Validation and Extension of the Memorial Sloan-Kettering Prognostic Factors Model for Survival in Patients With Previously Untreated Metasta
http://www.100md.com
《临床肿瘤学》
the Taussig Cancer Center, The Cleveland Clinic Foundation, Cleveland, OH
ABSTRACT
PATIENTS AND METHODS: Data were collected on 353 previously untreated metastatic RCC patients enrolled onto clinical trials between 1987 and 2002.
RESULTS: Four of the five prognostic factors identified by Motzer were independent predictors of survival. In addition, prior radiotherapy and presence of hepatic, lung, and retroperitoneal nodal metastases were found to be independent prognostic factors. Using the number of metastatic sites as surrogate for individual sites (none or one v two or three sites), Motzer’s definitions of risk groups were expanded to accommodate these two additional prognostic factors. Using this expanded criteria, favorable risk is defined as zero or one poor prognostic factor, intermediate risk is two poor prognostic factors, and poor risk is more than two poor prognostic factors. According to Motzer’s definitions, 19% of patients were favorable risk, 70% were intermediate risk, and 11% were poor risk; median overall survival times for these groups were 28.6, 14.6, and 4.5 months, respectively (P < .0001). Using the expanded criteria, 37% of patients were favorable risk, 35% were intermediate risk, and 28% were poor risk; median overall survival times of these groups were 26.0, 14.4, and 7.3 months, respectively (P < .0001).
CONCLUSION: These data validate the model described by Motzer et al. Additional independent prognostic factors identified were prior radiotherapy and sites of metastasis. Incorporation of these additional prognostic factors into the Motzer et al model can help better define favorable risk, intermediate risk, and poor risk patients.
INTRODUCTION
Unfortunately, approximately one third of patients have metastatic disease at the time of diagnosis, and approximately 50% of patients undergoing potentially curative surgery for less advanced disease can be expected to relapse distantly.4,5 Coupled with the lack of effective systemic therapy and the highly variable natural history of RCC, the poor outlook for patients with metastatic disease highlights the need to better define patient and disease factors that are associated with outcome. Identification of a reliable, validated prognostic model for outcome in patients with metastatic RCC will yield an important tool that can be used to help optimize patient selection for specific treatment strategies and aid in the interpretation of clinical trials by helping to determine the extent to which therapy is impacting the natural history of the disease.
Reports in the literature vary with respect to the identification of patient and disease characteristics that are prognostic for survival in patients with metastatic RCC.7–15 Although some factors have consistently been found to be of prognostic value (eg, performance status [PS]), all of the reports have evaluated different sets of factors, and definitions of some factors, such as metastasis-free interval, have varied.
Recently, Motzer et al13 identified five prognostic factors that correlated with overall survival in patients with metastatic RCC treated with interferon alfa as initial systemic therapy. The factors were Karnofsky PS, time from diagnosis of RCC to treatment with interferon alfa, serum lactate dehydrogenase, corrected serum calcium, and hemoglobin. Using a dichotomized version of each factor and giving them all equal weight, Motzer et al stratified patients into three different risk groups (favorable, intermediate, and poor risk) depending on the number of poor prognostic factors present (Table 1). The model was validated internally using a bootstrap resampling procedure; however, external validation in an independent set of patients would provide a valuable confirmation of the model before it is adopted and used in the design of future clinical trials in this disease.
The primary goal of this investigation was to validate the model developed by Motzer et al13 in an independent group of patients using survival as the primary end point. However, given the varied reports of prognostic factors for survival in the literature, we also considered all previously reported factors that are readily available as part of the patient’s normal work-up to determine whether additional independent prognostic factors could be identified that could be used to extend or modify the model.
PATIENTS AND METHODS
Survival distributions were estimated using the Kaplan-Meier method.17 The relationship between survival and the factors listed in Table 2 were analyzed using the log-rank test18 and the Cox proportional hazards model.19 Clinical and pathologic characteristics that were categoric by nature, such as sex, PS, and histology, were individually 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 individually analyzed as continuous variables using the Cox proportional hazards model and as categoric variables using the log-rank test. The cut points used for categorizations were based on cutoffs previously described in the literature and/or recursive partitioning. The Cox proportional hazards model with stepwise variable selection was used to simultaneously assess multiple factors. A significance level of P = .10 was used as the criterion for determining variable entry and removal from the model. Because patients were treated over a fairly long period of time, the stratified version of the model was used to adjust for any inherent changes in prognosis over time. The treatment periods used to define the strata were 1987 to 1991, 1992 to 1996, and 1997 to 2002. For convenience, only the categoric forms of continuous variables were included in the multivariable analyses.
RESULTS
Tables 4 and 5 list the patient and disease characteristics and the biochemical factors examined. Seventy-three percent of the patients were male, and median age at diagnosis was 54 years (range, 23 to 76 years). Most patients (81%) had prior nephrectomy and most were entered onto a clinical trial within a few months of their initial diagnosis (median, 4.2 months). By design, patients tended to have good PS (all were ambulatory with ECOG PS of 0 or 1), and few patients (4%) had CNS metastases.
