- © 2004 by American Society of Clinical Oncology
Interleukin-6, Interleukin-10, and Vascular Endothelial Growth Factor in Metastatic Renal Cell Carcinoma: Prognostic Value of Interleukin-6—From the Groupe Français d'Immunothérapie
- Sylvie Negrier,
- David Perol,
- Christine Menetrier-Caux,
- Bernard Escudier,
- Michel Pallardy,
- Alain Ravaud,
- Jean-Yves Douillard,
- Christine Chevreau,
- Christine Lasset and
- Jean-Yves Blay
- From the Cytokines and Cancer Research Unit, INSERM U.590, Medical Oncology Department, and Public Health Department, Centre Léon Bérard; Hôpital Edouard Herriot, Lyon; Institut Gustave Roussy, Villejuif; Institut Bergonié, Bordeaux; Centre Gauducheau, Nantes Saint-Herblain; Centre Claudius Régaud, Toulouse, France
- Address reprint requests to Sylvie Négrier, MD, Cytokines and Cancer Research Unit, INSERM U.590 & Medical Oncology Department, Centre Léon Bérard, 69373 Lyon Cedex 08, France; e-mail: negrier{at}lyon.fnclcc.fr
Abstract
Purpose Few clinical prognostic factors have been identified for patients with metastatic renal cell carcinoma (MRCC), and no biomarker is known in this disease. Several endogenous cytokines have demonstrated interesting and significant correlations with survival in these patients. Our objective was to analyze the prognostic value of circulating vascular endothelial growth factor (VEGF), interleukin-10 (IL-10), and interleukin-6 (IL-6).
Patients and Methods Serum levels of IL-6, IL-10, and VEGF were measured in patients with MRCC. Their prognostic value for response to treatment and progression-free and overall survival was evaluated. Pretreatment samples were obtained from 138 patients of a large randomized multicentric trial. Endogenous cytokine levels were determined using immunoassays. Univariate and multivariate analyses were performed to evaluate the prognostic value of each factor further controlled by an internal validation test. Threshold values for serum IL-6 and VEGF were determined using the quartile method.
Results Serum IL-6 was detectable in 70% of the patients. IL-10 and VEGF were elevated in 8% and 71% of the patients, respectively. None of these circulating factors was correlated with response to treatment. IL-10 was not significantly correlated with progression-free or overall survival. Despite significant correlation with survival, VEGF was not an independent prognostic factor in the multivariate analysis. Finally, IL-6 was significantly correlated with progression-free survival and overall survival, and has prognostic value for overall survival.
Conclusion Circulating IL-6 level appears to be an important independent prognostic factor in patients with MRCC; if confirmed in further studies, it could be considered for treatment decisions in these patients.
INTRODUCTION
Although renal cancer represents only 3% of all adult malignancies, its incidence increased by 35% from 1973 to 1991.1 About one third of patients diagnosed with renal cancer will die from the disease after the occurrence of metastases. Metastases from renal cancer are, indeed, resistant to all forms of chemotherapy tested until now.2,3
A few durable and complete remissions have been achieved by immunotherapy using mainly interferon alfa (IFN-α) or interleukin-2 (IL-2) and, more recently, by the way of allogeneic lymphoid transplants.4,5 However, only a small minority of patients experienced therapeutic benefit, whereas most patients were concerned with the important adverse effects of the treatments.
The Groupe Français d'Immunothérapie conducted a large multicentric trial, the CRECY trial, to evaluate the interest of IL-2 and/or IFN-α treatment in these patients.6 One of the secondary objectives of the study was to identify clinical prognosis factors that would help physicians select the patients who were candidates for these treatments. In addition to these clinical factors, the prognostic value of some routine blood tests or circulating factors was also analyzed.
Interleukin-6 (IL-6) is an autocrine tumor growth factor produced by renal tumor cells. Elevated serum levels of IL-6 have been correlated with a poor outcome in patients with metastatic renal carcinoma (MRCC).7-10 The prognostic value of serum IL-6 levels was therefore evaluated in this large prospective multicentric study, in addition to two other cytokines, vascular endothelial growth factor (VEGF) and interleukin-10 (IL-10), previously reported to be detectable in renal cell carcinoma patients.
