Fax: (216) 444-9464
Clinical factors associated with outcome in patients with metastatic clear-cell renal cell carcinoma treated with vascular endothelial growth factor-targeted therapy
Article first published online: 18 JUN 2007
Copyright © 2007 American Cancer Society
Volume 110, Issue 3, pages 543–550, 1 August 2007
How to Cite
Choueiri, T. K., Garcia, J. A., Elson, P., Khasawneh, M., Usman, S., Golshayan, A. R., Baz, R. C., Wood, L., Rini, B. I. and Bukowski, R. M. (2007), Clinical factors associated with outcome in patients with metastatic clear-cell renal cell carcinoma treated with vascular endothelial growth factor-targeted therapy. Cancer, 110: 543–550. doi: 10.1002/cncr.22827
- Issue published online: 18 JUL 2007
- Article first published online: 18 JUN 2007
- Manuscript Accepted: 20 APR 2007
- Manuscript Revised: 17 APR 2007
- Manuscript Received: 6 FEB 2007
- advanced renal cell carcinoma;
- vascular endothelial growth factor;
- prognostic factors;
- prognostic model
Therapy targeted against the vascular endothelial growth factor (VEGF) pathway is a standard of care for patients with metastatic renal cell carcinoma (RCC). The identification of patients who are more likely to benefit from these agents is warranted.
In total, 120 patients with metastatic clear-cell RCC received bevacizumab, sorafenib, sunitinib, or axitinib on 1 of 9 prospective clinical trials at the Cleveland Clinic. Clinical features associated with outcome were identified by univariate analysis; then, a stepwise modeling approach based on Cox proportional hazards regression was used to identify independent prognostic factors and to form a model for progression-free survival (PFS). A bootstrap algorithm was used to provide internal validation.
The overall median PFS was 13.8 months, and the objective response according to the Response Criteria in Solid Tumors was 34%. Multivariate analysis identified time from diagnosis to current treatment <2 years; baseline platelet and neutrophil counts >300 K/μL and >4.5 K/μL, respectively; baseline corrected serum calcium <8.5 mg/dL or >10 mg/dL; and initial Eastern Cooperative Oncology Group performance status >0 as independent, adverse prognostic factors (PF) for PFS. Three prognostic subgroups were formed based on the number of adverse prognostic factors present. The median PFS in patients with 0 or 1 adverse prognostic factor was 20.1 months compared with 13 months in patients with 2 adverse prognostic factors and 3.9 months in patients with >2 adverse prognostic factors.
Five independent prognostic factors for predicting PFS were identified and were used to categorize patients with metastatic RCC who received VEGF-targeted therapies into 3 risk groups. These prognostic factors can be incorporated into patient care and clinical trials that use such novel, VEGF-targeted agents. Cancer 2007. © 2007 American Cancer Society.
Renal cell carcinoma (RCC) is the tenth most common cancer in the U.S., accounting for >38,000 new diagnoses and >12,800 deaths annually.1 Epidemiologic studies suggest a continued rise in RCC incidence in the U.S.2 Although most patients with early-stage RCC can be cured surgically, approximately 33% of patients present with metastatic disease for which the treatment usually is not curative.3 In addition, approximately 50% of patients who undergo potentially curative surgery for less advanced disease can be expected to develop a recurrence with distant metastases.4 Five-year survival for patients with de novo metastatic or recurrent disease ranges between 0% and 20%.3, 5
Many authors have identified clinical factors that are associated with outcome in patients with metastatic RCC when they are treated with cytokines (interferon, and interleukin), chemotherapy, or a variety of historic therapies. These schemas included tumor-, patient-, and disease-related factors, such as performance status (PS), time from diagnosis to therapy, number of metastatic sites, visceral metastasis, hemoglobin, calcium, lactate dehydrogenase (LDH), inflammation markers, and others. Prognostic models have been derived from these factors and have greatly improved patient prognostication and counseling.6–15
Recent advances in the understanding of the biology of clear-cell RCC have established the importance of an inactivated von Hippel-Lindau (VHL) gene and its role in the pathogenesis of sporadic clear-cell RCC. VHL gene alterations lead to increased levels of hypoxia-induced factor α (HIFα) and subsequent activation of several of the hypoxia-regulated genes.16, 17 The overexpression of vascular endothelial growth factor (VEGF) and other growth factors results in endothelial cell migration, tumor growth, and tumor angiogenesis.18, 19 Consequently, VEGF inhibition has become an attractive therapeutic target in patients with metastatic RCC.20 The inhibition of this pathway with agents such as bevacizumab,21 sunitinib,22, 23 and sorafenib,24 has demonstrated significant antitumor activity in advanced RCC.
