radiology review committee. We assessed patient-reported outcomes using validated questionnaires. Baseline
See Comment page 450
characteristics and development of hypertension on treatment were studied as prognostic factors. Efficacy was
Memorial Sloan-Kettering assessed in the intention-to-treat population, and safety was assessed in patients who received at least one dose of the
Cancer Center, New York, NY,
USA (R J Motzer MD); Medical
study drug. This ongoing trial is registered on ClinicalTrials.gov, number NCT00678392.
Oncology Department, Institut
Gustave Roussy , Villejuif, Findings Median overall survival was 20·1 months (95% CI 16·7–23·4) with axitinib and 19·2 months (17·5–22·3)
France (B Escudier MD); Klinika with sorafenib (hazard ratio [HR] 0·969, 95% CI 0·800–1·174; one-sided p=0·3744). Median investigator-assessed
Onkologii, Uniwersytet
Medyczny, Poznań, Poland
PFS was 8·3 months (95% CI 6·7–9·2) with axitinib and 5·7 months (4·7–6·5) with sorafenib (HR 0·656, 95% CI
(P Tomczak MD); 0·552–0·779; one-sided p<0·0001). Patient-reported outcomes scores were similar in the treatment groups at
Baylor-Sammons and Texas baseline, were maintained during treatment, but decreased at end-of-treatment. Common grade 3 or higher treatment-
Oncology Physician’s
related adverse events were hypertension (60 [17%]), diarrhoea (40 [11%]), and fatigue (37 [10%]) in 359 axitinib-
Association, Sammons Cancer
Center, Dallas, TX, USA treated patients and hand–foot syndrome (61 [17%]), hypertension (43 [12%]), and diarrhoea (27 [8%]) in
(T E Hutson DO); Massachusetts 355 sorafenib-treated patients. In a post-hoc 12-week landmark analysis, median overall survival was longer in patients
General Hospital Cancer Center, with a diastolic blood pressure of 90 mm Hg or greater than in those with a diastolic blood pressure of less than
Boston, MA, USA
90 mm Hg: 20·7 months (95% CI 18·4–24·6) versus 12·9 months (10·1–20·4) in the axitinib group (p=0·0116), and
(M D Michaelson MD); Service
d’Oncologie Médicale, Centre 20·2 months (17·1–32·0) versus 14·8 months (12·0–17·7) in the sorafenib group (one-sided p=0·0020).
Léon Bérard, Lyon, France
(Prof S Negrier MD); Georges Interpretation Although overall survival, a secondary endpoint for the study, did not differ between the two groups,
Pompidou European Hospital,
investigator-assessed PFS remained longer in the axitinib group compared with the sorafenib group. These results
University of Paris Descartes,
Paris, France establish axitinib as a second-line treatment option for patients with metastatic renal cell carcinoma.
(Prof S Oudard MD); Royal
Marsden Hospital, London, UK Funding Pfizer Inc.
(Prof M E Gore MD); Pfizer
Oncology, San Diego, CA, USA
(J Tarazi MD, B Rosbrook MS, Introduction [HR] 0·665, 95% CI 0·544–0·812; one-sided p<0·0001).5
S Kim MD); Pfizer Inc, New York, Axitinib, a potent and selective second-generation in- On the basis of these results, axitinib was approved in
NY, USA (S Hariharan MD); hibitor of VEGF receptors 1, 2, and 3,1 has showed the USA,6 Japan, the European Union, and other
Pfizer Inc, Global Outcomes
Research, New York, NY, USA
activity in second-line treatment of metastatic renal cell countries for the second-line treatment of advanced
(C Chen PharmD); and Cleveland carcinoma in phase 2 studies.2–4 The randomised phase 3 renal cell carcinoma. Overall survival, a secondary
Clinic Taussig Cancer Institute, Axitinib Second-line (AXIS) trial compared efficacy endpoint, was not mature (only 53% of the prespecified
Cleveland, OH, USA (B I Rini MD) and safety of axitinib versus sorafenib, another VEGF overall survival events had occurred) as of the Aug 31,
Correspondence to: receptor tyrosine kinase inhibitor, in 723 patients with 2010, cutoff date for primary endpoint analysis and was
Dr Robert J Motzer, Memorial
metastatic renal cell carcinoma who had disease pro- therefore not reported at that time.5
Sloan-Kettering Cancer Center,
1275 York Ave, New York, gression after one previous systemic treatment: either We report overall survival results from the ongoing
NY 10021, USA sunitinib, cytokines, bevacizumab plus interferon alfa, AXIS trial, with a Nov 1, 2011, cutoff date, as well as
motzerr@mskcc.org or temsirolimus.5 Median progression-free survival updated safety and quality-of-life results and PFS and
(PFS) in the overall population by independent radiology objective response rate by investigator assessment.
review committee assessment, the primary endpoint of Additionally, we investigated patients’ baseline charac-
this trial, was longer with axitinib than it was with teristics as prognostic factors for overall survival, and we
sorafenib: 6·7 months versus 4·7 months (hazard ratio did retrospective analyses to assess whether overall

