Surrogate Endpoint For Survival Research Articles

Do We Have Good Surrogate Endpoints for Survival in Breast Cancer Studies?

Do We Have Good Surrogate Endpoints for Survival in Breast Cancer Studies?

OUTCOME ASSESSMENT IN CANCER MANAGEMENT

The goals of cancer treatment may be prevention, cure, or palliation. In each case, the length and quality of survival are the ultimate outcomes to assess when evaluating the success of treatment. Some surrogate or intermediate endpoints, however, may be considered, such as toxicity or the tumor response. The first section of this article considers traditional outcomes and endpoints and intermediate or surrogate endpoints for survival and quality of life. The second section specifically addresses the issue of the quality-of-life outcome, which has become increasingly important over the last 2 decades. The last section suggests that combining survival and quality of life may not be easy. Costs are increasingly important in management of diseases, and the outcomes of cancer treatment frequently are measured against them. Although the definitions of cost-effectiveness and cost-utility are mentioned at the end of this article, the article does not deal with the problems of decision making for health resource allocation.

Investigation of disease-free survival as a surrogate endpoint for survival in cancer clinical trials

A stochastic model wiuh exponential components is used to describe our data collected from a phase III cancer clinical trial. Criteria which guarantee that disease-free survival (DFS) can be used as a surrogate for overall survival are explored under this model. We examine several colorectal adjuvant clinical trials and find that these conditions are not satisfied. The relationship between the hazard ratio of DFS for an active treatment versus a control treatment and the cumulative hazard ratio of survival for the same two treatments is then explored. An almost linear relationship is found such that a hazard ratio for DFS of less than a threshold R corresponds to a non-null treatment effect on survival The threshold value R is determined for our colorectal adjuvant trial data. Based on this relationship, a one-sided test of equal hazard rate of survival is equivalent to a test of hazard ratio of DFS small than R This approach assumes that recurrence information is unbiasedly and accurately assessed; an assumpion which is sometimes difficult to ensure for multicenter clinical trials, particularly for interim analyses.

Relation between tumor size, quality of life, and survival in patients with colorectal liver metastases.

This study assessed the relationship between survival, tumor size, and quality of life (QoL; Rotterdam Symptom Checklist [RSC], Hospital Anxiety and Depression Scale [HAD], and Sickness Impact Profile [SIP]) in 50 patients with colorectal liver metastases (CLM). Physical symptom score (RSC) was a stronger survival predictor than tumor size measured on computed tomographic (CT) scan. The best model for predicting survival included QoL questions about diarrhea, eating, restlessness, and ability to work and sleep. The only clinically measured variable included in this best survival prediction model was serum alkaline phosphatase level. This is the first study to show that QoL indices predict survival in CLM. The findings suggest that differences in tumor products and not just in tumor size could influence patient fitness and survival in CLM. Such differences are more accurately estimated by QoL assessment than measurement of tumor size. QoL provides a better survival estimate than measurement of tumor size and could be used as a surrogate end point for survival in treatment trials.

What can we learn from a meta-analysis of trials testing the modulation of 5-FU by leucovorin?

We previously reported a meta-analysis of all randomized clinical trials comparing 5-FU to 5-FU plus intravenous d.l. leucovorin (LV) in patients with advanced colorectal cancer. The meta-analysis confirmed the advantage of 5-FU/LV over 5-FU alone in terms of response rate: overall, the response rate was 11% with 5-FU alone vs 23% with 5-FU/LV (p < 10(-7)). At the same time, it was showed that tumor regression can not be considered a surrogate end point for survival in patients with advanced colorectal cancer: no significant survival advantage was observed for patients allocated to 5-FU/LV. The present paper focuses on the interest of meta-analysis to study the role of the biomodulation of 5-FU by LV. This approach is compared with the analysis of individual clinical trials in terms of power, bias and credibility. It is argued that for the meta-analysis to be reliable, individual patient data from all available studies must be used, rather than summary data extracted from published papers. It is concluded that meta-analysis is very useful as a tool to explore and summarize available data on a given clinical problem, though its purpose is not to recommend any treatment modality. This meta-analysis has confirmed both the benefit of biomodulation of 5-FU by LV, and the limitations of the current modalities to impact significantly on overall patient survival. Further laboratory and clinical research is warranted on the biomodulation of 5-FU by LV.