Although an attempt was made to capture the histology and nuclear grade of the primary tumor, complete information was often not available. Of the 267 patients (87%) with histology available, 85% had clear-cell tumors, and 15% had other histologies, primarily papillary and sarcomatoid features. One hundred seventy-one patients had information on nuclear grade; 28% of these tumors were grade 1 or 2, and 72% were grade 3 or 4.
Median survival time for the 308 patients studied was 14.8 months. Eighty-one percent of the patients had died by the time of analysis, and 19% were still alive or had been lost to follow-up. Median follow-up time for these 60 patients was 17.9 months (range, 1 month to 14.1 years).
To validate the model proposed by Motzer et al13 and to determine whether additional factors could be identified to either extend or otherwise modify this model, a stepwise stratified Cox proportional hazards model, which considered the categoric forms of the factors listed in Tables 4 and 5, was used. However, histology and nuclear grade were not considered at this stage because of the large proportion of patients missing this information. Prior nephrectomy and total serum calcium were also not considered because prior nephrectomy is highly correlated with time from diagnosis to entry onto study (only one of the 58 patients who had prior nephrectomy was diagnosed with RCC more than 12 months before entering a clinical trial), and corrected calcium37 is highly correlated with total calcium (Pearson’s correlation coefficient, r = 0.92). The results of this analysis are listed in Tables 6 and 7. Using a significance level of P = .10 for determining variable entry into and deletion from the model, four of the five factors identified by Motzer et al were again identified as being independent prognostic factors for survival, time from diagnosis to entry onto study, hemoglobin, corrected serum calcium, and serum lactate dehydrogenase. PS, which was an important predictor in the Motzer et al model, was not found to be a statistically significant predictor. This was not surprising, however, because all patients in the current series had an ECOG PS of 0 or 1, and therefore, all patients had a favorable PS based on Motzer et al’s categorization.
Using the risk groups as defined by Motzer et al,13 58 patients (19%) had no poor prognostic factors present and, therefore, were categorized as favorable risk; 70% of patients had one or two poor prognostic factors present and, therefore, were considered intermediate risk; and 11% of patients were categorized as poor risk because more than two poor prognostic factors were present. Median survival times were 28.6, 14.6, and 4.5 months for the favorable, intermediate, and poor risk groups, respectively (P < .0001). These figures are similar to those reported by Motzer et al; 18% of 437 patients analyzed were favorable risk and had a median survival of 29.6 months, 62% of patients belonged to the intermediate risk group, which had a median survival of 13.8 months, and 20% of patients were considered poor risk and had a median survival of 4.9 months. Survival curves for the three risk groups as defined by Motzer et al are given in Figure 1.
In addition to the prognostic factors identified by Motzer et al,13 prior radiotherapy (P < .001) and the presence of hepatic metastases (P < .001), metastases to the lung (P = .003), and retroperitoneal nodal metastases (P = .04) were also observed to have a negative impact on survival. As a surrogate for the individual metastatic sites, the number of sites was also considered. Replacing the individual metastatic sites with the number of sites involved (zero or one v two or three sites) did not result in any appreciable loss of information based on the log partial likelihood, and therefore, in the final model, the number of metastatic sites was used rather than the individual sites. In addition to being an excellent surrogate for the individual sites of metastatic disease, the number of sites involved also provides an easy way to extend the Motzer et al model. That is, by again simply counting the number of poor prognostic factors present, three new risk groups can be defined. The favorable risk group now contains patients with zero or one poor prognostic factor, the intermediate group contains patients with two poor prognostic factors present, and the poor risk group contains patients with three or more poor prognostic factors present.
The result of applying the extended model and the new definitions of risk groups is summarized in Tables 7 and 8. On the basis of the extended model and the new risk group definitions, the favorable risk group is comprised of 37% of the 308 patients and has an estimated median survival of 26.0 months. Thirty-five percent of patients now fall into the intermediate risk group, which has a median survival of 14.4 months, and 28% of patients are considered poor risk and have a median survival of 7.3 months (P < .001).
As can be seen from Tables 7 and 8, the new stratification essentially identifies and reclassifies as favorable and poor risk the better and poorer prognosis patients originally considered to be intermediate risk. For example, from Table 8, 58 patients who were considered to be intermediate risk as originally defined by Motzer et al13 were considered favorable risk by the new extended model and definitions. One-year and median survival for these patients was 81% and 24.0 months, respectively, which is similar to the figures of 82% and 28.6 months observed for the 57 patients classified as favorable by both definitions. Similarly, 50 patients initially classified as intermediate risk are now considered poor risk. One-year and median survival for these patients were 29% and 8.1 months, respectively, compared with 16% and 4.5 months, respectively, for the 35 patients considered poor risk by both models and 58% and 14.4 months, respectively, for the 107 patients considered intermediate risk by the two models. The similarities in prognosis based on the two models are shown graphically in Figure 2.
Histology and nuclear grade were not considered initially because of the large number of patients for whom this information was not available. However, applying the extended model to the patients for whom histology and nuclear grade were available suggests that histology, but not nuclear grade, may impact on survival. Adjusting for the impact of the prognostic factors included in the extended model, patients with clear-cell tumors had a significantly better prognosis than patients with other histologies (P = .003, Fig 3).
