| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Memorial Sloan-Kettering Cancer Center New York, NY 10021
Weill Medical College of Cornell University New York, NY
10021
WEB—This is a Web exclusive article.
In the Materials and Methods section, the authors state that 30 consecutive patients with metastatic colon cancer were recruited for this study from a larger cohort enrolled in a multi-institutional trial; only patients with a minimum of five well-defined, nonconfluent, and nonnecrotic liver metastases were eligible for inclusion in the study. The resulting study population was thus highly selective, and the lesion assessment reflective neither of the goal of the RECIST criteria nor of the vast majority of therapeutic trials-which, namely, is to assess the response to therapy in an entire patient, not just in one organ. In colon cancer, metastases often occur in various nonhepatic sites, such as lung and lymph nodes; disease in all sites must be measured to gauge the response to therapy. Thus, assessing therapeutic response only on the basis of hepatic metastases would not be valid in many patients with metastatic colon cancer.
Also, even if one elected to assess the minimum number of lesions to be measured only in the liver, the lesion selection criteria chosen by the authors are quite biased because lesions were selected only if well-defined, nonconfluent, and nonnecrotic. In colon cancer (as well as many other primary tumors), liver metastases frequently are ill-defined, confluent, and necrotic. The selection of only homogeneous metastases thus favored "easier" measurements, and it is not at all surprising that the number of such lesions that needed to be measured was low. The authors' conclusion that only one liver metastasis must be measured (and in only one dimension) cannot be generalized to the common, real-world situation in which liver metastases from colon cancer are typically ill-defined, confluent, or necrotic.
The authors understate the conclusions of our research on this topic [2], in which we evaluated the critical issue of variance, using both unidimensional (RECIST) and bidimensional (World Health Organization [WHO]) criteria in response assessment. Our conclusions about the number of tumor deposits (selected from all sites of metastases) that needed to be measured were applicable to either RECIST or WHO criteria, and were based upon the particular level of variance that would be acceptable for the purposes of a specific study or clinical trial. Moreover, the authors seem to have overlooked our theoretical derivation [3], in which we provided a formula for calculating the incremental increase in variability that results when a subset of tumors, rather than all available tumors, is selected for measurement. The number of lesions selected to be measured depends in large part upon the percentage of variability in response assessment that is acceptable (or unacceptable) for a given purpose: The less variability that is acceptable, the larger the number of lesions that will have to be measured.
Because the authors reached their conclusions on the basis of an artificial, highly selective subset of patients with nonrepresentative hepatic metastases, and because measurement of hepatic lesions alone does not constitute overall response assessment for a patient, those conclusions cannot be generalized to all patients with metastatic colon cancer. Thus, the authors' final statement in the Discussion section that "This fact has significant clinical implications and needs to be applied in routine oncologic practice and pharmaceutical trials" far oversteps its actual scientific underpinnings, and we caution anyone from implementing this recommendation.
 Top References |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