552 www.thelancet.com/oncology Vol 14 May 2013

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

survival and PFS are associated with the development of in 24 h; axitinib was resumed at one lower dose level
hypertension during treatment. when adverse events resolved to grade 2 or lower, blood
pressure was 150/100 mm Hg or lower, or less than 2 g
Methods proteinuria in 24 h was present. Patients receiving
Study design and patients axitinib who did not have treatment-related adverse
AXIS was a multicentre, phase 3 open-label randomised events above grade 2 for at least 2 weeks and had blood
trial of axitinib versus sorafenib in patients with advanced pressure 150/90 mm Hg or lower without antihyper-
renal cell carcinoma who had failed one previous systemic tensive drugs could have their dose increased to 7 mg
treatment. Study design and eligibility criteria, which are twice daily and then to a maximum of 10 mg twice daily.
described elswhere,5 included patients with renal cell The sorafenib dose was interrupted in patients with
carcinoma with clear cell histology, measurable disease,7 grade 3 toxicities and resumed at the same dose when
Eastern Cooperative Oncology Group performance status the toxicity was grade 1 or lower (grade 2 or lower for
(ECOG PS) 0–1, blood pressure 140/90 mm Hg or lower haematological toxicities). The sorafenib dose was
(antihypertensive drugs were allowed), and progressive reduced to 400 mg once daily and then to 400 mg every
disease after previous treatment with sunitinib, other day in patients with recurrent grade 3 toxicities. In
bevacizumab plus interferon alfa, temsirolimus, or patients with grade 4 toxicities or grade 3 skin toxicities,
cytokines. We recruited patients from hospitals and the sorafenib dose was interrupted and then resumed at
outpatient clinics in 22 countries. one lower dose level when the toxicity was grade 1 or
This study was done in accordance with the Declaration lower (grade 2 or lower for haematological toxicities). In
of Helsinki, the International Conference on Harmon- patients with a grade 2 skin toxicity that did not improve
isation Guidelines on Good Clinical Practice, and within 7 days or a second or third occurrence of a grade
applicable local regulatory requirements and laws. All 2 skin toxicity, the sorafenib dose was interrupted and
participants provided written informed consent. The then resumed at one lower dose level when the toxicity
protocol, amendments, and informed consent forms were was grade 1 or lower. Crossover between the two drugs
approved by an institutional review board or independent on study was not allowed.
ethics committee at each investigational centre. Safety was assessed throughout the study period,
and severity of adverse events was graded according
Randomisation and masking to CTCAE.8 Physical examinations, laboratory assess-
As previously described,5 patients were stratified by ments, and blood pressure measurements were done at
ECOG PS and previous treatment, and were then screening, week 2, week 4, and every 4 weeks thereafter.
randomly allocated in a one-to-one ratio to receive Tumour assessments were done at screening, week 6,
axitinib or sorafenib. Randomised lists were computer-
generated by an independent randomisation group with
a permuted block design of size four within each stratum. 723 patients randomised
Patients and investigators were not masked to study
treatment. The radiologists who participated in the
independent review committee assessment of PFS (the 361 assigned to axitinib 362 assigned to sorafenib
primary endpoint) were masked to group assignment.
2 did not receive axitinib 7 did not receive sorafenib
Procedures
Axitinib was given orally at a starting dose of 5 mg twice
359 received axitinib 355 received sorafenib
daily and sorafenib was given orally at a starting dose of
400 mg twice daily. Dose adjustments are described
elsewhere.5 The axitinib dose was reduced to 3 mg twice 318 discontinued axitinib 325 discontinued sorafenib
daily and then further to 2 mg twice daily in patients 240 disease progression/relapse 226 disease progression/relapse
27 adverse events 45 adverse events
with grade 3 non-haematological treatment-related ad- 17 deaths 17 deaths
verse events according to the Common Terminology 13 refusal of treatment for reason 12 refusal of treatment for reason
other than adverse event other than adverse events
Criteria for Adverse Events (CTCAE; version 3.0),8 or in 12 global deterioration in health 8 global deterioration in health
patients with two readings of systolic blood pressure status status
4 protocol violations 3 protocol violations
150 mm Hg or higher or diastolic blood pressure 1 lost to follow-up 3 lost to follow-up
100 mm Hg or higher who were receiving maximal 4 other reasons 11 other reasons
antihypertensive treatment. The axitinib dose was inter-
rupted in patients with grade 4 non-haematological
361 in PFS and OS analyses 362 in PFS and OS analyses
treatment-related adverse events or grade 4 haemato- 359 in safety analysis 355 in safety analysis
logical adverse events, two readings of systolic blood
pressure 160 mm Hg or higher or diastolic blood pres- Figure 1: Trial profile
sure 105 mm Hg or higher, or 2 g or higher proteinuria PFS=progression-free survival. OS=overall survival.

www.thelancet.com/oncology Vol 14 May 2013 553

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

week 12, and every 8 weeks thereafter according to Cancer Therapy Kidney Symptom Index (FKSI-15) ques-
Response Evaluation Criteria in Solid Tumors (RECIST, tionnaire,9 and the FKSI–Disease-Related Symptoms
version 1.0).7 We assessed patient-reported outcomes at (FKSI-DRS) subscale,10 which measures metastatic renal
baseline, day 1 of every 4 week cycle,5 and end of treat- cell carcinoma-associated symptoms and health status.
ment using the validated Functional Assessment of Higher scores represent fewer symptoms or better health
status. A minimally important difference was 3–5 points
for the FKSI-15 scale (range 0–60) and 2–3 points for the
A
Median OS, months (95% CI) FKSI-DRS (range 0–36) subscale.9,10
100
90
Axitinib 20·1 (16·7–23·4) An independent radiology review committee assess-
Sorafenib 19·2 (17·5–22·3)
ment of PFS, the primary endpoint, concluded on June 3,
80
HR 0·969 (95% CI 0·800–1·174) 2011, at which time PFS results were much the same as
70 p=0·3744, one-sided stratified log-rank test those previously reported (Aug 31, 2010, cutoff) by Rini
Overall survival (%)

60
and colleagues.5 PFS by investigator assessment and
50
safety data were updated for this report and correspond
40
to the final overall survival analysis data cutoff (Nov 1,
30 2011). Secondary endpoints included overall survival,
20 investigator-assessed PFS, objective response rate, and
10 safety.
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
Number at risk Statistical analysis
Axitinib 361 351 326 302 273 252 237 214 193 178 152 127 86 70 45 36 14 5 0 Sample size required for the primary endpoint and
Sorafenib 362 340 318 298 278 259 240 218 202 181 145 118 88 62 41 26 9 4 0 statistical methods were described previously.5 The
B sample size also allowed for the assessment of overall
100 Median OS, months (95% CI) survival as a secondary endpoint, which required
Axitinib 15·2 (12·8–18·3)
90
Sorafenib 16·5 (13·7–19·2)
417 events for an overall one-sided significance level of
80 0·025 with 80% power to detect a 31·7% improvement in
HR 0·997 (95% CI 0·782–1·270) median overall survival from 18 months with sorafenib11
70
Overall survival (%)

p=0·4902, one-sided stratified log-rank test

60 to 23·7 months with axitinib. Efficacy was assessed in the
50 intention-to-treat population (n=723). Safety was assessed
40
in patients who received at least one dose of study drug
30
(n=714). The data cutoff date was Nov 1, 2011. We
summarised time-to-event endpoints using Kaplan-
20
Meier methods. Unstratified and stratified log-rank tests
10
(one-sided, α=0·025) were used to compare overall
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 survival between treatment groups; we estimated HR
Number at risk and 95% CI for overall survival.
Axitinib 194 187 172 157 137 123 114 98 88 80 67 56 38 32 22 19 10 4 0
We assessed a prespecified set of baseline charac-
Sorafenib 195 183 167 154 140 129 117 106 97 88 66 56 39 28 19 12 5 2 0
teristics for their relevance as prognostic factors of overall
C survival. We used a Cox proportional hazards model,
100
90
including treatment as a stratum, to do a univariate
80
analysis of overall survival. Variables significant at a 10%
level were then used to construct the multivariable model
70
for overall survival. We applied a backward elimination
Overall survival (%)

60
process to these variables, including treatment, using a
50
5% significance level to stay in the model, to identify the
40
Median OS, months (95% CI) final set of relevant factors. We explored treatment-by-
30 Axitinib 29·4 (24·5–NE)
Sorafenib 27·8 (23·1–34·5) factor interactions only for the set of factors included in
20
the final model. Estimated HR and two-sided 95% CI and
10 HR 0·813 (95% CI 0·555–1·191)
p=0·1435, one-sided stratified log-rank test
p values were calculated for each relevant prognostic
0 factor. For patients previously treated with sunitinib,
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
time to progression on previous sunitinib treatment was
Time (months)
Number at risk categorised into quartiles—overall survival, based on
Axitinib 126 124 120 114 108 104 101 94 88 83 74 62 44 36 21 15 4 1 0 time of randomisation in the AXIS study, was then
Sorafenib 125 117 115 111 108 103 99 91 86 77 68 53 41 27 17 13 4 2 0
summarised within each quartile corresponding to the
time to progression on first-line sunitinib.
Figure 2: Kaplan-Meier estimates of overall survival in all patients (A), patients previously treated with
sunitinib (B), and patients previously treated with cytokines (C) Similarly, the patient-reported outcomes data analyses
HR=hazard ratio. NE=not estimable. OS=overall survival. were done on the intention-to-treat population with the

