An important goal of noncurative therapy for metastatic renal cell carcinoma (mRCC) is tumor burden (TB) control. However, to the authors' knowledge, the impact of TB characteristics on clinical outcome has not been studied in patients with mRCC who were treated with vascular endothelial growth factor-targeted therapy.
Patients with clear cell mRCC who were treated with sunitinib between June 2004 and October 2007 were retrospectively identified. Computed tomography scans were re-reviewed from baseline, at the time of maximal TB shrinkage (TS) while receiving sunitinib, and at the time of progressive disease (PD). Measurements were recorded as per Response Evaluation Criteria In Solid Tumors (RECIST).
A total of 69 patients were identified. The majority (54%) were classified as being of favorable risk using Cleveland Clinic Foundation Tyrosine Kinase Inhibitor (CCF TKI) risk group criteria. All patients underwent prior nephrectomy and 77% received prior systemic therapy. There were a median of 8 metastatic deposits across all organs (range, 1-27 deposits). The median TB at the initiation of therapy was 14.0 cm (range, 3.0 cm-42.2 cm). On multivariable analysis, baseline characteristics of disease confined to above the diaphragm (P = .03) and a total TB <13 cm (P = .09) were found to be independent positive predictors of progression-free survival. A+ baseline, total number of metastases <10 (P < .001) and TB above the diaphragm <6.5 cm (P = .05) were found to be independent positive predictors of overall survival (OS). Increased TS while receiving sunitinib was found to be significantly associated with OS (P < .001). At the time of PD, tumor location and pattern of disease progression were not found to be associated with survival as measured from the date of PD. However, total TB (P = .003) and total number of metastatic deposits (≤12 vs >12; P < .001) were found to be significant predictors of survival after PD.
The management of metastatic renal cell carcinoma (mRCC) has undergone significant advancements in recent years. This has been achieved primarily through the elucidation of the vascular endothelial growth factor (VEGF) pathway and its integral role in the biology of RCC. As a result, several VEGF pathway inhibitors have been developed to treat patients with mRCC. Through several randomized controlled trials, VEGF pathway inhibitors such as sunitinib,1, 2 sorafenib,3 bevacizumab,4, 5 and pazopanib6 have demonstrated improved clinical outcomes. Given these superior outcomes, VEGF pathway inhibitors have now become the standard of care in patients with mRCC.
Despite these advances, there is still a great deal of information to be discovered regarding the behavior of mRCC treated with VEGF-inhibiting agents. Prognostic criteria incorporating baseline clinical and laboratory characteristics have been developed in the era of VEGF-targeted therapy.7, 8 Although these models identify select criteria to help estimate prognosis, to the best of our knowledge, the impact of TB on clinical endpoints such as progression-free survival (PFS) and overall survival (OS) have not been well studied to date. The idea of tumor burden (TB) correlating with survival was suggested in a study examining growth rate constants of renal carcinomas.9 In this study, an estimate of TB at the time of death was determined based on the calculated growth rate and the median survival of patients on the study. This was used to compare how growth rates and survival change while patients are on or off treatment with bevacizumab. Another study retrospectively examined both TB and growth rate constants using computed tomography (CT) measurements in patients with mRCC who were treated with ixabepilone.10 Here, TB was identified as having a strong correlation with OS, suggesting that the amount of tumor impacts clinical outcome in patients with RCC. Given these observations, it was hypothesized that the amount and location of mRCC, both before and during therapy with sunitinib, would be associated with clinical outcome independent of known prognostic variables. Therefore, a retrospective review of patients with mRCC who were treated with sunitinib was conducted comparing TB characteristics with patient outcomes.
MATERIALS AND METHODS
Patients with clear cell mRCC who were treated with sunitinib at the Cleveland Clinic Taussig Cancer Institute between June 1, 2004 and October 5, 2007 were retrospectively identified. CT images were re-reviewed by an investigator blinded to clinical outcome from 4 predefined time points: 1) baseline (the CT scan immediately before the initiation of therapy with sunitinib); 2) at the time of maximal TB shrinkage (TS) while receiving sunitinib; 3) at the time of progressive disease (PD) as per Response Evaluation Criteria In Solid Tumors (RECIST); and 4) at the time of last radiographic assessment before death/last visit. TB and percentage of TS were measured as per RECIST (version 1.0).11 Patient, tumor, and treatment characteristics were collected via review of the electronic medical record under a protocol approved by the Institutional Review Board.
