The emperor’s new speedos

There’s no doubt that the emergence of genomics was a huge scientific breakthrough. It supercharged research in any number of disciplines, and essentially created new ones such as microbiomics.

But it hasn’t done the one thing it was supposed to do: transform the diagnosis and treatment of disease. The first genome was sequenced in 2001. Many thousands more genomes have followed, and an entire discipline of Big Data tools and techniques have been developed to curate and analyze this tidal wave of data. No one should have expected this technology to be translated into clinical results in just a couple of years. But it has been fifteen years now, long enough for the technology to mature and begin yielding a few benefits in the clinic. We should be seeing some results by now.

Here are deaths from the #1 killer in the US, cardiovascular disease

From Heart Disease and Stroke Statistics-2016 Update

Do you see any acceleration in the decline of the death rate? Me neither. In fact, it appears that death rates are plateauing. Progress is slowing down, not speeding up.

OK, how about the #2 killer, cancer?

From Cancer Facts & Figures 2016

Rates began dropping in 1990, mostly because of the success of efforts to control smoking. They have continued to drop, but the rate is pretty steady. There is no sign of a faster drop in the genomic era.

That’s the 30,000 foot view and it may be unfair. After all, we know that heart disease and cancer are both significantly influenced by environment and behavior. Genomics might have an impact on these – we can imagine using genome results to counsel people to avoid certain foods or behaviors. But we already know that eating a balanced diet, avoiding smoking and processed foods, and getting some exercise is good. The only surprise would be if genetic testing identified those of us who could smoke, eat steaks and Cheetos every day while laying around on the couch and still live to be 90. That’s a genome test I would pay for.

But when we look for new therapies that have been enabled by genomics, the page is pretty much blank. A good place to find new breakthroughs would be the Genome Advance of the Month site, run by the National Human Genome Research Institute. They should know of all the latest advances, and be eager to publicize them. The archives go back to 2011, and contain lots of good stories about excellent and interesting science being done by genomics researchers. But new therapies? There is a story here about a few individuals who received genome-based diagnoses that changed their therapy for the better. When I say “a few”, that’s what I mean: less than ten. And that’s it.

All the other stories are about new discoveries that could, at some future date, lead to new therapies. In other words, they are the same stories about genomics that have been getting written for a couple of decades now.

But what about personalized medicine? Even if genomics has not resulted in the discovery of new medicines, isn’t it making the use of existing medicines more effective? Well, yes, it is. But just not as much as you might be led to believe. There are currently two limitations on the impact of personalized medicine: (1) there aren’t that many conditions that can be personalized; and (2) the personalized therapies are not that much better than generic therapies.

HER2+ breast cancer is a good example. Human Epidermal Growth Factor Receptor 2 is a molecule displayed on some breast cancer cells that stimulates growth of the cancer cells, making these tumors aggressive and difficult to treat. Antibodies that bind to HER2 block this stimulation, and when combined with cell-killing chemotherapeutics, improve patient outcomes. Herceptin (the antibody which blocks the receptor) therapy has long been considered one of the biggest advances in cancer treatment and the prime example of the potential of personalized medicine.

But there aren’t that many HER2-positive patients – they account for only about 15% of all breast cancers. Even if Herceptin was a cure, it would not bend the cancer death rate curve down by all that much. Unfortunately, Herceptin is far from being a cure. Median disease-free survival increases from 4.6 months to 7.6 months when Herceptin therapy is added to standard treatments. Three added months of relative good health is not nothing, and a few percent of patients even experience cures. Herceptin is a valuable advance, but no breakthrough.

That’s the best case story for personalized medicine right now: a few percent of patients see a modest benefit. If we can keep adding more treatments like Herceptin to the list of treatments then of course we will see a greater cumulative impact, and perhaps we are seeing it. After all, the death rate from cancer is dropping steadily. There’s just little evidence that genomics has anything to do with this drop.

I think it is fair to say that genomics has had next to no impact on public health so far (even Herceptin was developed well before the first genome was sequenced). And I think that this lack of impact will continue well into the future, because many of the premises of genomic therapy and diagnosis are fundamentally flawed. I’ll explain why in a future post.