Although these trials have enrolled patients and/or have reported results according to existing classification schemas, such schemas have been developed from patients who were treated with cytokines and/or chemotherapy, and it is unclear whether the same factors reported previously are relevant to patients who are treated with VEGF-targeted therapy. Because the vast majority of patients who have advanced RCC currently are receiving anti-VEGF therapy, the identification of relevant clinical factors in the anti-VEGF era is warranted. In the current study, clinical factors that were associated with outcome in a cohort of patients with clear-cell RCC who were receiving anti-VEGF agents were identified, and a prognostic model was constructed using progression-free survival (PFS) as an endpoint.
MATERIALS AND METHODS
We reviewed the records of patients with metastatic RCC who were treated with anti-VEGF agents (bevacizumab, sunitinib, sorafenib, and axitinib [AG-013736]) and who were enrolled onto Institutional Review Board-approved clinical trials at the Cleveland Clinic Taussig Cancer Center between October 2003 and January 2006. The trials were composed of 3 phase 2 trials, 2 phase 3 trials, 2 randomized phase 2 trials, and 2 compassionate use studies. The 3 phase 2 trials included patients who were treated with sunitinib,22 axitinib,25 and bevacizumab in combination with interleukin-2 (unpublished data). The 2 phase 3 trials included a randomized study of sunitinib versus interferon-α as first-line therapy23 and randomized study of sorafenib versus placebo in cytokine-refractory patients.24 The 2 randomized phase 2 studies included bevacizumab with or without erlotinib26 and sorafenib versus interferon-α,27 both as first-line therapy. Finally, 2 compassionate use studies included patients who were treated with either sorafenib or sunitinib as part of 2 extended-access programs. Eligibility criteria for the trials were fairly uniform and generally included the following: histologic documentation of RCC; clinical or biopsy evidence of metastatic disease; clear-cell component; measurable disease; an Eastern Cooperative Oncology Group (ECOG) PS of 0 or 1; adequate renal, hepatic, and bone marrow function; absent or previously treated central nervous system metastasis; no prior history of cancer (except basal cell carcinoma or carcinoma in situ of the cervix); absence of significant cardiac disease; and no recent surgery. Data collected included standard pretreatment patient and disease characteristics, baseline biochemical parameters, first date of treatment, best response to treatment, date of progression according to Response Criteria in Solid Tumors (RECIST) criteria,28 date of death or last follow-up, and other factors previously reported as prognostic for survival in patients with metastatic RCC (Table 1). C-reactive protein data were not collected, because this information was not available for most patients. The Memorial Sloan-Kettering Cancer Center (MSKCC)10 and Cleveland Clinic Foundation (CCF)14 risk groups were applied to these data to investigate the applicability of these 2 models in patients with RCC who received VEGF-targeted therapy. An objective response was defined by RECIST criteria for all patients.28 Overall survival was defined as the time from initiation of treatment to the date of death or last follow-up. PFS was defined from the initiation of treatment to the date of progression or death, whichever came first.
|Demographic and clinical hactors||Biochemical and hematologic hactors||Histologic hactors|
|Sex||Serum albumin (g/dL)||Histology|
|No. of metastatic sites||Serum alkaline phosphatase (U/L)||Nuclear grade|
|Age at study entry||Serum lactate dehydrogenase (U/L)|
|ECOG PS||Serum calcium (mg/dL)|
|Time from diagnosis to study entry||Corrected serum calcium (mg/dL)|
|Prior nephrectomy||Serum creatinine (mg/dL)|
|Prior radiotherapy||Hemoglobin (g/dL)|
|Kidney initially involved||Neutrophil count (K/μL)|
|Lung metastasis||Lymphocyte count (K/μL)|
|Mediastinal metastasis||Platelet count (K/μL)|
|Retroperitoneal lymph node metastasis|
|Other metastatic sites|
PFS distributions were estimated using the Kaplan-Meier method.29 The correlations between survival and the individual factors listed in Table 1 were analyzed using the log-rank test30 and the Cox proportional-hazards model.31 Clinical and pathologic characteristics that, by nature, are categorical, such as sex, and PS were analyzed using the log-rank test. Biochemical parameters and other characteristics that are measured on a continuum, such as age and time from diagnosis to study entry, were analyzed as continuous variables using the Cox proportional-hazards model and as categoric variables using the log-rank test. The cut-off points used for categorizations were based on previously described cut-off points in the literature and/or recursive partitioning. The Cox proportional-hazards model with step-wise variable selection was used to assess multiple factors simultaneously. A significance level of .05 was used as the criterion for determining variable entry and removal from the model. For convenience only, the categoric forms of continuous variables were included in the multivariable analyses. The proportionality assumption was assessed graphically using log (−log) plots and quantitatively using the Z-statistic proposed by Harrell.32 All tests of statistical significance were 2-sided. All data analyses were conducted using SAS statistical analysis software (version 8; SAS Inc., Raleigh, NC). Once a final model was determined, internal validation was performed using a bootstrap procedure in which samples of 120 patients in size were generated randomly (with replacement) from the original study population (also n = 120 patients) and analyzed using the stepwise logistic regression procedure described above. Five hundred such samples were generated and analyzed, and the frequency of each factor's inclusion in the resulting models was calculated. Factors that were present in ≥50% of the models were considered significant and were used to build a final “bootstrap-based” model.
