SAN FRANCISCO – The International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) model is a valid tool for predicting survival in patients starting second-line targeted therapy for this disease, new data show.
In fact, it outperforms the leading model derived a decade ago, researchers reported at the 2014 Genitourinary Cancers Symposium sponsored by the American Society of Clinical Oncology.
The new model uses six clinical and laboratory factors to stratify patients into prognostic risk groups. It builds on the widely used three-factor Memorial Sloan-Kettering Cancer Center (MSKCC) second-line prognostic model (J. Clin. Oncol. 2004;22:454-63).
Results of the retrospective cohort study showed that the IMDC model performed well for risk stratification in this setting, with patients in the intermediate- and high-risk groups having roughly two and six times the risk of death, respectively, compared with peers in the favorable risk group, reported Dr. Jenny J. Ko of the British Columbia Cancer Agency, Vancouver. Also, it was superior to the MSKCC model for predicting survival as assessed by three statistical tests.
"The six-factor IMDC prognostic model has been validated in and can be applied to patients previously treated with targeted therapy, in addition to previously validated populations in the first-line targeted therapy and non–clear cell setting," she commented.
"The model fits better into the context of the contemporary treatment era, where targeted therapies are used in a sequential fashion," Dr. Ko added.
The study population was heterogeneous, she acknowledged. But "this can also mean that the cohort was generalizable, as our cohort includes trial and off-trial patients, and it also includes community and academic centers."
Invited discussant Dr. Daniel Canter of the Fox Chase Cancer Center, Philadelphia, noted that the durations of second-line therapy and survival after starting this therapy were fairly short, highlighting the need for better therapies.
The choice of first-line therapy for metastatic renal cell carcinoma (mRCC) may shift in the future away from sunitinib (Sutent) and sorafenib (Nexavar), used in the vast majority of the patients studied, he noted.
"My observation is that the medical oncologists that I work with seem to be using pazopanib (Votrient) more now as a first-line therapy because of the improved toxicity," Dr. Canter noted, and doing so may ultimately affect survival. "Obviously, a future question is the role of pazopanib. And ... an algorithm to define second-line choices is clearly something where there is an opportunity for work to be done."
Dr. Ko and her team studied 1,021 consecutive patients with mRCC from 19 centers who were starting second-line targeted therapy. (About a fifth had received immunotherapy before their first-line targeted therapy and were therefore on third-line therapy of any type.)
Eighty-five percent of patients had undergone nephrectomy. Their first-line targeted therapy had most often been sunitinib (67%) or sorafenib (20%).
The second-line therapies given included sorafenib (29%), sunitinib (22%), everolimus (Afinitor) (22%), and temsirolimus (Torisel) (12%), among others.
In a multivariate analysis, five of the six IMDC factors (Karnofsky performance status less than 80%; interval between diagnosis and treatment of less than 1 year; anemia; thrombocytosis; and neutrophilia) individually predicted an elevated risk of death after starting second-line therapy, reported Dr. Ko, who disclosed no conflicts of interest related to the research. Hypercalcemia did not, but few patients had this risk factor.
The median time on second-line targeted therapy was 3.9 months, and the median overall survival after starting this therapy was 12.5 months.
Compared with patients in the favorable IMDC risk group (having none of the risk factors), patients in the intermediate-risk group (having one or two) and patients in the high-risk group (having three or more) had significantly elevated risks of death (hazard ratios, 1.89 and 5.73).
In a likelihood ratio test, comparing the two models, the chi-square value was highly significant, suggesting that inclusion of the three additional factors in the IMDC model enhanced predictive performance over that with the MSKCC model.
Finally, in a reclassification calibration analysis, the chi-square value was much lower for the IMDC model than for the MSKCC model (6.2 vs. 54.2), indicating that the observed vs. predicted number of deaths at 1 year was better for the former.