January 1, 1970 (Vol. , No. )
Zachary N. N. Russ Bioengineering graduate student UC Berkeley
Pharmaceutical firms, patients, and even the FDA couldn’t be more excited about their new friend: the cancer companion diagnostic. These diagnostics allow drugmakers to better characterize patient tumors and direct their therapies to the patients who would see significant benefits. The first diagnostic/treatment pair, Herceptin and HercepTest, was approved in 1998. Since then, drug and diagnostic pairs have appeared on a regular basis, and partnerships continue to be announced.
What Is Wrong and How Can We Fix It?
Of course, diagnostics are nothing new. Human medicine is the art and science of helping repair the human body. As with fixing anything, you try to identify the cause of the problem and fix it, or, if it cannot be fixed, use that knowledge to ameliorate the symptoms.
The continuous refinement of diagnostics and treatments is an arms-race of sorts, but an unbalanced one: diagnostics will always be in the lead, as they guide the application of treatments. Applying treatments without a good diagnostic is an expensive form of Russian roulette.
Not only do the treatments have costs and side effects, but, as in the case of many cancers, the time lost in applying the incorrect treatment may be a death sentence when it impedes the delivery of an effective treatment. This knowledge is valuable even in the absence of an effective alternative treatment, because many patients would prefer to make the best of their remaining time rather than endure debilitating therapies.
A Personal Companion
That’s why better diagnostics are generally good to have. But when developed alongside new treatments, they take on a new role: They help treatments pass through clinical trials by removing patients who would see no benefit from the medications.
Antibiotics don’t help people with viral infections; that’s why strep tests are done. Even though the differences between cancers are more subtle than those between viruses and bacteria, the treatment efficacy is just as much at stake.
What was once simply called a carcinoma became “estrogen receptor-positive” and then “Her2-positive” as the disease became better understood. But this isn’t the endpoint, at least, not while the costs of misdiagnosis are so high. As whole-genome sequencing costs drop below $10,000 per genome, and since Phase II and III per-patient costs can run as high as $85,000, it will become feasible to run a full battery of tests to get as much useful information as possible. In the coming age of personalized medicine, it is beginning to make sense to characterize not only the cancer, but the person.
Diseases do not operate in a vacuum, and neither do their treatments. Just as cytochrome P450 polymorphisms altered the way some people respond to warfarin (and hundreds of other drugs), mutations in antibody receptors change the way a patient responds to antibody therapies (including Herceptin, which targets Her2-positive cancers). Certain P450 mutations manage to play a role in cancer therapy as well by increasing conversion of tamoxifen to its metabolites and improving the activation of the drug.
So, just as E. coli strain W3110 has “F- Lambda- rph-1 INV(rrnD, rrnE)” as its genetic characteristics, we might see a patient cancer profile with “ER+++ HER2_IHC2+/FISH+ || BRCA1+/+ CYP2C19*17 …” including not only the cancer characteristics but also the overall genetic profile.
Three’s a Crowd
There’s just one hitch: Ordering all of these tests individually would be expensive, time-consuming (if ordered sequentially as many healthcare systems do), and require sample preparations for each separate test. A single, unified test system would address these problems, but the way to create one isn’t clear-cut, either.
The tests themselves don’t necessarily lend themselves to multiplexing as some are immunohistochemistry-based (most of Dako’s and Ventana’s lineup), while others are nucleic-acid based (MDxHealth and Dako). There are also competing tests for the same characteristic, such as Herceptest, CB11, and 4B5 antibodies for detecting Her2; some papers have even suggested using these tests to verify each other.
The problem is not intractable, however. Technician time is expensive, and automation could provide the answer in preparing multiple tests simultaneously at lower cost—the overall mass of the samples necessary isn’t so large that a single sample cannot be split up to be used for multiple protocols, and some nucleic acid tests are easily combined into a microarray. Patent pooling could be used to bring together the necessary IP, and data collected would permit better understanding of cancer mechanisms.
If a unified test could be designed to recommend the top three most effective treatments for a given patient, it would assist patients and doctors while saving money in the long run by eschewing ineffective treatments. Even dosages and adjuvants may be affected, as some nontarget polymorphisms affect pharmacokinetics and efficacy of small molecule drugs and antibodies. Such a test would be valuable for everyone . A given treatment might apply to 20% of the patient population, but the test would be appropriate for 100%.
Whether a unified test appears or not, the information granted by companion diagnostics will remain valuable in both product development and patient treatment. If knowledge is power, then companion diagnostics are powerful friends, and it never hurts to have powerful friends.