IL-10 has immunosuppressive properties that facilitate the progression of different human tumors; some evidence has been obtained in renal carcinoma models11-14 and, more recently, in human tumor samples.15 As a consequence, IL-10 represented a putative prognostic factor of disease progression or survival in these patients.
VEGF is a strong angiogenic factor that has been found to be abnormally abundant in patients suffering from different types of cancers, including renal cell carcinomas.16-20 A correlation between elevated serum levels of VEGF and patient survival has been evidenced by some authors.19-20 In addition, due to the hyperexpression of pro-angiogenic factors, specially VEGF, under the action of hyper-expressed hypoxia inducible factors, renal cell carcinomas are hypervascularized tumors.21-24 Circulating VEGF levels could, therefore, represent an interesting indicator of the tumor behavior.
In the present study, serum levels of IL-6, IL-10, and VEGF were analyzed together with other clinical, blood, or circulating routine factors, some of them having previously been established as prognostic factors for renal carcinoma patients. The prognostic value of these parameters was tested in terms of response to treatment, progression-free survival, and overall survival.
This study had several objectives: first, to assay the prognostic value of these parameters in a prospective study; and second, to appreciate their potential clinical interest as compared to other clinical factors.
PATIENTS AND METHODS
Patients with histologically proven and clearly progressive MRCC who were found eligible and had given written consent were randomly assigned to intravenous (IV) IL-2 (Proleukin Chiron Therapeutics, Suresnes, France; arm A) or subcutaneous IFN-α (Roferon, Roche, Neuilly-sur-Seine, France; arm B), or a combination of IL-2 and IFN-α (arm C). Details of the treatments, as well as eligibility criteria, were previously reported.6 Response evaluation was performed 12 weeks after the beginning of treatment in each group.
Sample Analyses
Serum samples were obtained from 138 patients enrolled into the CRECY trial within the 2 weeks preceding treatment start, centrifuged, then immediately stored at −80°C until assay. Serum samples from a control population of healthy donors were also obtained and stored under the same conditions.
IL-6, VEGF, and IL-10 serum levels were determined using specific quantitative sandwich enzyme immunoassay technique (EASIA IL-6, Medgenix, Fleurus, Belgium; Quantikine® VEGF, R&D, Abingdon, UK; IL-10, Immunotech, Marseille, France). For the IL-10 assay, both human IL-10 and viral IL-10 were detected. The lowest cytokine concentrations significantly differed from the zero standard, with a 95% probability of 3 pg/mL, 9 pg/mL, and 5 pg/mL, respectively. The mean concentrations of these cytokines in apparently healthy human donors (a group of 12 volunteers was tested) were 6 pg/mL (range, 3 to 8.5 pg/mL) for IL-6, 224 pg/mL (range, 62 to 707 pg/mL) for VEGF, and under the detection limit for IL-10, respectively.
Statistical Analysis
Survival analysis.
The end point of interest was survival time, with two criteria: (1) overall survival, defined as the time from the date of randomization into the CRECY trial to the date of death or the date of last follow-up for patients alive at last contact. Since only eight deaths were not related to the disease in this subgroup, we used overall survival instead of cause-specific survival as an end point; (2) progression-free survival, defined as the time from the date of randomization to the date of disease progression or death, or the date of last follow-up for patients alive with no indication of disease progression. Follow-up was completed in September 1999. The median follow-up time was 66 months (95% CI, 58 to 74 months).
Univariate analysis.