DISCUSSSION
Although our study validates Motzer et al’s model and their criteria for defining risk groups, a difficulty with the model is that the majority of patients are classified as intermediate risk, and relatively few patients are considered as favorable or poor risk. In the present study, 70% of patients were classified as intermediate risk, and only 19% were considered favorable risk. Still fewer patients (11%) were considered poor risk. These proportions are similar to the distribution of risk groups reported by Motzer et al (ie, 18%, 62%, and 20% for favorable, intermediate, and poor risk, respectively). The disproportionately large number of patients in the intermediate risk group suggests that it may be somewhat heterogeneous with respect to outcome.
In the present study, intermediate risk patients with a single poor prognostic factor (n = 99) did have a significantly better prognosis than intermediate risk patients (n = 116) with two poor prognostic factors present (P = .003). Extension of the model by incorporating prior radiotherapy and the number of metastatic sites into the definition of risk groups overcomes this difficulty because the expanded definitions essentially reclassify the better prognosis subset of intermediate risk patients as favorable and the poorer prognosis subset as poor risk. This is true even if one separates the intermediate risk group into two subgroups based on whether one or two poor prognostic factors are present. That is, of the 99 patients with one poor prognostic factor based on the original definition, 58 are classified as favorable risk based on the expanded criteria. Median survival time for these patients was 24.0 months, which is similar to the median survival time of 28.6 months observed in the 57 patients who were categorized as favorable risk by both models. Similarly, of the 116 patients with two poor prognostic factors present, 42 are considered poor risk by the expanded criteria. Median survival time for these patients was 8.4 months, which is similar to the median survival time of 4.5 months observed for the 35 patients considered poor risk by both models. One hundred seven patients considered intermediate risk by both models were similar regardless of the number of poor prognostic factors present based on Motzer et al criteria. Median survival time was 14.8 months for the 33 patients with one poor prognostic factor present and 14.4 months for the 74 patients with two poor prognostic factors present.
Another finding in the present study is that, although pathologic features of the primary tumor were often not available, correcting for the factors in the extended model, patients with clear-cell tumors seem to have a significantly better prognosis than patients with other histologies (P = .003). Although the proposed extension to the prognostic model proposed by Motzer et al13 seems to improve the model’s discriminatory power and patients with clear-cell tumors seem to have a better prognosis than patients with tumors of other histologies, these results are based on the retrospective analysis of highly selected patients, and confirmation of the results is needed.
In conclusion, six prognostic factors were identified for predicting survival in patients with RCC and most have been validated in different studies. We were able to validate Motzer’s data and find additional prognostic factors (prior radiation therapy and number of sites involved) to also be independent prognostic factors in the survival of patients with previously untreated RCC. This can be helpful in refining the definition of intermediate and high risk groups. The good discriminatory power that risk group status seems to have indicates that these are important factors that should be considered in the management of patients with advanced RCC and in the design and analysis of future clinical trials. An international consortium of investigators has been organized to further examine prognostic factors in patients with metastatic and untreated RCC and to develop a common approach.
Authors’ Disclosures of Potential Conflicts of Interest
NOTES
Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 31-June 3, 2003.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Greenlee RT, Murray T, Bolden S, et al: Cancer statistics, 2000. CA Cancer J Clin 50:7-33, 2000
2. Chow WH, Devesa SS, Warren JL, et al: Rising incidence of renal cell cancer in the United States. JAMA 281:1628-1631, 1999
3. Motzer RJ, Bander NH, Nanus DM: Renal-cell carcinoma. N Engl J Med 335:865-875, 1996
4. Figlin RA: Renal cell carcinoma: Management of advanced disease. J Urol 161:381-386, 1999
5. Linehan M, Shipley W, Parkinson D: Cancer of the Kidney and Ureter. Philadelphia, PA, Lippincott-Raven Publishers, 1997
6. Lanigan D: Prognostic factors in renal cell carcinoma. Br J Urol 75:565-571, 1995
7. Elson PJ, Witte RS, Trump DL: Prognostic factors for survival in patients with recurrent or metastatic renal cell carcinoma. Cancer Res 48:7310-7313, 1988
8. de Forges A, Rey A, Klink M, et al: Prognostic factors of adult metastatic renal carcinoma: A multivariate analysis. Semin Surg Oncol 4:149-154, 1988
