USPSTF on Prostate Screening: "Meh."

The US Preventative Services Task Force (USPSTF) has recommended that men not get the PSA test for prostate cancer. The recommendation applies to men who have not already been diagnosed with prostate cancer, nor have "symptoms that are highly suspicious for prostate cancer." So, USPSTF recommends that the PSA test not be used at all for screening. USPSTF recommends that the test can be used for men who pretty obviously already have prostate cancer, but the test isn't all that important for such men anyway.

Here's what's going to happen: USPSTF is going to start accepting comments on October 11. There will be a firestorm of criticism from patient advocacy groups, individual prostate cancer survivors, and physicians and companies that make a good living treating prostate cancer. Then the USPSTF will stick to its guns and issue a recommendation statement very similar to the draft. Treatment guidelines will change, and most men will never get a PSA test.

In 2009, USPSTF issued a similarly controversial recommendation on mammogram screening for breast cancer, recommending that most women between 40 and 50 no longer receive mammogram screening. At the time, many people had trouble understanding how more diagnosis can be a bad thing. The short answer is, a diagnostic test can lead to treating disease that otherwise would not have become a health problem, and the treatment itself carries significant risks.

Now USPSTF has dropped the other shoe, and made a similar recommendation for PSA, for the same reasons. USPSTF went further with the PSA test, essentially recommending that men never be screened with the PSA test for prostate cancer.

Here's the problem: if you have a prostate, and you live long enough, you'll get prostate cancer. Cancer is almost a normal part of aging for this organ. Few of these cancers will ever become aggressive enough to be life-threatening, and the majority of deaths from prostate cancer are in men over the age of 75, who were not likely to live many decades longer if cured anyway.

So, if you screen men, you can find a lot of prostate cancer. If you treat these men, almost all of them will survive the treatment, because their prostate cancer wasn't going to kill them anyway. A few men with aggressive cancer will be detected earlier, but it doesn't appear that detecting aggressive prostate cancer earlier has much effect on survival rates.

And prostate cancer treatment is very invasive. A positive PSA test leads to a biopsy. If the biopsy confirms cancer, then surgery typically follows. Impotence and incontinence are frequent side-effects of surgery. Some men are treated with hormone therapies that cause erectile dysfunction, hot flashes, and the development of breasts. And, some men die from their treatment.

But that could still be worth it, right? All of that suffering and expense could still save lives, right? Well, it could, but it happens that more men die younger if they are screened than if they are not. And furthermore, the men who are not screened are happier: they have fewer cancer worries, and they do not suffer the serious side effects that result from treatment.

The biggest problem with the PSA test is, it has a high false-positive rate. Make that a very high false positive rate. For every 10 men with a positive PSA test, 2 actually have prostate cancer. But 8 think they have prostate cancer and will get biopsied. Some fraction of the men who are biopsied will have a false-positive biopsy, and continue on down the treatment path for no good reason. (These numbers apply to the "cut point" that is normally used for the PSA test.)

Diagnostics are tough. It is often the case that there is a rock-solid relationship between a testable biomarker and a medical condition, but a useful diagnostic test only comes years or decades later. The science of PSA is sound. The mistake was using PSA to screen for disease. The USPSTF is in the process of correcting that mistake.

Elizabeth Dragon

Elizabeth Dragon died of "natural causes" in February, shortly before she was due to be sentenced for securities fraud. Her death brings to mind the importance of scientific ethics in one's life and the lives of others.

Dragon had an illustrious career as a senior scientist at Roche before leading R&D at Sequenom. Eventually, certain scientific results became part of an internal investigation, leading to an SEC investigation. Dragon pled guilty to securities fraud and had her sentencing delayed, as she agreed to cooperate with the ongoing investigation.

What did she do to destroy her career, nearly wreck a prominent diagnostics company, change the course of her life utterly, and even possibly shorten her life? In describing results of Sequenom's trisomy test, she describe results as blinded, when in fact researchers had been unblinded. That simple yet critical element of scientific dishonesty had dramatic consequences for her and everyone she worked with.

This sad story reminds me that it is easier than we like to think for even brilliant and talented people to become so convinced of the rightness of their ideas that they bend the truth until it breaks away beneath them. As far as we know, Elizabeth Dragon set out not to deceive, but to do good in the world. But at some point, she decided she was above the rules of science, and above the law. She was wrong, at great cost.