554 www.thelancet.com/oncology Vol 14 May 2013

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

data cutoff date of Nov 1, 2011. We used descriptive interferon alfa; and 6·4 months (1·6–18·7) versus
statistics to summarise scores by treatment group, and 10·1 months (<0·1–20·3) for previous temsirolimus. As
did a comparison of post-treatment scores using a of Nov 1, 2011, median duration of treatment was
repeated-measures mixed-effects model. 8·2 months (range <0·1–33·4) with axitinib and
In a post-hoc exploratory analysis, development of 5·2 months (0·2–34·1) with sorafenib. Dose reductions
hypertension during treatment with axitinib or were reported in 121 (34%) of 359 axitinib-treated
sorafenib was associated with overall survival and PFS. patients and 192 (54%) of 355 sorafenib-treated patients.
Patients were grouped on the basis of in-clinic blood The axitinib dose was titrated above 5 mg twice daily in
pressure measurements: at least one diastolic blood 136 (38%) patients.
pressure measurement 90 mm Hg or higher versus 425 deaths occurred, 211 in the axitinib group and 214
diastolic blood pressure lower than 90 mm Hg, and at in the sorafenib group. We recorded no difference in
least one systolic blood pressure measurement 140 mm median overall survival between the two groups overall or
Hg or higher versus systolic blood pressure lower than when stratified by either previous sunitinib or previous
140 mm Hg. We did landmark analyses12 using the cytokines (figure 2), nor did we detect a between-group
maximum diastolic blood pressure and systolic blood difference when stratifying by previous bevacizumab plus
pressure achieved by weeks 8 and 12 to group patients, interferon alfa (14·7 months [95% CI 9·2–20·0] with
and we assessed overall survival and PFS from that axitinib vs 19·8 months [13·1–not estimable] with
point forward. All patients who died or had progression sorafenib; HR 1·825, 95% CI 0·942–3·535; one-sided
of disease, according to the particular endpoint, before p=0·9648) or by previous temsirolimus (14·0 months
the 8-week or 12-week landmark were excluded from [3·8–not estimable] vs 8·5 months [5·7–13·5]; HR 0·459,
these analyses. We estimated median overall survival 0·165–1·278; one-sided p=0·0638).
and PFS using Kaplan-Meier methods with time from As of Nov 1, 2011, median PFS by investigator
the 8-week or 12-week landmark. We used one-sided assessment was longer in the axitinib group (8·3 months
unstratified log-rank tests to compare overall survival [95% CI 6·7–9·2]) than it was in the sorafenib group
and PFS between blood pressure groups. Development (5·7 months [4·7–6·5]; HR 0·656, 95% CI 0·552–0·779;
of diastolic blood pressure 90 mm Hg or higher or
systolic blood pressure 140 mm Hg or higher during
Axitinib (n=359) Sorafenib (n=355)
treatment was included with the prespecified set of
baseline characteristics for additional multivariable All grades Grade ≥3 All grades Grade ≥3
analyses of overall survival. We used SAS (version 9.2) Diarrhoea 193 (54%) 40 (11%) 185 (52%) 27 (8%)
for all statistical analyses. Hypertension 149 (42%) 60 (17%) 107 (30%) 43 (12%)
This trial is registered on ClinicalTrials.gov, number Fatigue 133 (37%) 37 (10%) 98 (28%) 14 (4%)
NCT00678392. Decreased appetite 113 (31%) 15 (4%) 94 (26%) 7 (2%)
Nausea 109 (30%) 6 (2%) 67 (19%) 3 (1%)
Role of the funding source Dysphonia 102 (28%) 0 42 (12%) 0
The study sponsor was involved in study design and Hand–foot syndrome 100 (28%) 20 (6%) 182 (51%) 61 (17%)
collection and analyses of data. All authors approved the Hypothyroidism 72 (20%) 1 (<0·5%) 29 (8%) 0
final content of this report. The corresponding author Weight decreased 70 (19%) 12 (3%) 63 (18%) 9 (3%)
had full access to all data and final responsibility to Asthenia 66 (18%) 15 (4%) 47 (13%) 8 (2%)
submit the report for publication. Vomiting 63 (18%) 5 (1%) 47 (13%) 0
Mucosal inflammation 58 (16%) 5 (1%) 44 (12%) 3 (1%)
Results Stomatitis 55 (15%) 5 (1%) 44 (12%) 1 (<0·5%)
Between Sept 15, 2008, and July 23, 2010, 723 patients Rash 47 (13%) 1 (<0·5%) 110 (31%) 13 (4%)
were randomly allocated to receive axitinib (n=361) or Constipation 45 (13%) 1 (<0·5%) 47 (13%) 1 (<0·5%)
sorafenib (n=362; figure 1). Patients’ demographics and Proteinuria 45 (13%) 11 (3%) 27 (8%) 4 (1%)
baseline characteristics were much the same between Dysgeusia 41 (11%) 0 30 (8%) 0
treatment groups.5 Of the 723 patients, 658 (91%) had
Headache 39 (11%) 3 (1%) 25 (7%) 0
prior nephrectomy; first-line treatment before study
Arthralgia 36 (10%) 3 (1%) 18 (5%) 1 (<0·5%)
entry was with sunitinib (n=389 [54%]), cytokines (n=251
Dry skin 36 (10%) 0 36 (10%) 0
[35%]), bevacizumab plus interferon alfa (n=59 [8%]), or
Alopecia 16 (4%) 0 117 (33%) 0
temsirolimus (n=24 [3%]). Median duration of first-line
Pruritus 22 (6%) 0 46 (13%) 0
treatment in the axitinib versus the sorafenib groups
Pain in extremity 32 (9%) 1 (<0·5%) 36 (10%) 3 (1%)
was: 9·5 months (range 0·6–108·3) versus 10·1 months
Erythema 10 (3%) 0 36 (10%) 1 (<0·5%)
(0·5–57·3) for previous sunitinib; 6·0 months
(<0·1–244·1) versus 6·6 months (<0·1–127·9) for Data are n (%). *Reported in ≥10% of patients in either group.
previous cytokines; 5·7 months (1·8–74·0) versus
Table 1: Treatment-related adverse events*
6·8 months (0·9–17·5) for previous bevacizumab plus