PFS and OS were calculated from the date treatment with sunitinib was initiated to the date of confirmed PD or death, respectively; both outcomes were summarized using the Kaplan-Meier method. All patients who developed PD eventually died. In the analysis of survival after documented PD, survival from PD (SfP) was calculated from the date of PD rather than from the initiation of therapy with sunitinib because patients developed PD at various times and relating factors present at the time of PD to OS could be biased. A recursive partitioning algorithm was used to identify cutoff points for the different TB variables studied, which included whether disease was present above or below the diaphragm (or both); the overall baseline soft tissue TB; the soft tissue TB present above and below the diaphragm, respectively; the ratio of soft tissue TB above and below the diaphragm; the number of organ sites involved; the total number of metastases present; and the number of bony and nonbony metastases present. Univariable analyses of PFS, OS, and SfP were performed using the log-rank test and Cox proportional hazards model. Univariable analyses of objective response (OR) and TS were conducted using chi-square tests and the Wilcoxon rank sum test. The McNemar test was used to compare metastatic sites present before the initiation of sunitinib and those present at the time of PD. Multivariable analyses of PFS and OS initially considered all the factors in Tables 1 and 2 and used Cox proportional hazards models with stepwise variable selection to identify independent prognostic factors. The analyses of the association between maximum TS and outcome treated TS as a time-varying covariate. All tests of statistical significance were 2-sided and all analyses were conducted using SAS statistical software (version 9.1; SAS Institute Inc, Cary, NC).
Table 1. Patient Characteristics Prior to Initiation of Sunitinib (n=69)
Table 2. Sites of Metastases at Baseline Prior to Initiation of Sunitinib
Median No. of Metastases (Range)
Median Tumor Burden (Range), cm
NA, not applicable.
Mediastinal lymph nodes
Retroperitoneal lymph nodes
Total (all organs)
A total of 69 patients were identified. Baseline characteristics were typical of a mRCC population (Table 1). The majority of patients (54%) were classified as being of favorable risk based on Cleveland Clinic Foundation Tyrosine Kinase Inhibitor (CCF TKI) risk group classification. All 69 patients underwent prior nephrectomy. The majority of patients (73%) had received some form of prior systemic therapy, which was predominately immunomodulatory in nature (65%). At the time of last follow-up, 86% of patients had developed PD at a median of 13.5 months, and 49% of patients had died at a median of 31.0 months.
The lung (87% of patients) and mediastinal lymph nodes (52%) were the most common sites of metastatic disease (Table 2). Overall, patients had a median of 3 organ sites of metastatic disease (range, 1-8 organ sites) regardless of whether one considered sites of bone metastases separately or combined into a single “site.” The median total number of metastases across all organ sites was 8 (range, 1-21 metastases) and the median number of nonosseous metastases was 7 (range, 1-20 nonosseous metastases). The median total TB was 14.0 cm (range, 3.0 cm-42.2 cm), with a median TB above the diaphragm of 8.3 cm (range, 0 cm-22.6 cm) and a median TB below the diaphragm of 4.2 cm (range, 0 cm-26.0 cm), with TB above the diaphragm defined as lung and/or mediastinal metastases and TB below the diaphragm defined as metastases to the liver, pancreas, abdomen, retroperitoneal lymph nodes, kidney, and/or adrenal glands. The ratio of median TB above the diaphragm to that below the diaphragm was 2.2.
Baseline Characteristics in Relation to Clinical Outcome
Baseline factors were analyzed to determine their relation with PFS and OS via both univariable and multivariable analyses, and the univariable data are summarized in Table 3. On univariable analysis, CCF TKI risk group (P < .001), ≤2 sites of disease (P = .004), ≤6 metastases overall (P = .01), ≤8 metastases excluding bone (P = .02), TB ≤13 cm (P = .02), TB below the diaphragm <3 cm (P = .02), disease confined to above the diaphragm (P = .02), and a higher ratio of the TB above the diaphragm (P = .03) were found to be associated with improved PFS. On multivariable analysis, disease confined to above the diaphragm (P = .03) (Fig. 1 Top), total TB ≤13 cm (P = .09) (Fig. 1 Bottom), and CCF TKI risk group (P < .001) were found to be independent predictors of PFS.