One hundred thirty patients who had no prior history of treatment with anti-VEGF therapies who were enrolled onto Institutional Review Board-approved clinical trials between October 2003 and January 2006 were identified. Excluding patients with incomplete key data (n = 2), patients who were treated on more than 1 clinical trial (n = 1), and patients with pure nonclear-cell histology (n = 7), in total, 120 patients were available for the current analysis. Table 2 lists the distribution of anti-VEGF agents used by the cohort of patients. The majority of patients (101 of 120 patients; 84%) were treated with either sorafenib or sunitinib. Forty-five patients (37%) received anti-VEGF therapy as their first-line treatment compared with 75 patients (63%) who received previous nonanti-VEGF therapies. Seventy-one percent of patients were men, and 46% were aged >60 years. All patients had undergone prior nephrectomy, which, in general, was an eligibility criterion for participation in these clinical trials. The majority of patients had an ECOG PS of 0, and only 7% of patients had central nervous system metastases. Although an attempt was made to capture the nuclear grade of the primary tumor, complete information was not available on 39% of patients. Missing grade was attributed mostly to the unavailability of the original specimen for review (eg, surgery or biopsy done at an outside hospital, specimens >15 years since nephrectomy) or to the fact that some specimens were obtained by cytologic aspiration.
|Treatment||No. of patients (%)|
|First line||Second line (also Third, fourth, etc)||Total|
|Bevacizumab||12 (100)||0||12 (10)|
|Axitinib||0||7 (100)||7 (6)|
|Sunitinib||12 (20)||48 (80)||60 (50)|
|Sorafenib||21 (51)||20 (49)||41 (34)|
|Total||45 (37)||75 (63)||120 (100)|
Overall, 34% of patients achieved an objective response according to investigator-assessed RECIST criteria (3 complete responses and 38 partial responses). Thirty-one percent of previously untreated patients achieved a response (partial or complete) compared with 36% of previously treated patients (P = .69). Sixty-two patients (52%) have developed disease progression, and 17 patients (14%) have died. The overall estimated PFS is 13.8 months (95% confidence interval [95% CI], 10.7–19.0 months) (Fig. 1). The median follow-up for the 58 patients who remain progression free is 10.9 months (range, 4.1–32 months).
Univariate Analysis of Risks Factors for PFS
A number of clinical and laboratory variables were associated with a worse PFS on univariate analysis, as shown in Table 3. In addition, the MSKCC and CCF models (comparing favorable-risk vs intermediate-risk vs unfavorable-risk groups) proposed for previously untreated and treated patients, respectively, also were prognostic for PFS on univariate analysis. Nuclear grade did not appear to have an impact on PFS; however, it did have borderline significance. Because data for nuclear grade were missing for 39% of patients, grade was not included in the multivariate analysis. Similarly, PFS was unaffected by prior systemic therapy (14.7 months vs 13.8 months for untreated vs previously treated patients, respectively; P = .79), suggesting that exposure cytokines or chemotherapy does not have an impact on subsequent response to VEGF-targeted therapy (Fig. 2).