The Long Tail of Global Health Equity

Partners in Health is a great and thoughtful organization that has a real impact on human health. In this conference they discussed chronic disease among the poorest billion people. PIH's message: the suffering of chronic disease among the poorest can be understood and addressed.

KIF6: Game Over?

JACC recently reported a gigantic meta-analysis study looking at the relationship between cardiovascular disease and the popular test from Celera. Short answer: the study found no relationship whatsoever between test results and cardiovascular disease. None. A test that has been conducted over 150,000 times in the last two years has, according to the authors, no validity.

The KIF6 test measures one SNP in the KIF6 gene called Trp719Arg. Individuals who have the 719Arg mutation, comprising about half of many human populations, were alleged to be at elevated risk for coronary artery disease (CAD). This study involved meta analysis of 19 other studies involving 19,000 CAD cases and 39,369 controls. The study was large enough for the authors to conclude with high confidence that any possible relationship between 719Arg and CAD is extremely weak. The study concludes with a finding of "uniform lack of elevated risk of clinical CAD among carriers of the KIF6 719Arg allele compared with noncarriers in 19 case-control studies performed around the world."

There are some aspects of the study that are open to question, but not much. This study is damning. Either 69 presigious authors completely messed up a high-profile study or KIF6 has no known diagnostic value.

Celera responds that, since the meta analysis did not control for statin use, the study might be masking the importance of the KIF6 variant.

There are two problems with Celara's line of reasoning. First, The Sample quotes Eric Topol and Samir Damani's conclusion that it's not likely the case that statin status significantly affected the study results.

Second, statins are the current standard of care. If 719Arg carriers are getting on statins at a high enough rate to mask the disease impact of the variant, then that calls into question the clinical utility of the KIF6 test. If everyone who needs statins is getting them, then there's literally no point in testing for what would happen to certain genetic subpopulations, were they not to be on statins.

If the KIF6 test could find people who would otherwise get statins but can't possibly benefit from them, then there might be a case for the test, but there's no reason to suspect that the KIF6 test can do that. Even then, it would be hard to justify the expense of the test on this basis.

KIF6 may not be over, but it's not looking good. Celera was widely expected to apply for FDA approval of KIF6. That's suddenly not such an obviously good idea. As the FDA expands regulation of laboratory "homebrew" tests like KIF6, it's not clear that FDA will continue to permit the test to be offered. And besides, even if FDA doesn't shut down the test, cardiologists aren't likely to order a test if they have no idea how to use results in clinical decisions.

Full disclosure: I work for a company developing a CVD test that would arguably compete with KIF6.

Checklist Manifesto

I reviewed this great new book by Atul Gawande here. For the P4 blog, I'll add a few comments realative to P4 Medicine.

Briefly, Atul Gawande's new book is about how simple checklists can bring a surgical team together, reduce infection rates, and save lives.

What does a book ostensibly about surgery have to say about P4 medicine? Well, surgery is personalized, so it's part-way to P4 anyway. But This book addresses issues that are even broader than medicine. Atul's concept of checklists comes straight out of aerospace, particularly pilots' checklists. So the application isn't limited to surgery or even medicine.

In reading this book, one of my thoughts was that many of the same issues with diligent quality control apply to diagnostic tests as well. It's well known that variability in phlebotomy has a dramatically bad effect on many diagnostic tests. Many reference lab tests are quite complex, and especially if technicians don't do these tests every day, it's often easy to make mistakes that can have drastic consequences for people's well-being.

Personalized medicine is going to be personalized in a sense we don't always think of: it will still be performed by people. Personalized medicine means that, in many cases, both patients and clinicians will confront more complex processes. Tools like well-designed checklists are worth understanding, because we'll need them.

Gene Patent Hullaballo

So Judge Robert Sweet has ruled that Myriad's BRCA1 and BRCA2 gene patents are invalid. Well, you aren't supposed to be able to patent genes anyway. Myriad's patent was on a form of matter: the gene encoded by isolated DNA. Myriad and the US Patent office argued that the sequence of DNA isolated from people wasn't just a "law of nature" and was a patentable invention. Judge Sweet was having none of it. Myriad argued that companies would have no incentive to develop genetic tests if they were unable to patent them. Judge Sweet was unmoved.