www.thelancet.com/oncology Vol 14 May 2013 555

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

one-sided p<0·0001). When stratified by previous investigator-assessed objective response than those
treatment, patients previously treated with either treated with sorafenib (82 [23%] of 361 patients in the
sunitinib or cytokines had a longer median PFS in the axitinib group vs 45 [12%] of 362 patients in the sorafenib
axitinib group than in the sorafenib group: 6·5 months group; one-sided p=0·0001).
(95% CI 5·7–7·9) versus 4·4 months (2·9–4·7) with Treatment-related adverse events occurring in more
previous sunitinib (HR 0·719, 95% CI 0·572–0·903; than 30% of 359 axitinib-treated patients were
one-sided p=0·0022), and 12·2 months (95% CI diarrhoea, hypertension, fatigue, decreased appetite,
10·2–15·5) versus 8·2 months (6·6–9·5) with previous and nausea (table 1). Treatment-related adverse events
cytokines (0·505, 0·373–0·684; one-sided p<0·0001). We occurring in more than 30% of 355 sorafenib-treated
detected no difference in PFS between the axitinib and patients were diarrhoea, hand–foot syndrome, alopecia,
sorafenib groups when assessing the small subsets of rash, and hypertension (table 1). The most common
patients who had received previous bevacizumab plus grade 3 or higher treatment-related adverse events were
interferon alfa (8·3 months [95% CI 2·8–10·5] vs hypertension, diarrhoea, and fatigue in the axitinib
4·5 months [3·0–6·5]; HR 0·815, 95% CI 0·429–1·550; group, and hand–foot syndrome, hypertension, and
one-sided p=0·2656) or those who had received previous diarrhoea in the sorafenib group (table 1). Treatment-
temsirolimus (2·6 months [1·5–17·1] vs 5·7 months related hypertension, nausea, dysphonia, and
[2·6–8·3]; HR 1·210, 95% CI 0·433–3·382; one-sided hypothyroidism were more common with axitinib than
p=0·6342). More patients receiving axitinib had an they were with sorafenib (greater than 10% difference
between treatment groups), whereas hand–foot syn-
drome, alopecia, and rash were more common with
Two-sided Hazard ratio sorafenib.
p value (95% CI)
Completion rates for the FKSI-15 and FKSI-DRS
Previous treatment questionnaires were 86% (FKSI-15: 312 of 361 patients in
Bevacizumab vs sunitinib 0·8664 0·971 (0·688–1·371) the axitinib group and 306 of 355 eligible patients in the
Cytokine vs sunitinib <0·0001 0·497 (0·397–0·622) sorafenib group) or higher at baseline and remained
Temsirolimus vs sunitinib 0·1326 1·444 (0·895–2·330) high (90% or higher) throughout most treatment cycles
ECOG PS (1 vs 0) <0·0001 2·022 (1·668–2·450) (data not shown). There were fewer responses to
Age (≥65 years vs <65 years) 0·2153 1·134 (0·930–1·383) questionnaires in the sorafenib group than in the axitinib
Sex (male vs female) 0·9729 1·004 (0·813–1·239) group, because patients treated with sorafenib ended
Race (non-white vs white) 0·1638 0·849 (0·675–1·069) treatment earlier, mainly due to disease progression: on
MSKCC risk score day 1 of cycles 10, 20, and 30 of treatment, there were
Intermediate vs favourable <0·0001 1·858 (1·408–2·452) 171 (47%), 73 (20%), and 18 (5%) of 361 patients in the
Poor vs favourable <0·0001 3·921 (2·988–5·146)
NA vs favourable <0·0001 3·733 (2·106–6·618) Two-sided Hazard ratio
Time from diagnosis to treatment <0·0001 1·998 (1·630–2·448) p value (95% CI)
on AXIS study (<1 year vs ≥1 year) Previous treatment
Metastatic sites (>1 vs 1) <0·0001 2·451 (1·690–3·554) Bevacizumab vs sunitinib 0·2921 0·821 (0·568–1·185)
Sites of metastases Cytokine vs sunitinib <0·0001 0·503 (0·395–0·641)
Liver (yes vs no) <0·0001 1·727 (1·417–2·105) Temsirolimus vs sunitinib 0·8046 1·065 (0·647–1·754)
Lung (yes vs no) 0·1774 0·854 (0·680–1·074) ECOG PS (1 vs 0) 0·0005 1·452 (1·177–1·790)
Bone (yes vs no) <0·0001 1·615 (1·327–1·967) Time from diagnosis to treatment on 0·0001 1·554 (1·240–1·947)
Lymph nodes (yes vs no) 0·2058 1·134 (0·933–1·378) AXIS study (<1 vs ≥1 year)
Previous nephrectomy (no vs yes) <0·0001 2·165 (1·617–2·901) Metastatic sites (>1 vs 1) 0·0057 1·744 (1·176–2·588)
Previous radiation (yes vs no) 0·0207 1·308 (1·042–1·643) Liver metastases (yes vs no) 0·0204 1·298 (1·041–1·618)
Corrected calcium (>10 mg/dL vs <0·0001 3·815 (2·832–5·139) Bone metastases (yes vs no) 0·0056 1·357 (1·093–1·685)
≤10 mg/dL) Corrected calcium (>10 mg/dL vs <0·0001 2·743 (1·971–3·817)
Alkaline phosphatase (>ULN vs ≤ULN) <0·0001 2·243 (1·799–2·796) ≤10 mg/dL)
LDH (>1·5 × ULN vs ≤1·5 × ULN) <0·0001 4·264 (2·938–6·188) Alkaline phosphatase (>ULN vs ≤ULN) 0·0059 1·412 (1·104–1·805)
Haemoglobin (<LLN vs ≥LLN ) <0·0001 2·428 (1·981–2·976) LDH (>1·5 × ULN vs ≤1·5 × ULN) <0·0001 2·677 (1·764–4·062)
Platelets (>ULN vs ≤ULN) <0·0001 2·325 (1·718–3·148) Haemoglobin (<LLN vs ≥LLN ) <0·0001 1·689 (1·352–2·111)
Neutrophils (>ULN vs ≤ULN) <0·0001 3·440 (2·518–4·700) Neutrophils (>ULN vs ≤ULN) 0·0048 1·688 (1·173–2·428)

ECOG PS=Eastern Cooperative Oncology Group performance status. LDH=lactate ECOG PS=Eastern Cooperative Oncology Group performance status; LDH=lactate
dehydrogenase. LLN=lower limit of normal. MSKCC=Memorial Sloan-Kettering dehydrogenase. LLN=lower limit of normal. ULN=upper limit of normal. *Final
Cancer Center. ULN=upper limit of normal. model constructed by backward elimination process with a 5% level of selection.