|Characteristic||No. of patients (%)||No. of failures (%)||Median PFS, mo||Log-rank P|
|Men||85 (71)||45 (53)||13|
|Women||35 (29)||17 (49)||19.9||.39|
|≤ 60||65 (54)||36 (55)||11|
|>60||55 (46)||26 (47)||17.5||.08|
|Right||64 (53)||31 (48)||11|
|Left||54 (45)||30 (56)||16.5||.94|
|Time from diagnosis to current treatment, mo|
|<24||58 (48)||37 (64)||9.2|
|≥24||62 (52)||25 (40)||19.9||.009|
|Prior systemic therapy|
|No||45 (37)||25 (56)||14.7|
|Yes||75 (63)||37 (49)||13.8||.79|
|No||86 (72)||45 (52)||14.7|
|Yes||34 (28)||17 (50)||10.7||.45|
|0||85 (71)||38 (45)||16.5|
|≥1*||35 (29)||24 (69)||7.6||<.001|
|1/2||17 (14)||8 (47)||17.6|
|3/4||56 (47)||29 (52)||16.5||.70|
|No||34 (28)||15 (44)||20.1|
|Yes||86 (72)||47 (54)||13||.22|
|No||93 (77)||46 (49)||14.7|
|Yes||27 (23)||16 (59)||11.5||.75|
|No||77 (64)||36 (47)||16.5|
|Yes||43 (36)||26 (60)||10.7||.08|
|No||111 (93)||57 (51)||14.7|
|Yes||9 (7)||5 (56)||7.2||.06|
|Retroperitoneal lymph nodes|
|No||96 (80)||47 (49)||16.5|
|Yes||24 (20)||15 (63)||6.5||.12|
|Mediastinal lymph nodes|
|No||62 (52)||34 (55)||13|
|Yes||58 (48)||28 (48)||13.8||.99|
|No||58 (48)||28 (58)||13.8|
|Yes||62 (52)||34 (55)||16.5||.30|
|No. of sites|
|≤ 2||56 (47)||25 (45)||16.5|
|>2||64 (53)||37 (58)||11||.06|
|<LLN||56 (47)||32 (57)||11.0|
|≥LLN||64 (53)||30 (47)||16.5||.19|
|≤10.5||109 (91)||55 (50)||16.5|
|>10.5||11 (9)||7 (64)||8.1||.01|
|≤6||98 (82)||50 (51)||16.5|
|>6||22 (18)||12 (55)||8.1||.03|
|≤4.5||61 (51)||25 (41)||17.6|
|>4.5||59 (49)||37 (63)||8.3||.004|
|≤1.5||68 (57)||35 (51)||11.5|
|>1.5||52 (43)||27 (52)||14.7||.19|
|≤400||112 (93)||56 (50)||14.7|
|>400||8 (7)||6 (75)||7.8||.07|
|≤300||100 (83)||47 (47)||14.7|
|>300||20 (17)||15 (75)||5.9||.004|
|≤1.5||89 (74)||47 (53)||11.5|
|>1.5||25 (21)||13 (52)||16.5||.76|
|≤1.5||114 (95)||58 (51)||13.8|
|>1.5||6 (5)||4 (67)||5.6||.20|
|Alkaline phosphatase, U/L|
|≤120||95 (79)||47 (49)||16.5|
|>120||25 (21)||15 (60)||9.2||.01|
|8.5–10.5||109 (91)||55 (50)||14.7|
|<8.5 or >10.5||11 (9)||7 (64)||2.3||.008|
|<4||29 (23)||18 (62)||5.7|
|≥4||98 (77)||49 (50)||16.5||<.001|
|Corrected calcium, mg/dL|
|8.5–10||104 (87)||49 (47)||17.6|
|<8.5 or >10||16 (13)||13 (81)||7.4||<.001|
|MSKCC risk group†|
|Favorable||48 (40)||19 (40)||20.1|
|Intermediate||66 (55)||38 (58)||11|
|Unfavorable||6 (5)||5 (83)||2.3||<.001|
|MSKCC risk group‡|
|Favorable||62 (52)||30 (46)||14.7|
|Intermediate||51 (43)||29 (55)||17.5|
|Unfavorable||7 (6)||6 (86)||2.2||.003|
|CCF risk group†|
|Favorable||41 (34)||16 (39)||16.5|
|Intermediate||40 (33)||20 (50)||14.7|
|Unfavorable||39 (33)||26 (67)||8.4||.007|
|CCF risk group‡|
|Favorable||56 (47)||25 (45)||16.5|
|Intermediate||43 (36)||23 (53)||17.6|
|Unfavorable||21 (17)||14 (67)||6.5||.009|
Multivariate Analysis of Risks Factors for PFS
Although many factors were associated individually with PFS on univariate analysis, only 6 factors were identified as independent predictors of a poor outcome on subsequent multivariate analysis. With the least favorable feature listed first, the following factors were identified: ECOG PS (≥1 vs 0), time from diagnosis to current treatment (<2 years vs ≥2 years), abnormal baseline corrected serum calcium (<8.5 mg/dL or >10 mg/dL vs 8.5–10 mg/dL), high platelet count (>300 K/μL vs ≤300 K/μL), higher absolute neutrophil count (ANC) (>4.5 K/μL vs ≤4.5 K/μL), and lower absolute lymphocyte count (>1.5 K/μL vs ≤1.5 K/μL). However, during internal validation using a bootstrap algorithm, lymphocyte count came into a fairly small number of models (38%) and, thus, was dropped from the final model (Table 4). It is noteworthy that neither the MSKCC risk score nor the CCF risk score was associated with PFS when it was considered in multivariate analysis.
