What constitutes a good cancer screening?
Tuesday, November 13, 2012 9:00 PM
My patients often ask me to do tests to make sure they don't have cancer. Our ability to detect cancer early when treatment is more likely to lead to prolonged survival or even a cure is not as simple as it seems. This has hopefully become more apparent to the general population as the arguments have unfolded in the popular press for and against screening tests like mammography for breast cancer detection and PSA blood tests for prostate cancer.
So, what constitutes a good cancer screening test? First of all, screening tests are developed for conditions that have a high incidence in the population and that, if found, respond to treatment that will extend the patient's life. Screening tests must also pose little risk to the patient (more on that below).
When evaluating whether or not to get a screening test, it's very important to know how accurate the test is. For that, you need to have a basic knowledge of statistics. The best screening tests have a high specificity or "true negative" rate (if the test is negative, it's less likely the patient has the condition) and high sensitivity "true positive" rate (if the test is positive, it's more likely the patient has the condition). So, the optimal screening test has 100 percent specificity (negative = no disease) and 100 percent sensitivity (positive = has disease). There is no such test.
For example, the sensitivity of mammography to detect breast cancer is about 79 percent (even lower in women 40 to 49 due to dense breast tissue) and the specificity is about 90 percent. So, with a sensitivity of 79 percent (a negative test will be truly negative about 79 percent of the time), around 20 percent of women who are told they have a normal mammogram will have breast cancer. The converse is also true - the 90 percent specificity of the test means ten percent of women who are told they have breast cancer on mammography actually do not.
In regard to men and prostate cancer, the reliability of PSA testing is much more complex than mammography. The sensitivity and specificity vary widely based on the context in which the PSA value is used. We often don't just look at the absolute PSA number - we look at the rate of change from test to test, the value in relation to the size of the prostate, whether or not the clinician finds a lump on the prostate, the age of the patient, etc.
I promised to address the risks of cancer screening. I think most people who know there is a screening test for a certain cancer want to have it so they can get early treatment and live a longer, hopefully cancer-free life. However, recall that pesky, true-positive "sensitivity" statistic.
Using mammography as an example, a sensitivity of 79 percent means that 21 percent of patients who have a positive test will not have cancer. This becomes problematic when we are obligated to start looking for a cancer that does not exist. Women worry, get breast biopsies that come back negative, are subjected to more intense testing, etc. There are also the risks of surgery - anesthesia, infection, scarring, etc.
PSA screening can be even more problematic, having an even lower sensitivity. Men often get prostate biopsies and treatment that may include surgery or radiation. In addition to the surgical risks, they can develop impotence, incontinence, bowel problems, etc.
Does screening and catching a cancer early prolong someone's life? One concept I need to address that helps answer this is termed "lead time bias." This is very important since many screening tests claim to "save lives." There is a stark difference between living longer after a diagnosis of cancer vs. living longer.
Using PSA screening as an example, say a man chooses to not see a doctor and develops prostate cancer at age 50. Since the majority of prostate cancers are slow-growing, he does not develop symptoms of cancer until age 75 and dies at 80. He lived 30 years with cancer, but was not aware he had it until 75 - as far as he was concerned (and the people who keep cancer statistics), he only lived for five years after his diagnosis.
Now suppose the same man sees his doctor and has a PSA test at age 50 that finds his cancer. He receives treatment for it and dies at 80. He lived for 30 years after his diagnosis, so it appears the treatment saved his life and gave him 25 more years than if he had not received treatment (30 years vs. 5). Here's the kicker - the man who received treatment at 50 developed treatment complications and had 30 years of urinary incontinence and impotence.
Who had the better quality of life? These are the questions physicians and patients have to mull over when deciding whether or not to pursue cancer screening. I promised you it's not as simple as it seems.
Dr. John Roberts is a family physician. He is also one of the owners of The Paper of Montgomery County. Send him your question today by e-mail at firstname.lastname@example.org.