The People

In 2001 I was working at a little biotech company in East Los Angeles. The founders were two professors at the USC Med School. The company had a pipeline of 4 antibodies against "denatured" collagen. The idea behind them was that they would prevent blood vessels from forming around tumors and thus kill them. The company was funded by an angel investor. The goal was for us to obtain some data that could be used to sell the company off and make our investor some money.

The experiment consisted of three groups of bald mice. Each group had 10 mice. There was a positive control group, treated with Rituxan. Rituxan is a monoclonal antibody that binds to VEGF. There was a negative control group, treated with PBS. There was an unknown group, treated with one of our antibodies. Day one we injected cancer cells subcutaneously on the backs of the mice. Each day the mice were injected with a large dose of the drug or PBS shot into their belly. In a couple days tumors began to grow, which we measured with a caliper.

It was extremely inhumane because no one had a painless way of doing this. The mice were horrified when the lights came on. I tagged along to write down the readings. I watched as each mouse was held upside down and a large needle was injected into a non-distinct region of their underbelly.

The experiments were done over and over. That is because we had to get the set of data that our superiors wanted. The investor had decided that the antibodies were a great idea and he sank millions into making a profit from them. The lab people needed to get the results. The problem was that we our antibodies didn't work. The Rituxan worked great. No tumors. The other two groups were statistically the same. But this would not do. We were sent back to have a graduate student from one of the founders lab show us how to "properly" use the caliper to make the measurements. She spoke as she measured, instructing us. As she instructed she was clearly favoring the hoped for measurements. Using a caliper is not a skill. When she had finished getting the data to look "better" and she ended her lesson on measuring tumors.

I had just one question. I handed her a mouse that she had already measured. She asked, "what is this?".

Me: It's a mouse with a tumor. Can you measure this one?

Her: Didn't we measure it?

Me: Yes, if you are measuring accurately you should get the same tumor size.

She measured again and was off by 30%. This number of course was not written down. It was for my own assessment. That was my N-Rays moment. I knew in my heart that they had been playing these games. The new measurement didn't count. I knew I was in a Cargo Cult.

The graduate student and the boss needed to show a reduction in tumor size. Somehow by measuring the mice again, they did not feel that they were simply fabricating data. After all, they had performed the measurements. To me, this was fraud. To the boss and the grad student, this was survival. They needed the data as much as anyone else.

My entire biotech career is littered with this ilk. They are good people until the data doesn't go their way. They convince themselves that these things are simply little white lies. They don't matter. In the big picture they know their drugs work and they'll be vindicated later. The little experiments where data is cherry picked is no big deal. It's one or two slides in a presentation that tells a much more interesting story. And the story is what sells. Investors open their pocket books and bigger companies start listening.