9. Palmer PA, Vinke J, Philip T, et al: Prognostic factors for survival in patients with advanced renal cell carcinoma treated with recombinant interleukin-2. Ann Oncol 3:475-480, 1992
10. Fossa SD, Kramar A, Droz JP: Prognostic factors and survival in patients with metastatic renal cell carcinoma treated with chemotherapy or interferon-alpha. Eur J Cancer 30A:1310-1314, 1994
11. Lopez Hanninen E, Kirchner H, Atzpodien J: Interleukin-2 based home therapy of metastatic renal cell carcinoma: Risks and benefits in 215 consecutive single institution patients. J Urol 155:19-25, 1996
12. Motzer RJ, Mazumdar M, Bacik J, et al: Survival and prognostic stratification of 670 patients with advanced renal cell carcinoma. J Clin Oncol 17:2530-2540, 1999
13. Motzer RJ, Bacik J, Murphy BA, et al: Interferon-alfa as a comparative treatment for clinical trials of new therapies against advanced renal cell carcinoma. J Clin Oncol 20:289-296, 2002
14. Negrier S, Escudier B, Gomez F, et al: Prognostic factors of survival and rapid progression in 782 patients with metastatic renal carcinomas treated by cytokines: A report from the Groupe Francais d’Immunotherapie. Ann Oncol 13:1460-1468, 2002
15. Atzpodien J, Royston P, Wandert T, et al: Metastatic renal carcinoma comprehensive prognostic system. Br J Cancer 88:348-353, 2003
16. Miller AB, Hoogstraten B, Staquet M, et al: Reporting results of cancer treatment. Cancer 47:207-214, 1981
17. Kaplan E, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
18. Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163-170, 1966
19. Cox D, Oakes D: Analysis of Survival Data (ed 1). New York, NY, Chapman and Hall, 1990
20. Figlin RA, Thompson JA, Bukowski RM, et al: Multicenter, randomized, phase III trial of CD8(+) tumor-infiltrating lymphocytes in combination with recombinant interleukin-2 in metastatic renal cell carcinoma. J Clin Oncol 17:2521-2529, 1999
21. Bukowski RM, Olencki T, Wang Q, et al: Phase II trial of interleukin-2 and interferon-alpha in patients with renal cell carcinoma: Clinical results and immunologic correlates of response. J Immunother 20:301-311, 1997
22. Agrawal N, Olencki T, Mekhail T, Peereboom D, et al: A phase I/II study of GM-CSF, interleukin-2 (IL-2) and interferon-alpha (INF-alpha) in metastatic renal cell cancer. Proc Am Soc Clin Oncol 21:5b, 2002 (abstr 1829)
23. Hutson T, Mekhail T, Messerli E, et al: Phase I trial of PEG-Intron and IL-2 in patients with metastatic renal cell carcinoma. Proc Am Soc Clin Oncol 21:24a, 2002 (abstr 2406)
24. Olencki T, Malhi S, Mekhail T, et al: Phase I trial of thalidomide and interleukin-2 (IL-2) in patients with metastatic renal cell carcinoma. Proc Am Soc Clin Oncol 22:387, 2003 (abstr 1554)
25. Motzer RJ, Rakhit A, Thompson JA, et al: Randomized multicenter phase II trial of subcutaneous recombinant human interleukin-12 versus interferon-alpha 2a for patients with advanced renal cell carcinoma. J Interferon Cytokine Res 21:257-263, 2001
26. Hutson TE, Molto L, Mekhail T, et al: Phase I trial of consensus interferon in patients with metastatic renal cell carcinoma: Toxicity and immunological effects. Clin Cancer Res 9:1354-1360, 2003
27. Bukowski RM, Sergi JS, Sharfman WJ, et al: Phase I trial of natural human interferon beta in metastatic malignancy. Cancer Res 51:836-840, 1991
28. Plautz GE, Bukowski RM, Novick AC, et al: T-cell adoptive immunotherapy of metastatic renal cell carcinoma. Urology 54:617-623, 1999
29. Motzer RJ, Rakhit A, Schwartz LH, et al: Phase I trial of subcutaneous recombinant human interleukin-12 in patients with advanced renal cell carcinoma. Clin Cancer Res 4:1183-1191, 1998
30. Tate J, Olencki T, Finke J, et al: Phase I trial of simultaneously administered GM-CSF and IL-6 in patients with renal-cell carcinoma: Clinical and laboratory effects. Ann Oncol 12:655-659, 2001
31. Wos E, Olencki T, Tuason L, et al: Phase II trial of subcutaneously administered granulocyte-macrophage colony-stimulating factor in patients with metastatic renal cell carcinoma. Cancer 77:1149-1153, 1996
32. Thompson JA FR, Galanis E, Bukowski R, et al: Phase II trial of plasmid DNA/lipid (leuvectin) immunotherapy in patients with metastatic renal cell cancer. Proc Am Soc Clin Oncol 19:462a, 2000 (abstr 1814)
33. Olencki T, Peereboom D, Wood L, et al: Phase I and II trials of subcutaneously administered rIL-2, interferon alfa-2a, and fluorouracil in patients with metastatic renal carcinoma. J Cancer Res Clin Oncol 127:319-324, 2001
34. Chang DZ, Olencki T, Budd GT, et al: Phase I trial of capecitabine in combination with interferon alpha in patients with metastatic renal cancer: Toxicity and pharmacokinetics. Cancer Chemother Pharmacol 48:493-498, 2001
35. Perez-Zincer F, Olencki T, Budd GT, et al: A phase I trial of weekly gemcitabine and subcutaneous interferon alpha in patients with refractory renal cell carcinoma. Invest New Drugs 20:305-310, 2002
36. Schacter LP, Bukowski R, Carducci MA, et al: Safety profile of DHA-paclitaxel (TXP) in 8 phase II trials. Proc Am Soc Clin Oncol 22:214, 2003 (abstr 856)
37. Orrell D: Albumins as an aid to the interpretation of serum calcium. Clin Chim Acta 35:483-489, 1971(Tarek M. Mekhail, Rony M.)