To evaluate the relationship between survival and baseline characteristics, categoric and continuous variables were included in univariate Cox proportional hazard regression models.25 IL-10 was included in the models as a dichotomous variable taking the value 0 when under the detection limit, or 1 otherwise. This choice was simplified by the fact that 127 (92%) of the 138 patients had no detectable IL-10. Because of nonlinear relationships in the Cox models between IL-6, VEGF, and the log hazard, we chose to introduce these two factors as categoric variables in the survival analysis. IL-6 was included in the models in three categories (≤ 3 pg/mL, 3 to 35 pg/mL, > 35 pg/mL), and VEGF in two categories (≤ 755 pg/mL, > 755 pg/mL). These categories were defined by first determining the quartiles of the sample distributions of IL-6 and VEGF. The overall survival distributions of these quartiles were further examined using the Kaplan-Meier method.26 In addition, the hazard ratios for time-to-death and 95% CIs were estimated for each upper three quartiles (Q2, Q3, and Q4) of each cytokine compared with the lowest quartile (Q1) used as the reference group. Regarding IL-6, results in Figure 1 show evidence that patients in the second and third quartiles had a similar risk excess; both estimated hazard ratios were similar in magnitude, and confidence intervals had much overlap. These observations suggested that we should create a three-category variable, taking the value 0 when IL-6 level was in the first quartile, 1 in the second and third quartiles, and 2 in the fourth quartile. Similarly, results in Figure 2 suggested that we should create a dichotomous variable based on VEGF serum level as continuous, taking the value 0 if VEGF level was in the first three quartiles and 1 in the fourth.
Results of the quartile analysis of interleukin-6 serum levels (pg/mL), with Kaplan-Meier overall survival analysis and univariate Cox proportional hazard regression model (n = 138).
Results of the quartile analysis of vascular endothelial growth factor serum levels (pg/mL), with Kaplan-Meier overall survival analysis and univariate Cox proportional hazard regression model (n = 138).
Multivariate analysis.
Independent prognostic variables of overall survival and progression-free survival were respectively identified by a Cox regression analysis, using a backward stepwise procedure to eliminate noninfluential variables. We used a stepwise modeling algorithm with a 0.10 significance level for entering and 0.05 for removing explanatory variable. The adequacy of the models was tested and the proportional hazards assumption was found to be reasonable for each model.27 Notably, since the results of the randomized CRECY trial had failed to demonstrate any significant difference in survival, we performed the multivariate survival analysis without adjustment on the treatment group of the CRECY trial. However, in the last step of the model, we verified that the association of the different prognostic factors with survival was independent of treatment. Data obtained were analyzed using SPSS 10.0 statistical software (SPSS Inc, Chicago, IL).
Bootstrapping
To internally validate the accuracy of the final multivariate models, a bootstrap resampling technique was used according to the two-step method of Chen and George,28 generalized by Sauerbei and Schumacher.29 One hundred bootstrap samples were created in the first validation step, 200 in the second. The first step formulated two models that included only the major prognostic factors for overall and progression-free survival, ie, the variables that were selected more than 65% of the time from the 100 samples. For each survival criteria analyzed, the results confirmed the variables chosen in the original procedure, with five and four variables, respectively, in the final models for overall and progression-free survivals (data not shown). In the second step, the final models were refit and regression parameters (bootstrapped hazard ratios and 95% CIs were estimated). Estimates were compared with those obtained in the final Cox models. The bootstrapping procedures were performed using S-PLUS statistical package (Insightful Corporation, Seattle, WA).
Response to Treatment
Factors predictive of response to treatment were also investigated. Univariate analysis and multivariate logistic regression analysis30 were performed using the backward logistic procedure of SPSS software to identify independent prognostic factors for response. Biologic factors were tested as categoric variables with the same categorization as previously validated in the survival analysis.
RESULTS
Four hundred twenty-five patients from 25 different French centers were enrolled in the CRECY trial over 4 years. Pretreatment serum samples from 138 patients originating from eight different centers were available for IL-6, IL-10, and VEGF tests. The characteristics of these patients are detailed in Table 1. In order to avoid biases due to the important amount of missing data, the analysis of prognostic factors was performed in the subgroup of the 138 patients with available serum samples. We verified that this subgroup did not differ from the other 287 patients enrolled in the CRECY trial (ie, without serum samples available). The different characteristics of both groups were tested in univariate analysis, and no significant differences were detected. Progression-free survival and overall survival also appear similar in both groups (log-rank test: P = .88 and 0.99, respectively). The median survival time of both groups is 13 months (95% CI, 9 to 17 months).
Main Characteristics of Patients (N = 138)
IL-10 levels were mostly not detectable, except in 11 patients (8%). The levels of VEGF and IL-6 were found abnormally high (above the mean concentration observed in healthy donors) in 98 (71%) and 96 patients (70%), respectively.