|Factor||Level associated with poor PFS||Parameter estimate||HR||P|
|Corrected calcium||<8.5 mg/dL or >10.0 mg/dL||1.33 ± 0.33||3.77||<.001|
|Neutrophils||>4.5 K/μL||0.84 ± 0.28||2.31||.002|
|Platelets||>300 K/μL||0.92 ± 0.31||2.50||.003|
|ECOG PS||>0||0.75 ± 0.28||2.11||.007|
|Time from diagnosis to study entry||<2 y||0.60 ± 0.27||1.82||.02|
Derivation of a Prognostic Model
Based on the magnitude of the parameter estimates and the observation that their variances were similar, it is convenient to combine these factors into a simple scoring system. By counting the number of unfavorable characteristics present (1 “point” for each adverse prognostic factor, except for corrected serum calcium <8.5 mg/dL or >10 mg/dL, which counts as 2 “points” because, according to Table 4, the magnitude of the associated regression coefficient is approximately twice that of the other parameters), 3 distinct prognostic groups were identified. Patients with 0 or 1 adverse prognostic factors (63 patients; 53%) have a median PFS of 20.1 months (95% CI, 19–22.3 months) and are considered a favorable-risk group. Patients with 2 adverse prognostic factors (27 patients; 23%) have a median PFS of 13 months (95% CI, 8.6–17.6 months) and are considered an intermediate-risk group. Finally, patients with >2 adverse prognostic factors (30 patients; 25%) have a dismal PFS of 3.9 months (95% CI, 1.8–7.2 months) and are considered a poor-risk group (Table 5, Fig. 3).
|Risk group||No. of adverse risk factors present*||No. of patients (%)||No. of failures (%)||One-year PFS, %||Median PFS, mo|
|1||0 or 1||63 (53)||22 (35)||67||20.1|
|2||2||27 (23)||15 (56)||44||13|
|3||3–5||30 (25)||25 (83)||16||3.9|
Advances in the treatment of patients with metastatic RCC and the introduction of novel agents targeting the VEGF pathway make it necessary to identify the contemporary clinical, laboratory, and molecular features that can be used to predict outcome. The current results demonstrate that prognostic factors and models that were described previously and that often are used prior to the availability of novel agents may not be entirely applicable to patients with advanced RCC who are starting VEGF therapy. This is important, especially because clinical trials evaluating these novel compounds have stratified patients using outdated models from the cytokine era. The suggested model has grouped patients into 3 risk groups based on 5 readily available clinical prognostic variables. These can be incorporated readily into clinical patient care and into stratification schema for future clinical trials of these novel agents.
It is noteworthy that the median PFS of 13 months in patients with metastatic RCC who received the VEGF-targeting agents was considerably superior to that reported in prior series involving earlier therapies. This finding is supported by clinical trials, which have indicated improved PFS with sunitinib, sorafenib, bevacizumab, and axitinib in patients with advanced RCC.21, 23–25 The absolute value of PFS, however, must be interpreted in light of the select group of patients with metastatic RCC who are fit enough to participate in clinical trials at a specialized tertiary referral center for RCC.
Another finding of interest was the statistically similar PFS of patients who received front-line VEGF-targeted therapy for metastatic disease compared with patients who received VEGF-targeted therapy as second-line treatment. This finding supports the hypothesis that the biology of RCC that mediates response to VEGF-targeted therapy perhaps may not be impacted adversely by prior exposure to cytokine therapy.