I have a hard time sympathizing with Myriad. Myriad had a patent on an aspect of biology that was crucial to the health and well-being of many, many women. (And many men as well.) What did they do? Did they license the test widely to facilitate new research and new innovations? No, they did not. I don't buy Myriad's argument that unpatentability will discourage innovation. Look at companies like Genomic Health that have attracted enormous investment in the private and public capital markets for genetic testing, without a single gene patent to their credit. As far as Myriad's claim that lack of patentability will discourage investment, I don't think the facts support Myriad's position. I think it's more likely that the reverse is true: uncertainty and the trouble caused by recalcitrant players like Myriad have discouraged investment in genetic tests. When you don't know for sure how deep your IP stack is going to be, that's uncertainty. Uncertainty equals risk, and risk drives away investment.

It will be very interesting to see what happens next.

Hood: ISB plans to double in size, focus on P4

In an interview with Genomeweb, Dr. Leroy Hood provided an update on ISB's expansion plans, as well as comments on Integrated Diagnostics and the P4 Medical Institute.

Celera's Dx Revenues Down

Celera's Q3 2009 diagnostic revenues declined 24% compared with Q3 a year earlier. Celera reports this revenue as lab services, conducted by their Berkeley Heartlab subsidiary. So this is a decline in the premier high-value cardiovascular Dx.

What happened? Celera's 10-Q tells the story. Sample volume slipped, but only a little. The big problem is getting tests paid for:

Revenues from our Lab Services segment for the three months ended September 26, 2009 decreased $5.9 million compared to the three months ended September 27, 2008. The decrease was primarily due to lower reimbursement rates, reflecting the continued impact of denied tests and historical collection activities.

It looks like Celera is facing something of a headwind in getting both payors and patients to pay for ordered tests. This is partly an effect of the economy, but probably not the whole story. Are payors becoming resistant to Berkeley's value proposition? I have no evidence that they are, but if so that's a caution for the whole emerging field of high-value Dx. Cheap tests with limited predictive value may be better than nothing, but if you are charging the big bucks, payors probably will keep the pressure on you to deliver measurable results.

Or maybe not. That seems an obvious lesson to draw from Celera's results, but there are other possible explanations.

To grow the business, Celera is licensing several US labs to perform some of Berkeley's tests. These deals involve several different tests, and Celera's payment issues may not relate to any of these licensed tests. It will be nonetheless interesting to see how these deals work out.

(Full disclosure: I work for a company developing cardiovascular diagnostics.)

Breast Cancer Fund: Are Mammograms So Great?

In the 12/6/09 San Jose Mercury News, the Breast Cancer Fund's Jeanne Rizzo writes:

A blanket guideline for mammography is the antithesis of personalized medicine and belies the complexity of breast cancer. The reliance on mammography as a tool for detection also undermines the need for directing our national health resources toward the development of noninvasive alternatives that truly address prevention.

Rizzo makes an important point: lost in the debate over mammography for women in their 40's is the unfortunate fact that x-ray, our front-line diagnostic for breast cancer, is carcinogenic. Despite dramatic progress, our tools for diagnosing and treating breast cancer are still barbaric compared to what we should want women to have available to them.

Report: The New Science of Personalized Medicine

PriceWaterhouseCoopers has a new report out on Personalized Medicine, which is described explicitly in terms of P4 medicine, although the authors concede that Personalized Medicine is a fluid concept, with many definitions.

The report sizes the Personalized Medicine market at $232 billion, with an 11% annual growth rate. Yes, a quarter of a trillion dollars, although most of that money is in nutrition and wellness services that some people would place outside of "medicine."

The core Dx and theranostic applications of personalized medicine are a still impressive $24 billion market growing at 10% per year. The report breaks this down as as follows:

• Esoteric lab services $6B, 10% CAGR
  
• Esoteric tests, including molecular diagnostics: $5B, 13% CAGR
  
• Target therapeutics $13B, 9% CAGR
  

The report is particularly favorable towards molecular diagnostics. The report acknowledges that molecular diagnostics are expensive, but notes that payors will pay for them if the tests improve patient outcomes, especially if the tests identify patients who don't need expensive therapies.

This is a nice report with a good data-to-speculation ratio, all the more impressive in a free report. Give it a read. It's a good source of quotes for business plans, and you'll probably learn something too.