ABSTRACT
PATIENTS AND METHODS: Data were collected on 353 previously untreated metastatic RCC patients enrolled onto clinical trials between 1987 and 2002.
RESULTS: Four of the five prognostic factors identified by Motzer were independent predictors of survival. In addition, prior radiotherapy and presence of hepatic, lung, and retroperitoneal nodal metastases were found to be independent prognostic factors. Using the number of metastatic sites as surrogate for individual sites (none or one v two or three sites), Motzer’s definitions of risk groups were expanded to accommodate these two additional prognostic factors. Using this expanded criteria, favorable risk is defined as zero or one poor prognostic factor, intermediate risk is two poor prognostic factors, and poor risk is more than two poor prognostic factors. According to Motzer’s definitions, 19% of patients were favorable risk, 70% were intermediate risk, and 11% were poor risk; median overall survival times for these groups were 28.6, 14.6, and 4.5 months, respectively (P < .0001). Using the expanded criteria, 37% of patients were favorable risk, 35% were intermediate risk, and 28% were poor risk; median overall survival times of these groups were 26.0, 14.4, and 7.3 months, respectively (P < .0001).
CONCLUSION: These data validate the model described by Motzer et al. Additional independent prognostic factors identified were prior radiotherapy and sites of metastasis. Incorporation of these additional prognostic factors into the Motzer et al model can help better define favorable risk, intermediate risk, and poor risk patients.
INTRODUCTION
Unfortunately, approximately one third of patients have metastatic disease at the time of diagnosis, and approximately 50% of patients undergoing potentially curative surgery for less advanced disease can be expected to relapse distantly.4,5 Coupled with the lack of effective systemic therapy and the highly variable natural history of RCC, the poor outlook for patients with metastatic disease highlights the need to better define patient and disease factors that are associated with outcome. Identification of a reliable, validated prognostic model for outcome in patients with metastatic RCC will yield an important tool that can be used to help optimize patient selection for specific treatment strategies and aid in the interpretation of clinical trials by helping to determine the extent to which therapy is impacting the natural history of the disease.
Reports in the literature vary with respect to the identification of patient and disease characteristics that are prognostic for survival in patients with metastatic RCC.7–15 Although some factors have consistently been found to be of prognostic value (eg, performance status [PS]), all of the reports have evaluated different sets of factors, and definitions of some factors, such as metastasis-free interval, have varied.
Recently, Motzer et al13 identified five prognostic factors that correlated with overall survival in patients with metastatic RCC treated with interferon alfa as initial systemic therapy. The factors were Karnofsky PS, time from diagnosis of RCC to treatment with interferon alfa, serum lactate dehydrogenase, corrected serum calcium, and hemoglobin. Using a dichotomized version of each factor and giving them all equal weight, Motzer et al stratified patients into three different risk groups (favorable, intermediate, and poor risk) depending on the number of poor prognostic factors present (Table 1). The model was validated internally using a bootstrap resampling procedure; however, external validation in an independent set of patients would provide a valuable confirmation of the model before it is adopted and used in the design of future clinical trials in this disease.
The primary goal of this investigation was to validate the model developed by Motzer et al13 in an independent group of patients using survival as the primary end point. However, given the varied reports of prognostic factors for survival in the literature, we also considered all previously reported factors that are readily available as part of the patient’s normal work-up to determine whether additional independent prognostic factors could be identified that could be used to extend or modify the model.
PATIENTS AND METHODS
Survival distributions were estimated using the Kaplan-Meier method.17 The relationship between survival and the factors listed in Table 2 were analyzed using the log-rank test18 and the Cox proportional hazards model.19 Clinical and pathologic characteristics that were categoric by nature, such as sex, PS, and histology, were individually 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 individually analyzed as continuous variables using the Cox proportional hazards model and as categoric variables using the log-rank test. The cut points used for categorizations were based on cutoffs previously described in the literature and/or recursive partitioning. The Cox proportional hazards model with stepwise variable selection was used to simultaneously assess multiple factors. A significance level of P = .10 was used as the criterion for determining variable entry and removal from the model. Because patients were treated over a fairly long period of time, the stratified version of the model was used to adjust for any inherent changes in prognosis over time. The treatment periods used to define the strata were 1987 to 1991, 1992 to 1996, and 1997 to 2002. For convenience, only the categoric forms of continuous variables were included in the multivariable analyses.
RESULTS
Tables 4 and 5 list the patient and disease characteristics and the biochemical factors examined. Seventy-three percent of the patients were male, and median age at diagnosis was 54 years (range, 23 to 76 years). Most patients (81%) had prior nephrectomy and most were entered onto a clinical trial within a few months of their initial diagnosis (median, 4.2 months). By design, patients tended to have good PS (all were ambulatory with ECOG PS of 0 or 1), and few patients (4%) had CNS metastases.
Although an attempt was made to capture the histology and nuclear grade of the primary tumor, complete information was often not available. Of the 267 patients (87%) with histology available, 85% had clear-cell tumors, and 15% had other histologies, primarily papillary and sarcomatoid features. One hundred seventy-one patients had information on nuclear grade; 28% of these tumors were grade 1 or 2, and 72% were grade 3 or 4.