In the current study, we identified additional prognostic factors in patients with metastatic RCC, such as high platelet and neutrophil counts. These 2 factors may be considered “inflammation markers.” Recently, it was demonstrated that thrombocytosis was an independent, poor PF in a retrospective study of 700 patients with metastatic RCC.33 Patients who had elevated platelet counts had significantly shorter survival than patients ho had normal platelet counts (median survival, 8.4 months and 14.6 months, respectively; P < .001). The exact mechanism of thrombocytosis in association with RCC is unclear, although this may be a reflection of an overproduction of interleukins and other growth factors by the tumor.34 In turn, platelet overproduction can enhance the adherence and penetration of malignant cells through the endothelial wall,35 and platelet granules contain a variety of angiogenic factors, such as VEGF, platelet-derived growth factor, transforming growth factor β, and others that have been implicated in various steps of tumor progression.36, 37
Similarly, increased an ANC was cited as an independent adverse risk factor in 2 metastatic RCC prognostic models for patients who received treatment with cytokine-based regimens.12, 13 Recently, immunologic variables derived from blood (ANC) and tumor analyses (intratumoral neutrophils and CD57-positive natural killer cell counts) added significantly to prognostic models that were based on clinical risk factors only in patients who were treated with interleukin-2-based regimens.38 Therefore, the status of the immune system may exert a critical role in the response to cytokine-based therapy. In the current study, we observed that immune parameters such as ANC continued to play an important role in patient prognostication despite the presumed mechanism of action of VEGF-targeted agents. This finding supports other data demonstrating a potential immunomodulatory effect of sunitinib in promoting an immunostimulatory T-helper 1 bias and reducing immunosuppressive T-regulatory cells.39 Further investigation of a potential immunostimulatory mechanism of VEGF-targeted agents is warranted.
This study had many limitations. First, patients were treated with different VEGF-targeted agents, and the relevance and application of our factors and model may not be considered equal for all agents. Alternatively, a heterogeneous group of anti-VEGF agents may make these results more generalizable for this class of agents in metastatic RCC. The implications and findings of our study will require external validation before they can have widespread application. This study did not address or investigate the prognostic value of molecular markers for this disease, specifically, whether mutation or methylation of the VHL gene has prognostic implications. A recent retrospective study suggested that the time to progression on VEGF-targeted therapy may be prolonged in patients with VHL methylation or mutations that truncate or shift the VHL reading frame.40 The VHL status of tumor tissue from the current cohort of patients is being analyzed to incorporate these results into our proposed model. Prospective investigation of clinical and molecular features in a large number of patients with RCC will be required.
In summary, the current analyses identified 5 readily available clinical parameters that can classify patients with metastatic RCC into 3 risk groups with variable outcomes after they receive treatment with VEGF-targeting agents. Further validation of this model will help to identify which patients are most likely to respond to such therapy and will allow the appropriate application of treatment to those patients who are most likely to benefit.
- 25AG-013736, a multi-target tyrosine kinase receptor inhibitor, demonstrates anti-tumor activity in a phase 2 study of cytokine-refractory, metastatic renal cell cancer (RCC) Proc Am Soc Clin Oncol. 2005; 23: 16S. Abstract 4509., , , et al.
- 26Bevacizumab with or without erlotinib in metastatic renal cell carcinoma (RCC). Proc Am Soc Clin Oncol. 2006; 24: 18S. Abstract 4523., , , et al.
- 27Randomized phase II trial of the multi-kinase inhibitor sorafenib versus interferon (IFN) in treatment-naïve patients with metastatic renal cell carcinoma (mRCC). Proc Am Soc Clin Oncol. 2006; 24: 18S. Abstract 4501., , , et al.
- 28New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst. 2000; 92: 205–216., , , et al.
- 31Analysis of Survival Data. 1st ed. New York, NY: Chapman and Hall; 1990., .
- 32The phglm procedure. In: HastingsRP, ed. SAS Supplemental Library User's Guide, version 5. Cary, SC: SAS Institute Inc.; 1986: 437–466..
- 39T regulatory cells (Treg) in patients with metastatic renal cell carcinoma (mRCC) decrease during sunitinib treatment: correlations with clinical responses and T helper 1/T helper 2 (Th1/Th2) bias. Proc Am Soc Clin Oncol. 2006; 24: 18S. Abstract 2526., , , et al.
- 40Clinical response to therapy targeted at vascular endothelial growth factor in metastatic renal cell carcinoma: impact of patient characteristics and von Hippel-Lindau gene status. BJU Int. 2006; 98: 756–762., , , et al.
- 42Prognostic factors for survival in previously treated patients (pts) with metastatic renal cell cancer (RCC). Proc Am Soc Clin Oncol. 2003; 22: 16S. Abstract 1647., , , et al.