The median levels and ranges (pg/mL) of endogenous cytokines were 0 (range, 0 to 144 pg/mL), 13 (range, 0 to 2,100 pg/mL), and 435 (range, 0 to 3,413 pg/mL) for IL-10, IL-6, and VEGF, respectively.
Prognostic Value of the Relevant Factors
Few factors appeared to be correlated with the tumor regression induced by treatment, and none of the biologic parameters were significantly correlated with the response to treatment (data not shown). Only the presence of one metastatic organ site (P = .05) and treatment with a combination of IL-2 and IFN-α (P < .05) were confirmed as independently predictive of response to treatment.
Progression-free and overall survival correlated with a number of parameters, notably IL-6 and VEGF levels, with a high degree of significance (Table 2). No significant correlation was found with IL-10 levels. After multivariate analysis, neither IL-6 nor VEGF appear as independent prognostic factors for progression-free survival (Table 3). The predictive ability of the multivariate model is appreciable; the bootstrapped and nonbootstrapped hazard ratios are similar, and the bias/bootstrap SE is less than 5%.
Results of the Univariate Survival Analysis Testing 25 Factors for Significant Impact on Overall Survival and Progression-Free Survival
Results of the Multivariate Analysis for Progression-Free Survival (N = 138)
Of note, the serum VEGF level is correlated with the number of platelets (Spearman's correlation coefficient: 0.50; P < .001), as well as with the number of circulating neutrophils (Spearman's coefficient: 0.40; P < .001).
Finally, six different parameters are evidenced as independent prognostic factors for overall survival in this series (Table 4). The bootstrap results suggest that the predictive accuracy of the multivariate model is reasonable. A weight loss of more than 10% and a serum level of IL-6 above 35 pg/mL are linked with a more than three-fold increased probability of death, whereas an increased number of circulating neutrophils and the presence of liver metastases augmented the risk of death by more than two-fold. A moderate increase of serum IL-6 (3 to 35 pg/mL) is again found to be an independent prognostic factor for survival. Patients with elevated serum IL-6 levels have a median survival of 5 months (95% CI, 5 to 6 months) versus 13 (95% CI, 6 to 20 months) or 26 months (95% CI, 8 to 44 months) when IL-6 levels are between 3 and 35 pg/mL or under 3 pg/mL, respectively.
Results of the Multivariate Analysis for Overall Survival (N = 138)
DISCUSSION
Many different prognostic factors have been identified in patients with renal cell carcinoma. Some of the factors that were identified in several large retrospective and prospective series can be considered as validated and clinically relevant.6,31-35 Five different factors for survival were identified in at least three different series of patients: performance status, number of metastatic sites, metastasis-free interval, elevated C-reactive protein or erythrocyte sedimentation rate, and hemoglobin rate.36 On the contrary, other biologic factors, especially those that were described recently and require complex and very sensitive tests, gave contradictory results in prognostic studies. As recently indicated by other authors,37 different methodologic weaknesses, including limited sample size, inadequately reproducible assays and inappropriate multiple significance testing, may explain why any of these factors had first been considered of clinical interest.
In this study, we analyze the potential prognostic value of three different circulating factors that had previously been of interest in renal cell tumors. These results were prospectively obtained in 138 patients with histologically proven metastases from renal carcinomas.
In our series, serum IL-10 levels were found elevated in a limited number of patients with MRCC; they showed no significant correlation with the response status or with survival. These findings differ from previously published results;38 however, these previous results were obtained on 80 patients from a larger series, and no analysis was performed to verify whether this subgroup was representative of the overall study population. The role of IL-10 in renal tumor immune escape was previously suggested by several authors.11-15 The present study demonstrates that general IL-10 production does not influence response to immunotherapy with recombinant cytokines in MRCC patients and is not an important issue for the outcome of the patients.