Median survival time for the 308 patients studied was 14.8 months. Eighty-one percent of the patients had died by the time of analysis, and 19% were still alive or had been lost to follow-up. Median follow-up time for these 60 patients was 17.9 months (range, 1 month to 14.1 years).
To validate the model proposed by Motzer et al13 and to determine whether additional factors could be identified to either extend or otherwise modify this model, a stepwise stratified Cox proportional hazards model, which considered the categoric forms of the factors listed in Tables 4 and 5, was used. However, histology and nuclear grade were not considered at this stage because of the large proportion of patients missing this information. Prior nephrectomy and total serum calcium were also not considered because prior nephrectomy is highly correlated with time from diagnosis to entry onto study (only one of the 58 patients who had prior nephrectomy was diagnosed with RCC more than 12 months before entering a clinical trial), and corrected calcium37 is highly correlated with total calcium (Pearson’s correlation coefficient, r = 0.92). The results of this analysis are listed in Tables 6 and 7. Using a significance level of P = .10 for determining variable entry into and deletion from the model, four of the five factors identified by Motzer et al were again identified as being independent prognostic factors for survival, time from diagnosis to entry onto study, hemoglobin, corrected serum calcium, and serum lactate dehydrogenase. PS, which was an important predictor in the Motzer et al model, was not found to be a statistically significant predictor. This was not surprising, however, because all patients in the current series had an ECOG PS of 0 or 1, and therefore, all patients had a favorable PS based on Motzer et al’s categorization.
Using the risk groups as defined by Motzer et al,13 58 patients (19%) had no poor prognostic factors present and, therefore, were categorized as favorable risk; 70% of patients had one or two poor prognostic factors present and, therefore, were considered intermediate risk; and 11% of patients were categorized as poor risk because more than two poor prognostic factors were present. Median survival times were 28.6, 14.6, and 4.5 months for the favorable, intermediate, and poor risk groups, respectively (P < .0001). These figures are similar to those reported by Motzer et al; 18% of 437 patients analyzed were favorable risk and had a median survival of 29.6 months, 62% of patients belonged to the intermediate risk group, which had a median survival of 13.8 months, and 20% of patients were considered poor risk and had a median survival of 4.9 months. Survival curves for the three risk groups as defined by Motzer et al are given in Figure 1.
In addition to the prognostic factors identified by Motzer et al,13 prior radiotherapy (P < .001) and the presence of hepatic metastases (P < .001), metastases to the lung (P = .003), and retroperitoneal nodal metastases (P = .04) were also observed to have a negative impact on survival. As a surrogate for the individual metastatic sites, the number of sites was also considered. Replacing the individual metastatic sites with the number of sites involved (zero or one v two or three sites) did not result in any appreciable loss of information based on the log partial likelihood, and therefore, in the final model, the number of metastatic sites was used rather than the individual sites. In addition to being an excellent surrogate for the individual sites of metastatic disease, the number of sites involved also provides an easy way to extend the Motzer et al model. That is, by again simply counting the number of poor prognostic factors present, three new risk groups can be defined. The favorable risk group now contains patients with zero or one poor prognostic factor, the intermediate group contains patients with two poor prognostic factors present, and the poor risk group contains patients with three or more poor prognostic factors present.
The result of applying the extended model and the new definitions of risk groups is summarized in Tables 7 and 8. On the basis of the extended model and the new risk group definitions, the favorable risk group is comprised of 37% of the 308 patients and has an estimated median survival of 26.0 months. Thirty-five percent of patients now fall into the intermediate risk group, which has a median survival of 14.4 months, and 28% of patients are considered poor risk and have a median survival of 7.3 months (P < .001).
As can be seen from Tables 7 and 8, the new stratification essentially identifies and reclassifies as favorable and poor risk the better and poorer prognosis patients originally considered to be intermediate risk. For example, from Table 8, 58 patients who were considered to be intermediate risk as originally defined by Motzer et al13 were considered favorable risk by the new extended model and definitions. One-year and median survival for these patients was 81% and 24.0 months, respectively, which is similar to the figures of 82% and 28.6 months observed for the 57 patients classified as favorable by both definitions. Similarly, 50 patients initially classified as intermediate risk are now considered poor risk. One-year and median survival for these patients were 29% and 8.1 months, respectively, compared with 16% and 4.5 months, respectively, for the 35 patients considered poor risk by both models and 58% and 14.4 months, respectively, for the 107 patients considered intermediate risk by the two models. The similarities in prognosis based on the two models are shown graphically in Figure 2.
Histology and nuclear grade were not considered initially because of the large number of patients for whom this information was not available. However, applying the extended model to the patients for whom histology and nuclear grade were available suggests that histology, but not nuclear grade, may impact on survival. Adjusting for the impact of the prognostic factors included in the extended model, patients with clear-cell tumors had a significantly better prognosis than patients with other histologies (P = .003, Fig 3).