Table 3. Univariable Analyses: Tumor Burden Parameters and PFS and OSa
Tumor burden parameters are presented using the categorization found to be optimal for the particular outcome. However, if a parameter was not associated with an outcome at all, the categorization used with the other outcome is presented. For example, the number of bony metastases was not associated with PFS, but was found to be prognostic for OS. Therefore, the categorization used in the OS analysis is also presented in the PFS analysis.
The feature with the poorer outcome is listed first except as noted.
P value was determined using the Wald test. Bold P values indicate statistical significance.
Higher burden was found to be associated with poorer OS.
Location of soft tissue disease (below the diaphragm ± above vs only above)
No. of organ sites involved (>2 vs ≤2)
Total no. of metastases (>6 vs ≤6)
No. of bony metastases (>2 vs ≤2)
No. of nonbony metastases (>8 vs ≤8)
Total soft tissue tumor burden (>13 vs ≤13 cm)
Total soft tissue tumor burden above the diaphragm (>6.5 vs ≤6.5 cm)
Total soft tissue tumor burden below the diaphragm (≥3 vs <3 cm)
Ratio of soft tissue burden above to below the diaphragm
Total soft tissue tumor burden above the diaphragm (>6.5 vs ≤6.5 cm)
Total soft tissue tumor burden below the diaphragm (≥3 vs <3 cm)
Ratio of soft tissue burden above to below the diaphragm
Similar results were observed with respect to OS. Several TB-related factors were found to be associated with outcome on univariable analysis: CCF TKI risk group (P < .001), total number of metastases <10 (P < .001), ≤2 bone metastases (P < .001), ≤4 metastatic sites (P < .001), ≤8 metastases excluding bone (P = .005), lower total TB (P = .01), and TB above the diaphragm ≤6.5 cm (P < .001). However, on multivariable analysis only total number of metastases <10 (P < .001) (Fig. 2 Top) and tumor volume above the diaphragm ≤6.5 cm (P = .05) (Fig. 2 Bottom) were found to be independently associated with OS, as was CCF TKI risk group (P = .003). Thus, accounting for established prognostic variables, baseline amount and location of TB remained significantly associated with clinical outcome.
TB Characteristics at Best Response in Relation to Clinical Outcome
The overall RECIST-defined OR rate was 52%, with a median time to best response of 3.2 months (range, 0.9 months-22.1 months). Approximately 87% of patients had some degree of TS, with a median TS of 31%. No baseline TB characteristics were found to be correlated with OR or TS. To address whether TS while receiving sunitinib was associated with OS, maximum TS was treated as a time-varying covariate in a proportional hazards model. From this analysis, increased maximum TS while receiving sunitinib was found to be significantly associated with OS (P < .001).
TB Characteristics at Time of PD in Relation to Clinical Outcome
Approximately 86% of patients had developed PD at the time of data analysis. The median survival from the time of RECIST-defined PD was 13.1 months (95% confidence interval, 7.4 months–20.8 months). With the exception of 1 patient who developed acute cholecystitis and required surgery, all other patients discontinued sunitinib at the time of RECIST-defined PD. Volume increases at existing sites of metastases were noted in 83% of patients, 46% of patients had metastases in different organs at the time of PD, and 52% of patients had new metastases in previously involved organs. Table 4 compares the pattern of disease involvement at baseline and at time of PD. Bone, mediastinal lymph nodes, and the brain had significantly more involvement at time of PD compared with at baseline (P = .05, .05, and .005, respectively). The most common sites of new organ involvement were the spine (20%), brain (15%), and lymph nodes (mediastinal, 8% and retroperitoneal, 8%). Lung, bone, and lymph nodes (mediastinal and retroperitoneal) were the most commonly involved sites both at baseline and at PD. Tumor location at the time of PD was not found to be associated with OS from the time of PD. However, total TB (P = .003) and total number of metastatic deposits (≤12 vs >12; P < .001) at the time of PD were found to be significantly associated with OS from the time of PD.
Table 4. Patterns of PD in Patients With mRCC Receiving Sunitinib
No. of Patients With Organ Involvement at Baseline (N=59)
No. of Patients With Organ Involvement at PD (N=59)
A deeper understanding of RCC characteristics in relation to outcome with VEGF-targeted therapy is desired to optimize clinical outcomes with targeted therapy. In this retrospective analysis of patients with mRCC who were treated with sunitinib, several features of TB were found to be associated with clinical outcome after adjustment for known prognostic variables.