Table 2: Univariate analysis for all prespecified baseline prognostic Table 3: Multivariable analysis for baseline prognostic factors for overall
factors for overall survival survival*

556 www.thelancet.com/oncology Vol 14 May 2013

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

100 Median OS, months (95% CI)

Favourable 31·6 (27·2–NE)
90 Intermediate 21·4 (18·9–24·0)
Poor 10·6 (9·1–12·1)

80

70
Overall survival (%)

60

50

40

30

20

10

0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36
Time (months)
Number at risk
Favourable 201 198 197 190 188 180 173 164 155 144 116 99 73 58 39 28 12 5 0
Intermediate 264 261 247 235 214 202 194 179 157 145 121 94 69 48 33 26 9 4 0
Poor 238 217 186 163 139 120 101 80 75 64 55 48 30 25 14 8 2 0 0

Figure 3: Kaplan-Meier estimates of overall survival by MSKCC risk score for the overall trial population
MSKCC=Memorial Sloan-Kettering Cancer Center. NE=not estimable. OS=overall survival.

axitinib group, respectively, remaining on treatment, and

Axitinib* Sorafenib†
112 (32%), 51 (14%), and 11 (3%) of 355 eligible patients in
the sorafenib group, respectively. At baseline, mean n Median overall n Median overall
survival in survival in
FKSI-15 scores were 43·2 (SD 8·4) in axitinib group and months (95% CI) months (95% CI)
43·3 (8·2) in the sorafenib group, and FKSI-DRS scores
≤25th percentile (≤5·6 months) 46 10·9 (7·7–17·3) 55 12·4 (9·1–16·5)
were 28·9 (5·2) in the axitinib group and 29·0 (5·2) in
>25th percentile (>5·6 months) to 50th percentile 51 10·8 (7·6–13·9) 43 18·5 (11·1–22·8)
the sorafenib group, and scores were much the same (10·0 months)
between the two groups throughout treatment (data not
>50th percentile (>10·0 months) to 75th percentile 50 23·0 (17·8–28·3) 48 19·2 (14·9–29·0)
shown). Scores remained similar to baseline in both (17·9 months)
groups during treatment, indicating that patients’ quality >75th percentile (>17·9 months) to 100th percentile 45 16·6 (12·5–21·7) 48 18·3 (13·1–25·0)
of life or metastatic renal cell carcinoma-associated (49·4 months)
symptoms did not deteriorate when receiving axitinib or
*Two patients excluded because of protocol deviations. †One patient excluded because of missing start date of
sorafenib. At the end-of-treatment assessment, at which previous treatment.
time patients were coming off treatment mainly due to
tumour progression,5 scores decreased in both treatment Table 4: Overall survival by time to progression on previous sunitinib treatment

groups, indicating a worsening of quality of life or

symptoms at the time of progression. The mean end-of- Center (MSKCC) risk score,13 time from diagnosis to
treatment FKSI-15 score was 38·9 (SD 9·5) with axitinib treatment, number and sites (liver and bone) of metas-
and 39·1 (8·9) with sorafenib, and the mean end-of- tases, prior nephrectomy, prior radiation, concentrations
treatment FKSI-DRS score was 26·6 (SD 5·8) with of corrected calcium, alkaline phosphatase, lactate
axitinib and 26·9 (5·6) with sorafenib. A repeated dehydrogenase, haemoglobin, platelets, and neutrophils
measures mixed-effects model showed no statistical (table 2). In the multivariable analysis, prognostic factors
difference between treatment groups; post-treatment for individual characteristics associated with shorter
least squares means in the axitinib versus sorafenib overall survival were: type of previous treatment (cytokine
groups were 42·2 versus 41·8 (p=0·4922) with FKSI-15, vs sunitinib), ECOG PS of 1, less than 1 year from initial
and 28·6 versus 28·4 (p=0·6741) with FKSI-DRS. diagnosis to treatment on AXIS study, more than one
Univariate analyses showed the following baseline metastatic site, hepatic metastases, bone metastases,
characteristics to be prognostic factors for overall haemoglobin less than lower limit of normal, corrected
survival: type of previous treatment (cytokine vs calcium above 10 mg/dL, lactate dehydrogenase greater
sunitinib), ECOG PS, Memorial Sloan-Kettering Cancer than 1·5 × upper limit of normal, and alkaline

www.thelancet.com/oncology Vol 14 May 2013 557

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

Axitinib Sorafenib
8-week landmark 12-week landmark 8-week landmark 12-week landmark
n Median survival Hazard ratio n Median survival Hazard ratio n Median survival Hazard ratio n Median survival Hazard ratio
in months (95% CI); in months (95% CI); in months (95% CI); in months (95% CI);
(95% CI) p value* (95% CI) p value* (95% CI) p value* (95% CI) p value*
Overall survival
Diastolic blood 0·775 0·716 0·724 0·657
pressure (0·588–1·020); (0·537–0·957); (0·547–0·960); (0·492–0·876);
p=0·0342 p=0·0116 p=0·0120 p=0·0020
≥90 mm Hg 189 21·3 (16·9–25·3) 203 20·7 (18·4–24·6) 182 21·1 (17·2–27·1) 187 20·2 (17·1–32·0)
<90 mm Hg 161 13·9 (11·1–21·1) 132 12·9 (10·1–20·4) 154 15·8 (13·2–18·7) 141 14·8 (12·0–17·7)
Systolic blood 0·781 0·753 0·726 0·715
pressure (0·590–1·034); (0·556–1·019); (0·542–0·973); (0·527–0·970);
p=0·0417 p=0·0329 p=0·0155 p=0·0151
≥140 mm Hg 217 20·7 (16·1–24·3) 231 20·7 (16·4–24·0) 225 20·8 (17·4–23·4) 230 19·9 (16·8–22·9)
<140 mm Hg 133 15·7 (12·5–21·1) 104 17·0 (11·8–20·7) 111 14·8 (10·8–17·7) 98 14·8(10·2–17·5)
Progression-free survival
Diastolic blood 1·009 1·028 0·922 0·952
pressure (0·769–1·324); (0·754–1·401); (0·694–1·225); (0·694–1·307);
p=0·5230 p=0·5644 p=0·2846 p=0·3778
≥90 mm Hg 159 8·1 (6·5–10·0) 160 8·9 (7·1–9·2) 138 4·8 (4·5–6·3) 124 5·2 (3·6–5·6)
<90 mm Hg 121 8·3 (6·3–10·7) 80 9·0 (5·9–11·0) 105 4·7 (4·5–6·6) 74 5·4 (3·7–7·1)
Systolic blood 1·148 1·064 0·897 0·960
pressure (0·866–1·521); (0·773–1·466); (0·666–1·208); (0·682–1·351);
p=0·8303 p=0·6456 p=0·2322 p=0·4027
≥140 mm Hg 179 8·1 (6·3–10·0) 168 8·9 (7·0–9·6) 167 4·8 (4·5–6·5) 145 5·3 (3·7–6·2)
<140 mm Hg 101 8·3 (6·5–10·7) 72 7·9 (7·1–12·6) 76 4·8 (4·4–6·4) 53 5·4 (3·6–7·1)

* p values from one-sided unstratified log-rank test.

