MD Anderson Study: scientists unable to verify cancer research studies by others

I  agree that our precious clinical trial dollars should only be used to test drugs that have really solid and encouraging pre-clinical data to recommend them. If the pre-clinical data (i.e. studies done in test tubes, cell culture and animal models) is wrong, then the whole idea goes out the window.

However, as a professional scientist, I have personally experienced and witnessed how difficult it can sometimes be to reproduce experients– sometimes for the oddest reasons. One time I spent months of time and thousands of dollars of grant money failing to reproduce some important published data. It eventually turned out that the problem was with the BRAND of salt I was using; when I used the brand of salt that the original investigator used, the experiment worked the first time.

A similar thing happened to a friend of mine. He was studying cockroaches in Germany and got some really interesting results. He came to the US to continue the study and suddenly nothing worked. Same man. Same cockroaches. Nothing worked. In that case (again after thousands and thounds of wasted dollars and hours) the difference was in the brand of paper towels he used to line the insect cages; pine trees in Germany produce a different chemical than do American pine trees. When his friends in Germany sent him German paper towels, his experiments worked. 

My point is that while there may be an ocassional case where an investigator publishes a finding that they know to be unreliable or even incorrect, that is very rare. The vast majority of the time published results can be replicated. Most of the rest of the time the results can be replicated but only after a long, difficult and expensive period of picking apart every single detail of every single item that goes into an experiment. Requiring pharmaceutical companies to do that for all the pre-clinical data they use when selecting drugs for clinical trials would not be a good use of precious clinical trial dollars, either.  

I  agree that our precious clinical trial dollars should only be used to test drugs that have really solid and encouraging pre-clinical data to recommend them. If the pre-clinical data (i.e. studies done in test tubes, cell culture and animal models) is wrong, then the whole idea goes out the window.

However, as a professional scientist, I have personally experienced and witnessed how difficult it can sometimes be to reproduce experients– sometimes for the oddest reasons. One time I spent months of time and thousands of dollars of grant money failing to reproduce some important published data. It eventually turned out that the problem was with the BRAND of salt I was using; when I used the brand of salt that the original investigator used, the experiment worked the first time.

A similar thing happened to a friend of mine. He was studying cockroaches in Germany and got some really interesting results. He came to the US to continue the study and suddenly nothing worked. Same man. Same cockroaches. Nothing worked. In that case (again after thousands and thounds of wasted dollars and hours) the difference was in the brand of paper towels he used to line the insect cages; pine trees in Germany produce a different chemical than do American pine trees. When his friends in Germany sent him German paper towels, his experiments worked. 

My point is that while there may be an ocassional case where an investigator publishes a finding that they know to be unreliable or even incorrect, that is very rare. The vast majority of the time published results can be replicated. Most of the rest of the time the results can be replicated but only after a long, difficult and expensive period of picking apart every single detail of every single item that goes into an experiment. Requiring pharmaceutical companies to do that for all the pre-clinical data they use when selecting drugs for clinical trials would not be a good use of precious clinical trial dollars, either.  

I  agree that our precious clinical trial dollars should only be used to test drugs that have really solid and encouraging pre-clinical data to recommend them. If the pre-clinical data (i.e. studies done in test tubes, cell culture and animal models) is wrong, then the whole idea goes out the window.

However, as a professional scientist, I have personally experienced and witnessed how difficult it can sometimes be to reproduce experients– sometimes for the oddest reasons. One time I spent months of time and thousands of dollars of grant money failing to reproduce some important published data. It eventually turned out that the problem was with the BRAND of salt I was using; when I used the brand of salt that the original investigator used, the experiment worked the first time.

A similar thing happened to a friend of mine. He was studying cockroaches in Germany and got some really interesting results. He came to the US to continue the study and suddenly nothing worked. Same man. Same cockroaches. Nothing worked. In that case (again after thousands and thounds of wasted dollars and hours) the difference was in the brand of paper towels he used to line the insect cages; pine trees in Germany produce a different chemical than do American pine trees. When his friends in Germany sent him German paper towels, his experiments worked. 

My point is that while there may be an ocassional case where an investigator publishes a finding that they know to be unreliable or even incorrect, that is very rare. The vast majority of the time published results can be replicated. Most of the rest of the time the results can be replicated but only after a long, difficult and expensive period of picking apart every single detail of every single item that goes into an experiment. Requiring pharmaceutical companies to do that for all the pre-clinical data they use when selecting drugs for clinical trials would not be a good use of precious clinical trial dollars, either.