I talked about the importance of early cancer diagnosis in a previous post, and reader Ian wrote back to ask for more detail. Early detection can be life-saving, but accuracy in these tests is also a serious problem - a test that misses existing cancers is obviously bad, but one that detects cancers that aren't there or suggests a future cancer that will never develop can expose patients to unnecessary procedures, some of them invasive. I'm not knocking cancer screening — just noting that, for a given test, the potential for early detection is not the whole story. Early unambiguous detection is the goal. Let's take a look at a few of the methods we can apply to improve cancer diagnosis.Genetic screening: find out which people are likely to develop certain cancers in the future. Certain genetic mutations predispose a person to developing various cancers. While no mutation is a guarantee that cancer is in a patient's future, in a few cases the tendency is so strong that pre-emptive treatment is indicated, though earlier and more frequent screening is more common. It's estimated that 5-10 percent of cancers are strongly hereditary, along with another 20-30 percent that are weakly hereditary. Certain mutations of the BRCA1 and BRCA2 genes , for example, increase the lifetime chance of a woman developing breast cancer from around 13 percent to somewhere between 36 and 85 percent. Research for other risky versions of genes continues, and will eventually branch out into risky combinations of genes that appear in single nucleotide polymorphism genotyping (for a few hundred dollars) and complete genome sequences (soon to be a few thousand dollars).

When personal genomics and medical histories (finally) come together, with a little luck we'll be able to find new correlations between genes and cancers. Family history is a good but imprecise indicator of risk; a better understanding of the specific genetic factors behind it will improve a physician's ability to assess an individual and design a regimen of treatment or prevention appropriate to the patient. Tumor markers: search for indirect signs of an existing or developing tumor. Usually this refers to a blood or urine test, though saliva or even smell are also options. If you can find a substance that increases or decreases in abundance when a certain kind of tumor is present, assays that look for these tumor markers can be used for routine evaluation. One of the most common markers is the prostate specific antigen (PSA) - elevated levels of PSA in the blood are often (but not always) associated with prostate cancer. Because the test is prone to false positives, a worrisome blood test result is usually followed by ultrasound imaging. Looking for multiple tumor markers at the same time can potentially be more accurate, while new technologies decrease the time and expense of the necessary assays. For tumors caused by infections, the presence of the infection itself can be used to determine at least risk - while a high-risk human papillomavirus infection certainly doesn't guarantee a cervical cancer, it does suggest that more frequent screening may be in order. Similar profiling of the tumors themselves could indicate what sorts of drugs would be effective against an individual's particular tumor. Imaging: look for the damn things. This is a tricky business, because the difference between a benign and a cancerous (or pre-cancerous) mass can be difficult to tell without going in, taking out a sample, and subjecting it to various tests. In a colonoscopy, many suspicious lesions turn out to be harmless, but they still require a biopsy, which is even less fun than a colonoscopy without a biopsy. If you want to avoid bringing the cells to the microscope, bring the microscope to the cells - the pCLE system is effectively a fiberoptic microscope that can examine lesions in the colon without a biopsy. If possible, avoid the colonoscopy entirely - take a series of radiographs of the colon and turn them over to the computer to reconstruct its 3D structure. Examination of a virtual colon is far more comfortable than the alternative.

If a computer can be used to verify a doctor's "eyeball spectroscopy" (as in computer-assisted mammography, currently in clinical trials), the visual analysis of an expert can be combined with digital image analysis to separate the healthy from the suspicious.


Displaying mammography results in 3D is also potentially useful in helping physicians to find suspicious masses, while improvements in MRI technology work to increase the resolution of images generated by the technique. Imaging can also be combined with treatment to increase its effectiveness. Cancer-seeking nanoparticles can also be detected using various imaginge techniques - a build-up of the particles is desired at the cancer site, but not elsewhere, and being able to detect where the treatment is collecting is a partial measure of its efficacy. Finally, technologies like the CyberKnife combine imaging and robotics to guide radiation delivery to the patient by zeroing in on its target and following it as the patient moves.


A good technique catches cancer early and is as unambiguous as possible. Unfortunately, early and unambiguous is a tall order. Do you have questions you've always wanted to ask a biogeek? You can email me.