A few years ago, Luke Timmerman of Xconomy wrote about the 21 Red Flags in biotech investing and Sally Church of the Pharmastrategy Blog wrote a related post on Red Flags in clinical trials and cancer research. Here are a few additions of my own that I’d like to add to those lists.
Context is Everything
Target identification itself is something of a black art, because in many cases you have a really incomplete picture of the target space for a particular disease. The literature may only tell half the story. Most of the papers I’ve been reading lately identify one or two novel genes or pathways and associate that gene or pathway with pancreatic cancer. When I read the Methods section, the authors have a tendency to natter on about standard protocols they followed, and neglect to tell us anything about the patients profiled in the paper. This leaves me with some nagging questions:
- What stage was patient in when the sample was taken? What I really want to know is can a particular result (a mutation or a differentially expressed gene) be tied to a specific process known to be in play at a given time in the progression of the disease.
- Was the patient from a family that had a history of one of the 5-6 known syndromes for pancreatic cancer? Do the results reflect one of those syndromes? [PMID: 23187834]
- What parts of the tumor were sampled? Tumors are rarely if ever homogeneous in nature. Colonies of cells continue to accumulate mutations as they attempt to gain the functions they need to invade local tissue and to metastasize. This means that if you take multiple samples from a tumor you are likely to get different answers, and those variations will give you a clearer picture of the tumor biology and progression. The more complete the picture, the greater your chances of success in clinical trials.
- What was the state of the primary tumor area? Do we see signs of local invasion? If so, in what tissues, and how does that correlate to the genotyping or gene expression data we’re seeing? Does a piece of tumor located near the liver show signs that the tumor is preparing to invade the liver? Does the primary tumor show signs of invading the celiac nerve plexus? Do we see a corresponding signature for perineural invasion?
- Do we have distal metastases? If so, where? Were samples taken from the metastases and compared with primary samples? [PMID: 20981101]
- Was the patient a smoker? If so, how much? Do the results show any connection to known smoking-related mutational signatures?[PMID: 19351817]
- Were the samples taken prior to treatment or after treatment, or both? If both, what was the effect of the treatment both at a phenotypic level, and at a genetic level?
- Were any circulating tumor cells also sampled? How does this compare with the primary tumor? How do those samples compare with one another? How do they compare with mets? Do certain CTCs genetically resemble site-specific mets? Is this CTC destined for a brain metastasis or a met in the omentum?
In general, I’d like to see more papers attempt to connect the dots not only between genes and pathways, but to specific processes in cancer. That connection to a disease process discloses the real value behind the research. Without it, it’s like reading a murder mystery and realizing that someone ripped out the last chapter. You feel cheated.
Lack of Reproducibility
Over the past few years pharmas have begun to depend on a greater extent on the research of external collaborators in academia and startup companies. In the case of the former, it’s been a real challenge to reproduce some of the results that pharmas are hoping to place a bet on. In academia, the emphasis is on publishing — and successful results get published, experimental failures don’t. This means that you often see results where the experiment failed 5 times, but succeeded on the 6th; and it was the 6th that was included in the paper. In industry, an experiment has to work 10 times out of 10 in order for someone to bet the “pharm” on it.
Getting The Right Advice
When selecting a target (and an indication for that target) it’s important that you get good advice from people who work in that discipline. Stephen Covey, the late author of the book, “The Seven Habits of Highly Effective People”, used to say, “start with the end in mind”. And in this case, that means talking with clinicians, and letting the disease guide the discovery.
From the clinicians you want to know what constitutes an underserved area. For example, your initial inclination might be to attempt to tackle metastatic disease in pancreatic cancer. However, a recent study showed that it was actually the degree of local invasion that was a greater predictor of mortality in patients. And this makes sense if you think about it. If the tumor is invading parts of the digestive tract, and into the sympathetic nervous system, the quality of life of the patient is rapidly diminished and the amount of pain they experience is increased. Since perineural invasion is a common event in pancreatic cancer patients, and most patients spend the last few weeks on large amounts of pain killers, it might be a better short-term strategy to interfere with perineural invasion, and have a longer-term research program focused on a cytotoxic therapy.
Ultimately, you want to end up with something that clinicians want to prescribe to their patients. And the more you know about the problems they’re dealing with, the more likely you are to have a winning product.