Three different studies have shown that serum IL-6 levels are highly correlated with survival in patients with renal cell carcinoma.7-9 However, in two studies that included a multivariate analysis, IL-6 failed to appear as an independent prognosis factor. In the first study, the number of patients was limited (N = 22).8 The population of the second study was larger but heterogeneous, since it mixed metastatic patients and patients with localized disease only. The correlation between survival and IL-6 was only significant in the subgroup with metastases.9 Our results show that elevated IL-6 levels seen in 70% of the patients correlated with overall survival and progression-free survival. According to two different threshold values, IL-6 appears as an independent prognostic factor for overall survival, and high levels of serum IL-6 (> 35 pg/mL) increase the risk of death by a factor close to four (3.96).
This is the first demonstration of the prognostic value of serum IL-6 in metastatic renal cell carcinoma. However, it is our opinion that the results are robust and relevant for these patients. Indeed, our results were verified using an internal validation test (ie, the bootstrap technique). This strongly supports a direct role of IL-6 on the behavior of MRCC. However, it is difficult to identify the main mechanisms involved; an immunosuppressive effect on dendritic cells, a pro-inflammatory role, and a growth factor effect could all contribute to these observations14,39,40
IL-6 is a multifunctional cytokine, inducing hematopoietic, pro-inflammatory, and complex immunomodulatory effects.41 As a consequence, the deleterious effect of augmented levels of circulating IL-6 may be related to the pro-inflammatory effects of the cytokine, which include induction of fever, weight loss, and specific biologic abnormalities related to inflammation. In addition, several authors previously described IL-6 as an important autocrine growth factor in renal cell carcinomas.42-45 Finally, the inhibitory activity of IL-6 on the differentiation of dendritic cells, as described in the renal cell carcinoma model, could also enhance the adverse effect of IL-6 in these patients.39 The adverse effect of in vivo IL-6, circulating at high levels, could therefore be accounted by at least three different biologic mechanisms. Finally, whereas circulating IL-6 is clearly related to tumor progression and, consequently, to survival, there is no link with response to treatment. This means that tumor regression induced by IL-2 and/or IFN-α does not depend on serum levels of IL-6, but on other mechanisms.
Various authors have demonstrated that some patients with renal carcinoma have abnormally elevated serum levels of VEGF; this correlated with poor survival.19-20 Obviously, the prognostic value of serum VEGF increased with the stage of the disease, but VEGF was not considered as an independent prognostic factor in any of these series, which included a limited number of patients with metastases.18,20 VEGF rates were found elevated in 71% of our patients, and augmented serum VEGF levels were significantly correlated with progression-free and overall survival. After multivariate analysis, elevated rates of VEGF no longer appear as an independent prognostic factor for survival.
We must note that a paraneoplastic thrombocytosis is present in 30% of the patients in our series; it is also significantly correlated with a shorter survival. In addition, and as expected from previous reports,19,46 there is a significant correlation between VEGF levels and the number of platelets (correlation rate: 0.670). Therefore, it would be important to confirm these results on platelet-poor plasma in order to detect circulating free VEGF only.47
The number of circulating neutrophils, which had also been identified as a prognostic factor for survival in uterine cervix carcinoma, non-small-cell lung cancer, and renal cancer,34,48,49 is a potent prognostic factor in these patients. In our series, an elevated neutrophil count is correlated with very high levels of serum VEGF, but the elevated number of neutrophils can also be related to the secretion of other cytokines, such as interleukin-8 or granulocyte macrophage colony-stimulating factor, since both are produced by some renal tumors.50,51 The role of abnormally high numbers of circulating neutrophils in renal tumor progression requires further investigations.
In conclusion, this study confirms the correlation between limited progression-free survival or short overall survival and augmented levels of serum IL-6 and VEGF. In addition, elevated serum levels of IL-6 appear as an important and independent prognostic factor in patients with MRCC. If confirmed in further studies, this parameter could be considered for treatment decisions in patients with MRCC.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgments
We thank Adeline Duc for excellent technical assistance, and Marie-Dominique Reynaud for editorial help.
Footnotes
-
Supported by a grant from the Comité de Saône et Loire of the French Ligue Nationale contre le Cancer.
Authors' disclosures of potential conflicts of interest are found at the end of this article.
- Received June 23, 2003.
- Accepted March 17, 2004.