Table 5: Landmark analyses of overall and progression-free survival in patients by diastolic and systolic blood pressures subgroups

100th percentile) while receiving first-line sunitinib than

Axitinib Sorafenib
(n=320) (n=332)
those whose time to progression was less than or equal to
the median (0–50th percentile). In the sorafenib group,
Any subsequent systemic 174 (54%) 188 (57%)
treatment
median overall survival seemed longer in patients in the
≥2 subsequent systemic treatments 75 (23%) 83 (25%)
greater than 50th to 75th quartile for time to progression
mTOR inhibitors 125 (39%) 137 (41%)
while receiving prior sunitinib versus the other three
quartiles (table 4).
VEGF or VEGF receptor inhibitors 104 (33%) 106 (32%)
Landmark analyses showed that patients who de-
Cytokines 12 (4%) 20 (6%)
veloped diastolic blood pressure 90 mm Hg or higher
Chemotherapy 11 (3%) 14 (4%)
within the first 8 weeks or 12 weeks of randomisation
Other 12 (4%) 15 (5%)
had longer median overall survival than those with
Data are n (%). diastolic blood pressure lower than 90 mm Hg (table 5).
Similarly, patients who developed systolic blood
Table 6: Systemic treatment after discontinuation of study drug
pressure 140 mm Hg or higher during treatment had
longer median overall survival than those with systolic
phosphatase or neutrophils greater than upper limit of blood pressure below 140 mm Hg (table 5). Multivariable
normal (table 3). Kaplan-Meier estimates of overall analyses including baseline characteristics showed that
survival by MSKCC risk score are shown in figure 3. development of diastolic blood pressure 90 mm Hg or
The most common first-line treatment for patients higher or systolic blood pressure 140 mm Hg or higher
entering this second-line study was sunitinib. We during treatment were independent predictors of overall
explored the relation between time to progression on survival: the HR for diastolic blood pressure 90 mm Hg
first-line sunitinib treatment and overall survival from or higher versus diastolic blood pressure below
randomisation to second-line axitinib or sorafenib 90 mm Hg was 0·627 (95% CI 0·507–0·776; two-sided
treatment (table 4). In the axitinib group, median overall p<0·0001); the HR for systolic blood pressure 140 mm
survival seemed longer in patients whose time to Hg or higher versus systolic blood pressure below
progression was greater than the median (>50th to 140 mm Hg was 0·490 (0·391–0·613; two-sided

558 www.thelancet.com/oncology Vol 14 May 2013

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

p<0·0001). By contrast for these findings for overall study treatment. Also, as described by Broglio and
survival, PFS was similar in the diastolic blood pressure colleagues15 and Negrier and colleagues,16 being able to
and systolic blood pressure subgroups in both treatment show a survival advantage is especially challenging when
groups in the 8-week and 12-week landmark analyses survival post-progression is long, that is, 12 months or
(table 5). longer, because the variability added during a long
174 (54%) of 320 patients who discontinued axitinib on survival post-progression might dilute the ability to
study or who were randomly allocated to the axitinib detect a statistically significant difference in overall
group but did not receive study drug, and 188 (57%) of survival. The statistical design calling for an 80% power
332 patients who discontinued sorafenib on study or who to detect a 31·7% improvement in median overall
were randomly allocated to the sorafenib group but did survival with axitinib would require a large benefit to be
not receive study drug received additional systemic statistically significant, as well. Also, there might be
treatment after discontinuation. We detected no major discordance between assessment of PFS by study-
differences in type of subsequent treatment (table 6). Of defined RECIST criteria,7 and that used in decision
patients who discontinued axitinib, 51 (16%) of making by a treating physician in clinical practice. As
320 received subsequent sorafenib and one (<0·5%) per the study protocol, patients were allowed to continue
received subsequent axitinib. Of patients who discon- treatment post-progression as long as the sum of longest
tinued sorafenib, no patients received subsequent axitinib diameters of measurable lesions was equal to or less
and 26 (8%) of 332 received subsequent sorafenib. than at baseline and if the patient had clinical benefit
from axitinib or sorafenib treatment and no alternative
Discussion treatment was available. Therefore, RECIST-defined PFS
Although median overall survival, assessed as a might be another factor affecting the discordance
secondary endpoint, did not differ between the two between PFS advantage and absence of overall survival
groups, median PFS by investigator assessment advantage seen in this study.
remained longer in the axitinib group than it was in the Interpretation of outcome data for overall survival (and
sorafenib group. Both tyrosine kinase inhibitors had PFS) should be viewed in the context of this trial’s use of
favourable safety profiles in chronic outpatient dosing. an active comparator. AXIS is the first phase 3 renal cell
We detected differences, however, in the relative inci- carcinoma trial to compare one targeted agent with
dence of some treatment-related adverse events. For another active targeted agent approved by the US Food
example, rash, hand–foot skin reaction, and alopecia and Drug Administration for the treatment of patients
were more common with sorafenib, whereas hyper- with advanced renal cell carcinoma. Sorafenib, the first
tension, nausea, hypothyroidism, and dysphonia were tyrosine kinase inhibitor to receive regulatory approval
more common with axitinib. Patient-reported outcomes for advanced renal cell carcinoma, was chosen as the
as assessed by the FKSI-15 and FKSI-DRS were much the comparator for this trial on the basis of its PFS superiority
same during treatment with axitinib or sorafenib. This compared with placebo in another phase 3 trial11 and its
finding was not unexpected because baseline scores were widespread use in oncology practice as a second-line
similar to those of the general population in the USA.14 therapy for metastatic renal cell carcinoma. Everolimus,
Therefore, an improvement in symptoms or quality of an orally administered mTOR inhibitor, could have been
life would not be expected. Because patients treated with considered as a comparator, but was not available as an
axitinib had longer PFS than did those treated with approved agent globally at the time of study initiation.
sorafenib, patients in the axitinib group seemed to have AXIS sets a new standard for phase 3 trials assessing
longer symptom maintenance or control, because novel therapeutics against other active targeted agents in
FKSI-15 and FKSI-DRS scores worsened at the end of patients with metastatic renal cell carcinoma, and future
treatment when patients were discontinuing treatment, trials should continue to use active comparators.
mainly because of disease progression.5 Eligibility for AXIS included patients who had pro-
The choice of agent is an important step in individualising gressed on one of four first-line treatments—sunitinib,
treatment for patients with metastatic renal cell carcinoma, bevacizumab plus interferon alfa, temsirolimus, or
and many factors contribute to treatment selection, cytokines. Although patients previously treated with
including safety profile of the drug and a patient’s temsirolimus comprised a small subset of the trial
tolerability to previous therapy, comorbidities, and patient- population, we know of no other study to report
specific circumstances. Results of this study substantiate tyrosine kinase inhibitor treatment after first-line
axitinib as a treatment option in previously treated patients temsirolimus, an mTOR inhibitor that showed overall
with metastatic renal cell carcinoma (panel), and showed survival benefit in patients with poor prognostic
axitinib to have better efficacy in PFS compared with an factors.17 In both treatment groups, overall survival
active comparator, sorafenib. seemed to be longer in patients who had received
Analysis of overall survival might have been previous cytokine treatment than in those who had
confounded by subsequent active treatments, which received prior sunitinib treatment. Median duration of
were given to the majority of patients who discontinued first-line therapy was longer in the prior sunitinib

