BioMed Critical Commentary™

_______________The Other 90% of Science_______________

View Linked Comments
PMID: 19509160

Rajeshkumar NV et al
Clin Cancer Res 2009 Jun 15;15(12):4138-46
Antitumor effects and biomarkers of activity of AZD0530, a Src inhibitor, in pancreatic cancer.

5 comments exist on this article.

No useful specificity shown

Rank: 1

BioMed Crit Comm 2010; 3:2271
Posted: Dec 30, 2010 CCID: 2271
Commentator: PhilW	Ave. score of this commentator: 2.7 for 64 scored comments.

The authors use the toxic compound AZD0530 in mice containing xenografted tumors, finding that the treated tumors grew slower. They conclude that there will be "pancreatic tumor sensitivity to AZD0530 in the clinic". Specific and useful effect on tumor cells, however, were not shown. 1) Figure 1 shows that in all treated mice, the tumors continued growing. No tumor regression was observed. To observe a slowing of tumor growth is a promiscuous effect of toxins when administered to tumor-bearing mice. It is not a sign of specificity, for no other growing cell population in the mouse was compared to the tumor cells. 2) Studies using cell culture were not presented. Such studies can be used to show that a compound has specificity for tumor cells as compared to normal cell types. 3) No comparison was performed of tumors having an abnormality of Src with those not having an abnormality. Thus, specificity for Src was neither sought nor observed for this "Src inhibitor".

No real invasion was studied

Rank: 2

BioMed Crit Comm 2010; 3:2273
Posted: Dec 30, 2010 CCID: 2273
Commentator: PhilW	Ave. score of this commentator: 2.7 for 64 scored comments.

Subcutaneous xenografts typically become invested with mouse vessels and stroma, supporting the tumor graft as it expands in diameter. Neighboring mouse tissues, such as muscle and skin, are pushed against but are not invaded. It can be misleading to refer to invasion in such a model, such as where the authors state, "Immunohistochemistry results clearly show a treatment effect for AZD0530 on the Src substrates key to the invasive phenotype." There was no observed invasive phenotype in the current studies, in treated or untreated tumors. Nor were the treated tumors shown to be less invasive in their growth patterns on a microscopic level within the tumor masses (that is, focusing on the invested murine tissue within the tumor and ignoring the neighboring tissues since they were not invaded).

Invalid: Data mining without followup independent validation set

Rank: 3

BioMed Crit Comm 2011; 4:2398
Posted: May 9, 2011 CCID: 2398
Commentator: Dave	Ave. score of this commentator: 2.6 for 24 scored comments.

The authors used data mining of a large database to identify a gene pair that was predictive in their initial data set. This is likely to be due to overfitting rather than due to a real relationship, for false positive results are to be expected in data-mining exercises. It is essential to validate the data-mining result in a second, independent sample set. This validation was not performed.

Minimal study: No dose-response relationship presented

Rank: 4

BioMed Crit Comm 2010; 3:2272
Posted: Dec 30, 2010 CCID: 2272
Commentator: PhilW	Ave. score of this commentator: 2.7 for 64 scored comments.

Only a single dose level was explored in the treatment study of xenografts. Conventionally, one prefers to explore whether there exists a dose-response relationship. Otherwise, it remains unclear whether any tumors were really more sensitive than others in a systematic manner, or whether instead there merely existed an idiosyncratic phenomenon at a particular dose level for reasons largely irrelevant to a preclinical investigation.

Invalid statistical procedures used

Rank: 5

BioMed Crit Comm 2011; 4:2397
Posted: May 9, 2011 CCID: 2397
Commentator: Dave	Ave. score of this commentator: 2.6 for 24 scored comments.

The authors used "gene set enrichment analysis" as a statistical tool. This was an invalid choice. 1) The underlying statistical foundations of GSEA are thought to be invalid and the GSEA can produce bizarre results (PMID: 16199517 CC). 2) There were only five sensitive cell lines. This number is too few to permit use of GSEA-like clustering techniques.

Key authors in this publication:

De Oliveira E, Hidalgo M, Jimeno A, Messersmith WA, Rajeshkumar NV, Tan AC.

View Linked Comments	PMID: 19509160
Rajeshkumar NV et al Clin Cancer Res 2009 Jun 15;15(12):4138-46 Antitumor effects and biomarkers of activity of AZD0530, a Src inhibitor, in pancreatic cancer.