At baseline, a lower total TB, fewer metastatic sites, and a lower TB above the diaphragm were significantly associated with improved survival. This may reflect an inherently more indolent underlying biology because patients who start therapy with less TB are likely to have slower growing disease. However, these factors remained significant even after adjusting for established baseline prognostic factors, which further supports the observed association. These data may also have implications for the timing of the initiation of systemic therapy. There is a subset of patients with mRCC who present with low-volume, slow-growing metastatic disease. Clinical practice in a subset of such patients could include an initial period of observation until an as yet undefined TB threshold or pace of growth is reached. Recent data have emerged suggesting an association between calculated growth rate constants and survival in mRCC patients.9, 10, 12 It is interesting to note that data from 1 of these trials found baseline TB to have a strong correlation with OS but found an even stronger correlation with OS when growth rate constant and initial TB were combined.10 Although it requires prospective study, this concept supports the notion that treatment decisions based on the growth and/or volume of disease in patients with mRCC may be plausible (eg, delaying the initiation of therapy or allowing treatment holidays). This is particularly important given the largely noncurative nature and toxicities of existing systemic therapies. The biology of a lower TB above the diaphragm at baseline relating to better outcomes is poorly understood. Given that disease above the diaphragm is more prevalent in this cohort, it may simply be a surrogate for overall TB. Conversely, different metastatic sites may indeed have an inherently different biology as related to response to VEGF inhibition, ultimately affecting clinical outcome. This is observed through the empiric observation that patients with liver metastases tend to have a poorer prognosis in some series compared with patients with lung or mediastinal involvement. This, in part, may account for why disease confined to above the diaphragm was found to be an independent predictor of PFS. From these observations, it is clear that the biology of individual organ metastases in patients with RCC requires further investigation.
In addition, these data suggest that a sunitinib-induced decrease in TB is associated with a reduction in the risk of death. This supports what has been increasingly recognized with the routine use of VEGF-targeting agents, namely, that TB reduction not meeting traditional RECIST criteria for an OR (≥30%) can be associated with clinical benefit.13 Furthermore, these data support that the degree of TB reduction with therapy remains an important predictor of outcome.
When we assessed patterns of PD, several interesting features emerged. Although progression of existing sites accounted for the majority of cases of PD, new organ sites, namely the bone and brain, represented a substantial and significant percentage of sites of PD. These data support in part an organ-specific effect of VEGF-targeting agents, and possibly that these sites may be sanctuaries in which the effect of VEGF targeting has a reduced antitumor effect. To our knowledge, it remains unclear how to manage a patient with an isolated site of PD (eg, bone or brain) with otherwise adequate control of the remaining TB, and further investigation is needed. Another plausible hypothesis is that greater overall tumor control with sunitinib allows patients to develop metastases in organ sites (eg, bone and brain) that would otherwise not have been observed with shorter survival. Tumor location at the time of PD was not found to be associated with survival, but greater TB and a greater number of disease sites were significant, underscoring the importance of amount of tumor over tumor location.
There are several limitations to the current study. This was a retrospective review of a small number of patients who were treated at a single RCC referral center and therefore is subject to the inherent limitations of such a study, including limited power, patient selection bias, and variable scan interval. In addition, evaluation of the amount and location of tumor and TB reductions were assessed by investigators, with radiology reports used as reference. Such an analysis may not accurately capture the exact locations of all tumors or the amount of tumor at each site. Finally, tumor necrosis without size change is believed to be indicative of an antitumor effect of sunitinib and other targeted agents, but such change would not be reflected in TB measurements based on size.
Overall, the findings of the current study provide insight into how TB characteristics over time influence outcomes in patients with mRCC who are treated with sunitinib. These data require further evaluation in larger, prospectively defined data sets to integrate them into the routine management of patients with mRCC.
CONFLICT OF INTEREST DISCLOSURES
Ms. Wood has acted as a member of the Speakers' Bureau for Bayer/Onyx Pharmaceutical Corporation, Genentech Pharmaceutical Corporation, Novartis Pharmaceutical Corporation, and Pfizer Pharmaceutical Corporation. Dr. Garcia has received research funding from Novartis, Pfizer, Genentech, Wyeth, Celgene, Imclone, and Amgen Pharmaceutical Corporations and has acted as a paid consultant for Novartis, Pfizer, and Genentech and as a member of the Speakers' Bureau for Novartis and Pfizer. Dr. Rini has acted as a consultant to and received research funding from Pfizer Pharmaceutical Corporation.