www.thelancet.com/oncology Vol 14 May 2013 559

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

We examined pretreatment clinical features as

Panel: Research in context prognostic factors. The MSKCC risk factors established
Systematic review before the availability of targeted agents13 were validated
As previously described,5 we searched PubMed for clinical trials here, as well as in patients treated with the mTOR
published in 2006–10 in English with the MeSH terms “renal inhibitor, everolimus, following progression on
cell carcinoma”, “vascular endothelial growth factor receptor”, sunitinib, sorafenib, or both agents.20 Other risk factors,
“targeted therapy”, and for specific targeted agents (sorafenib, such as the presence of bone metastases or previous
sunitinib, pazopanib, everolimus, temsirolimus). At the time, sunitinib treatment, were reported here and in the
no prospective randomised phase 3 trials comparing two analysis from a phase 3 trial of everolimus.20 Similarity
targeted drugs had been reported. Therefore, we did a in prognostic factors with treatments with different
randomised phase 3 study of axitinib versus sorafenib in mechanisms of action suggests that these factors are not
723 patients with advanced renal cell carcinoma who failed specific to a given treatment. Instead, these factors are
prior therapy with sunitinib, bevacizumab/interferon alfa, indicative of a more aggressive underlying renal cell
temsirolimus, or cytokine therapy. carcinoma phenotype. The risk factors identified here
could be used to counsel patients about prognosis, to
Interpretation design future clinical trials, and to interpret clinical trial
These results establish axitinib as a treatment option for data. However, they are not yet applicable for choosing
second-line therapy in advanced renal cell carcinoma. one agent versus another to treat patients with advanced
Optimisation of sequencing of targeted agents is a key focus renal cell carcinoma who have received previous
of research in advanced renal cell carcinoma, and data from systemic treatment.
the AXIS trial might help to improve treatment algorithms. Our data show that a shorter time on first-line sunitinib
Further assessment of approved and novel targeted therapies (0–25th percentile for time to progression on sunitinib)
through comparative effectiveness trials is needed. was associated with shorter overall survival on second-line
axitinib or sorafenib. These data might be indicative of a
subgroup compared with the prior cytokine subgroup, more aggressive natural history, cross-resistance to VEGF-
and this difference could have translated into a lead- targeted treatment, or both. PFS and overall survival data
time benefit for patients previously treated with and these prognostic factors can be used in the design of
cytokines. Additionally, patients in the prior cytokine phase 3 trials investigating new agents in this clinical
subgroup had their first exposure to VEGF-targeted setting.
therapy during the AXIS trial and thus did not have Hypertension is frequently associated with inhibition
previous resistance to this class of agents. Patients of the VEGF pathway resulting from multiple mechan-
previously treated with sunitinib, however, had already isms including inhibition of the vasodilator nitric oxide,
shown clinical resistance to VEGF-targeted therapy and as previously reported from retrospective analyses of
might have had shorter overall survival after treatment pooled trials with axitinib21 and sunitinib.22 In this
with either agent. phase 3 trial, the development of hypertension during
Controversy exists regarding the optimum sequencing axitinib or sorafenib treatment predicted improved
of targeted agents in metastatic renal cell carcinoma, and, overall survival in patients with metastatic renal cell
in particular, the use of mTOR inhibitors or alternative carcinoma, providing further support that the
anti-VEGF agents as second-line treatment after failure of association of treatment-induced hypertension and
first-line anti-VEGF therapy. Everolimus showed PFS improved clinical outcome is a class effect of VEGF
benefit when compared with placebo in patients who had pathway inhibitors. The high proportion of patients
progressed on sunitinib, sorafenib, or both agents;18 many with hypertension on axitinib treatment is consistent
patients had received two or more previous treatments, with its targeting profile as a potent inhibitor of the
representing a more heavily pretreated population than VEGF pathway. Data available elsewhere21,22 suggest that
the AXIS trial. A phase 3 trial comparing temsirolimus appropriate treatment of hypertension does not
versus sorafenib as second-line threatment for metastatic abrogate the benefit seen with tyrosine kinase
renal cell carcinoma was reported at the 2012 European inhibitors, and thus patients should be monitored for
Society for Medical Oncology annual meeting.19 The the emergence of hypertension with VEGF-targeted
primary endpoint of PFS was similar between the two therapy and treated with appropriate antihypertensive
treatment groups, and overall survival, assessed as a drugs. In this trial, patients who became hypertensive
secondary endpoint, was longer for patients treated with while receiving either axitinib or sorafenib, as defined
sorafenib than for those treated with temsirolimus. by the systolic blood pressure or diastolic blood
Optimisation of sequencing of targeted agents is a key pressure threshold and according to both the 8-week
issue to maximise survival in patients with metastatic and 12-week landmarks, had similar median overall
renal cell carcinoma. Additionally, identification of survival of about 21 months. Although interpretation is
patient-specific predictive biomarkers should be a major limited by small, retrospective subsets, these data
focus of future clinical trials. suggest that a phenotypic subset of patients with renal