DISCUSSSION
Although our study validates Motzer et al’s model and their criteria for defining risk groups, a difficulty with the model is that the majority of patients are classified as intermediate risk, and relatively few patients are considered as favorable or poor risk. In the present study, 70% of patients were classified as intermediate risk, and only 19% were considered favorable risk. Still fewer patients (11%) were considered poor risk. These proportions are similar to the distribution of risk groups reported by Motzer et al (ie, 18%, 62%, and 20% for favorable, intermediate, and poor risk, respectively). The disproportionately large number of patients in the intermediate risk group suggests that it may be somewhat heterogeneous with respect to outcome.
In the present study, intermediate risk patients with a single poor prognostic factor (n = 99) did have a significantly better prognosis than intermediate risk patients (n = 116) with two poor prognostic factors present (P = .003). Extension of the model by incorporating prior radiotherapy and the number of metastatic sites into the definition of risk groups overcomes this difficulty because the expanded definitions essentially reclassify the better prognosis subset of intermediate risk patients as favorable and the poorer prognosis subset as poor risk. This is true even if one separates the intermediate risk group into two subgroups based on whether one or two poor prognostic factors are present. That is, of the 99 patients with one poor prognostic factor based on the original definition, 58 are classified as favorable risk based on the expanded criteria. Median survival time for these patients was 24.0 months, which is similar to the median survival time of 28.6 months observed in the 57 patients who were categorized as favorable risk by both models. Similarly, of the 116 patients with two poor prognostic factors present, 42 are considered poor risk by the expanded criteria. Median survival time for these patients was 8.4 months, which is similar to the median survival time of 4.5 months observed for the 35 patients considered poor risk by both models. One hundred seven patients considered intermediate risk by both models were similar regardless of the number of poor prognostic factors present based on Motzer et al criteria. Median survival time was 14.8 months for the 33 patients with one poor prognostic factor present and 14.4 months for the 74 patients with two poor prognostic factors present.
Another finding in the present study is that, although pathologic features of the primary tumor were often not available, correcting for the factors in the extended model, patients with clear-cell tumors seem to have a significantly better prognosis than patients with other histologies (P = .003). Although the proposed extension to the prognostic model proposed by Motzer et al13 seems to improve the model’s discriminatory power and patients with clear-cell tumors seem to have a better prognosis than patients with tumors of other histologies, these results are based on the retrospective analysis of highly selected patients, and confirmation of the results is needed.
In conclusion, six prognostic factors were identified for predicting survival in patients with RCC and most have been validated in different studies. We were able to validate Motzer’s data and find additional prognostic factors (prior radiation therapy and number of sites involved) to also be independent prognostic factors in the survival of patients with previously untreated RCC. This can be helpful in refining the definition of intermediate and high risk groups. The good discriminatory power that risk group status seems to have indicates that these are important factors that should be considered in the management of patients with advanced RCC and in the design and analysis of future clinical trials. An international consortium of investigators has been organized to further examine prognostic factors in patients with metastatic and untreated RCC and to develop a common approach.
Authors’ Disclosures of Potential Conflicts of Interest
NOTES
Presented in part at the 39th Annual Meeting of the American Society of Clinical Oncology, Chicago, IL, May 31-June 3, 2003.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Greenlee RT, Murray T, Bolden S, et al: Cancer statistics, 2000. CA Cancer J Clin 50:7-33, 2000
2. Chow WH, Devesa SS, Warren JL, et al: Rising incidence of renal cell cancer in the United States. JAMA 281:1628-1631, 1999
3. Motzer RJ, Bander NH, Nanus DM: Renal-cell carcinoma. N Engl J Med 335:865-875, 1996
4. Figlin RA: Renal cell carcinoma: Management of advanced disease. J Urol 161:381-386, 1999
5. Linehan M, Shipley W, Parkinson D: Cancer of the Kidney and Ureter. Philadelphia, PA, Lippincott-Raven Publishers, 1997
6. Lanigan D: Prognostic factors in renal cell carcinoma. Br J Urol 75:565-571, 1995
7. Elson PJ, Witte RS, Trump DL: Prognostic factors for survival in patients with recurrent or metastatic renal cell carcinoma. Cancer Res 48:7310-7313, 1988
8. de Forges A, Rey A, Klink M, et al: Prognostic factors of adult metastatic renal carcinoma: A multivariate analysis. Semin Surg Oncol 4:149-154, 1988
9. Palmer PA, Vinke J, Philip T, et al: Prognostic factors for survival in patients with advanced renal cell carcinoma treated with recombinant interleukin-2. Ann Oncol 3:475-480, 1992
10. Fossa SD, Kramar A, Droz JP: Prognostic factors and survival in patients with metastatic renal cell carcinoma treated with chemotherapy or interferon-alpha. Eur J Cancer 30A:1310-1314, 1994
11. Lopez Hanninen E, Kirchner H, Atzpodien J: Interleukin-2 based home therapy of metastatic renal cell carcinoma: Risks and benefits in 215 consecutive single institution patients. J Urol 155:19-25, 1996
12. Motzer RJ, Mazumdar M, Bacik J, et al: Survival and prognostic stratification of 670 patients with advanced renal cell carcinoma. J Clin Oncol 17:2530-2540, 1999
13. Motzer RJ, Bacik J, Murphy BA, et al: Interferon-alfa as a comparative treatment for clinical trials of new therapies against advanced renal cell carcinoma. J Clin Oncol 20:289-296, 2002
14. Negrier S, Escudier B, Gomez F, et al: Prognostic factors of survival and rapid progression in 782 patients with metastatic renal carcinomas treated by cytokines: A report from the Groupe Francais d’Immunotherapie. Ann Oncol 13:1460-1468, 2002
15. Atzpodien J, Royston P, Wandert T, et al: Metastatic renal carcinoma comprehensive prognostic system. Br J Cancer 88:348-353, 2003
16. Miller AB, Hoogstraten B, Staquet M, et al: Reporting results of cancer treatment. Cancer 47:207-214, 1981
17. Kaplan E, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457-481, 1958
18. Mantel N: Evaluation of survival data and two new rank order statistics arising in its consideration. Cancer Chemother Rep 50:163-170, 1966
19. Cox D, Oakes D: Analysis of Survival Data (ed 1). New York, NY, Chapman and Hall, 1990
20. Figlin RA, Thompson JA, Bukowski RM, et al: Multicenter, randomized, phase III trial of CD8(+) tumor-infiltrating lymphocytes in combination with recombinant interleukin-2 in metastatic renal cell carcinoma. J Clin Oncol 17:2521-2529, 1999
21. Bukowski RM, Olencki T, Wang Q, et al: Phase II trial of interleukin-2 and interferon-alpha in patients with renal cell carcinoma: Clinical results and immunologic correlates of response. J Immunother 20:301-311, 1997
22. Agrawal N, Olencki T, Mekhail T, Peereboom D, et al: A phase I/II study of GM-CSF, interleukin-2 (IL-2) and interferon-alpha (INF-alpha) in metastatic renal cell cancer. Proc Am Soc Clin Oncol 21:5b, 2002 (abstr 1829)
23. Hutson T, Mekhail T, Messerli E, et al: Phase I trial of PEG-Intron and IL-2 in patients with metastatic renal cell carcinoma. Proc Am Soc Clin Oncol 21:24a, 2002 (abstr 2406)
24. Olencki T, Malhi S, Mekhail T, et al: Phase I trial of thalidomide and interleukin-2 (IL-2) in patients with metastatic renal cell carcinoma. Proc Am Soc Clin Oncol 22:387, 2003 (abstr 1554)
25. Motzer RJ, Rakhit A, Thompson JA, et al: Randomized multicenter phase II trial of subcutaneous recombinant human interleukin-12 versus interferon-alpha 2a for patients with advanced renal cell carcinoma. J Interferon Cytokine Res 21:257-263, 2001
26. Hutson TE, Molto L, Mekhail T, et al: Phase I trial of consensus interferon in patients with metastatic renal cell carcinoma: Toxicity and immunological effects. Clin Cancer Res 9:1354-1360, 2003
27. Bukowski RM, Sergi JS, Sharfman WJ, et al: Phase I trial of natural human interferon beta in metastatic malignancy. Cancer Res 51:836-840, 1991
28. Plautz GE, Bukowski RM, Novick AC, et al: T-cell adoptive immunotherapy of metastatic renal cell carcinoma. Urology 54:617-623, 1999
29. Motzer RJ, Rakhit A, Schwartz LH, et al: Phase I trial of subcutaneous recombinant human interleukin-12 in patients with advanced renal cell carcinoma. Clin Cancer Res 4:1183-1191, 1998
30. Tate J, Olencki T, Finke J, et al: Phase I trial of simultaneously administered GM-CSF and IL-6 in patients with renal-cell carcinoma: Clinical and laboratory effects. Ann Oncol 12:655-659, 2001
31. Wos E, Olencki T, Tuason L, et al: Phase II trial of subcutaneously administered granulocyte-macrophage colony-stimulating factor in patients with metastatic renal cell carcinoma. Cancer 77:1149-1153, 1996
32. Thompson JA FR, Galanis E, Bukowski R, et al: Phase II trial of plasmid DNA/lipid (leuvectin) immunotherapy in patients with metastatic renal cell cancer. Proc Am Soc Clin Oncol 19:462a, 2000 (abstr 1814)
33. Olencki T, Peereboom D, Wood L, et al: Phase I and II trials of subcutaneously administered rIL-2, interferon alfa-2a, and fluorouracil in patients with metastatic renal carcinoma. J Cancer Res Clin Oncol 127:319-324, 2001
34. Chang DZ, Olencki T, Budd GT, et al: Phase I trial of capecitabine in combination with interferon alpha in patients with metastatic renal cancer: Toxicity and pharmacokinetics. Cancer Chemother Pharmacol 48:493-498, 2001
35. Perez-Zincer F, Olencki T, Budd GT, et al: A phase I trial of weekly gemcitabine and subcutaneous interferon alpha in patients with refractory renal cell carcinoma. Invest New Drugs 20:305-310, 2002
36. Schacter LP, Bukowski R, Carducci MA, et al: Safety profile of DHA-paclitaxel (TXP) in 8 phase II trials. Proc Am Soc Clin Oncol 22:214, 2003 (abstr 856)
37. Orrell D: Albumins as an aid to the interpretation of serum calcium. Clin Chim Acta 35:483-489, 1971(Tarek M. Mekhail, Rony M.)