560 www.thelancet.com/oncology Vol 14 May 2013

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

cell carcinoma exists who develop a hypertensive MDM, MEG, JT, SH, BR, SK, CC, and BIR participated in data analysis
and interpretation. RJM, BE, TEH, SO, MDM, MEG, and BIR
response to any VEGF-targeting agent and that they
participated in provision of patients. All authors participated in writing
have improved survival compared with patients who do the paper and approved the final version of the paper.
not develop hypertension when treated with a VEGF
Conflicts of interest
inhibitor. RJM has served as a consultant or adviser for Pfizer and has received
Conversely, we did not detect a PFS benefit in patients research funding from Pfizer, Novartis, and GlaxoSmithKline. BE has
who developed hypertension during treatment with served as a consultant or adviser for Bayer, Pfizer, and Novartis, and has
received honoraria from Bayer, Roche, Pfizer, Genentech, Novartis, and
axitinib or sorafenib in the AXIS trial. A prior Cox
Aveo. TEH has served as a consultant or adviser for and received
proportional hazards model analysis, with onset of honoraria and research funding from Pfizer, Bayer, GlaxoSmithKline,
diastolic blood pressure 90 mm Hg or higher as a time- and Novartis. MDM has served as a consultant or adviser for Pfizer,
dependent covariate, and an 8-week landmark analysis Novartis, Aveo, Genentech, and Abbott and has received research
funding from GlaxoSmithKline, Genentech, Pfizer, Novartis, Eisai,
of diastolic blood pressure 90 mm Hg or higher showed Bayer, and Abbott. SN has served as a consultant or adviser for Novartis
numerically—but not statistically significantly—longer and Pfizer; has received honoraria from Novartis, Pfizer,
overall survival and PFS in a pooled analysis of patients GlaxoSmithKline, and Roche; and has received research funding from
with different tumour types who developed hypertension Novartis, Pfizer, and GlaxoSmithKline. SO has served as a consultant or
adviser for Roche, Sanofi-Aventis, Novartis, and Bayer; has received
while receiving axitinib.21 The difference was statistically honoraria from Pfizer, Bayer, Novartis, Keocyt, Amgen, Roche, and
significant for overall survival, but not PFS, in the time- Sanofi-Aventis; and has received research funding from Sanofi-Aventis
dependent Cox model covariate analysis; neither end- and Bayer. MEG has served as a consultant or adviser for and received
point was statistically significant in the landmark honoraria from Pfizer and Bayer. JT, SH, CC, BR, and SK are employees
of and own stock in Pfizer. BIR has served as a consultant or adviser for
analysis. A landmark analysis of the subset of patients and received research funding from Pfizer. PT declares that he has no
with metastatic renal cell carcinoma, showed longer conflicts of interest.
PFS in the group with diastolic blood pressure 90 mm Acknowledgments
Hg or higher, whereas the extension of overall survival See the appendix for a full list of participating countries and principal See Online for appendix
was not statistically significant.21 In the pooled investigators. This study was sponsored by Pfizer Inc. Medical writing
retrospective analysis of four trials of sunitinib for the support, also funded by Pfizer Inc, was provided by Joanna Bloom, of
UBC Scientific Solutions (Southport, CT, USA).
treatment of metastatic renal cell carcinoma, both a
time-dependent Cox model covariate analysis and a References
1 Hu-Lowe DD, Zou HY, Grazzini ML, et al. Nonclinical
12-week landmark analysis showed longer overall antiangiogenesis and antitumor activities of axitinib (AG-013736),
survival in patients who developed diastolic blood an oral, potent, and selective inhibitor of vascular endothelial
growth factor receptor tyrosine kinases 1, 2, 3. Clin Cancer Res 2008;
pressure 90 mm Hg or higher or systolic blood pressure 14: 7272–83.
140 mm Hg or higher, and longer PFS only in those who 2 Rixe O, Bukowski RM, Michaelson MD, et al. Axitinib treatment in
developed systolic blood pressure 140 mm Hg or patients with cytokine-refractory metastatic renal-cell cancer:
a phase II study. Lancet Oncol 2007; 8: 975–84.
higher.22 Inherent limitations of retrospective,
3 Rini BI, Wilding G, Hudes G, et al. Phase II study of axitinib in
exploratory, subgroup analyses might have contributed sorafenib-refractory metastatic renal cell carcinoma. J Clin Oncol
to discrepancies seen for hypertension as a predictor of 2009; 27: 4462–68.
improved clinical outcomes in patients treated with 4 Tomita Y, Uemura H, Fujimoto H, et al. Key predictive factors of
axitinib (AG-013736)-induced proteinuria and efficacy: a phase II
VEGF-targeted therapy. The distinct methods used in study in Japanese patients with cytokine-refractory metastatic renal
these analyses (Cox model vs the landmark analysis) cell carcinoma. Eur J Cancer 2011; 47: 2592–602.
might have added to these disparities; statistical power 5 Rini BI, Escudier B, Tomczak P, et al. Comparative effectiveness of
axitinib versus sorafenib in advanced renal cell carcinoma (AXIS):
is reduced with landmark analyses compared with time- a randomised phase 3 trial. Lancet 2011; 378: 1931–39.
dependent Cox model covariate analyses because 6 INLYTA® (axitinib) prescribing information, 2012. http://labeling.
patients who die or have disease progression are pfizer.com/ShowLabeling.aspx?id=759 (accessed Feb 14, 2013).
excluded. Correlations between blood pressure and 7 Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to
evaluate the response to treatment in solid tumors. European
clinical outcomes are being prospectively assessed in an Organization for Research and Treatment of Cancer, National
ongoing phase 2 trial (NCT00835978) of axitinib with or Cancer Institute of the United States, National Cancer Institute of
Canada. J Natl Cancer Inst 2000; 92: 205–16.
without dose titration in first-line metastatic renal cell
8 Trotti A, Colevas AD, Setser A, et al. CTCAE v3.0: development of a
carcinoma, and could provide further insight into comprehensive grading system for the adverse effects of cancer
hypertension as a biomarker of efficacy in patients treatment. Semin Radiat Oncol 2003; 13: 176–81.
treated with VEGF-targeted agents. 9 Cella D, Yount S, Du H, et al. Development and validation of the
Functional Assessment of Cancer Therapy-Kidney Symptom Index
PFS and overall survival data and prognostic factors (FKSI). J Support Oncol 2006; 4: 191–99.
validated in this trial can be applied to clinical trial 10 Cella D, Yount S, Brucker PS, et al. Development and validation of a
design for novel therapeutics for metastatic renal cell scale to measure disease-related symptoms of kidney cancer.
Value Health 2007; 10: 285–93.
carcinoma.
11 Escudier B, Eisen T, Stadler WM, et al. Sorafenib in advanced
Contributors clear-cell renal-cell carcinoma. N Engl J Med 2007; 356: 125–34.
RJM, BE, MDM, JT, SK, CC, and BIR participated in study conception 12 Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor
and design. BE, PT, TEH, SO, MDM, SN, JT, SK, CC, and BIR response. J Clin Oncol 1983; 1: 710–19.
participated in collection and assembly of data. RJM, BE, PT, TEH, SO,

www.thelancet.com/oncology Vol 14 May 2013 561

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
 Articles

13 Motzer RJ, Bacik J, Schwartz LH, et al. Prognostic factors for 18 Motzer RJ, Escudier B, Oudard S, et al. Efficacy of everolimus in
survival in previously treated patients with metastatic renal cell advanced renal cell carcinoma: a double-blind, randomised,
carcinoma. J Clin Oncol 2004; 22: 454–63. placebo-controlled phase III trial. Lancet 2008; 372: 449–56.
14 Butt Z, Peipert J, Webster K, Chen C, Cella D. General population 19 Hutson T, Escudier B, Esteban E, et al. Temsirolimus vs sorafenib
norms for the Functional Assessment of Cancer Therapy-Kidney as second line therapy in metastatic renal cell carcinoma: results
Symptom Index (FKSI). Cancer 2013; 119: 429–37. from the INTORSECT trial. Ann Oncol 2012; 23: ixe14 (abstr).
15 Broglio KR, Berry DA. Detecting an overall survival benefit that is 20 Motzer RJ, Escudier B, Oudard S, et al. Phase 3 trial of everolimus
derived from progression-free survival. J Natl Cancer Inst 2009; for metastatic renal cell carcinoma: final results and analysis of
101: 1642–49. prognostic factors. Cancer 2010; 116: 4256–65.
16 Negrier S, Tang Y, Chen C, Bycott P. Interpreting overall survival 21 Rini BI, Schiller JH, Fruehauf JP, et al. Diastolic blood pressure as a
(OS) results when progression free survival (PFS) benefits exists in biomarker of axitinib efficacy in solid tumors. Clin Cancer Res 2011;
today’s oncology landscape – metastatic renal cell carcinoma 17: 3841–49.
(mRCC) case study. Ann Oncol 2012; 23: ix453 (abstr). 22 Rini BI, Cohen DP, Lu DR, et al. Hypertension as a biomarker of
17 Hudes G, Carducci M, Tomczak P, et al. Temsirolimus, interferon efficacy in patients with metastatic renal cell carcinoma treated with
alfa, or both for advanced renal-cell carcinoma. N Engl J Med 2007; sunitinib. J Natl Cancer Inst 2011; 103: 763–73.
356: 2271–81.

562 www.thelancet.com/oncology Vol 14 May 2013

Downloaded for Anonymous User (n/a) at Prince of Songkla University from ClinicalKey.com by Elsevier on October 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.