The Myth of Early Detection

The (false) message that cancer screening cuts mortality is 'in our bone marrow.'

Talk about missing the point. In the fortnight since two studies, taken together, showed that the PSA test for prostate cancer does not decrease mortality from that disease, pundits and bloggers have been sputtering with outrage and grasping at straws. The studies were flawed, critics said. Neither followed the men long enough. A conclusion about patients in the aggregate can't apply to individual men.

Please. Let's stop arguing over arcane details and splitting hairs over whether the PSA test reduces cancer mortality by a little (as one study found) or nothing, and focus on the truly infuriating, head-scratching thing about all early-detection studies: that the results are not an unambiguous, slam-dunk positive. If the hype is right, then finding a tumor when it is small should mean a better chance that surgery, radiation and/or chemotherapy will eradicate it, allowing you to live until something else kills you. Early detection should produce clear, unquestionable benefits.

Yet it doesn't. Not the PSA, not early detection of lung or testicular or pancreatic cancer, or glioblastomas, a type of brain cancer. Even mammography is iffy, since trials showing a decrease in mortality were done before the age of adjuvant therapy. Colonoscopy might cut mortality, but it has never been tested in a randomized, controlled trial, notes Barnett Kramer of the National Institutes of Health, lead author of one of the new PSA studies. The Pap test for cervical cancer and the fecal occult blood test for colorectal are about the only screening tests shown to decrease mortality from the cancer they target. If the failure of most early detection to strongly affect whether you live or die makes no sense to you, you're not alone. "The medical community has done such an extraordinary job getting out the message that early detection is tantamount to cure, it's incorporated into people's bone marrow," says Kramer.

One reason early detection doesn't make a bigger difference is that, absent effective therapy, it hardly matters when a tumor is found: early or late, you're doomed. This is pretty much the case, sadly, for glioblastomas and pancreatic cancer. But the benefits of early detection also vanish if therapy is so terrific, or the malignancy so slow-growing, that the cancer can be vanquished even if caught late. This describes many testicular cancers. It is for cancers such as breast and colorectal, where therapies sometimes succeed and sometimes fail, that early detection should help.

But the natural progression of cancers also undercuts the value of early detection. "All the common cancers have a spectrum of aggressiveness," says Kramer. Some are slow-growing, while other cancers of the same organ are aggressive and deadly, speeding from birth to an advanced stage before you know what hit you. If you look for cancer in asymptomatic people—as screening does—you are much more likely to find an indolent one for the simple reason that indolent cancers, by definition, spend more time in an early, nonthreatening stage, whereas aggressive cancers speed through that early stage. "You may pick up tumors that can be cured, but which never needed to be treated in the first place," says Kramer. Failing to detect an indolent cancer early doesn't necessarily put you at much of a disadvantage.

Last week, another study underlined the futility of early detection, this time in ovarian cancer. Scientists at the Duke Comprehensive Cancer Center compared the gene-expression profiles—which genes are expressed at high levels and which at lower levels—in 166 ovarian cancers. They categorized each cancer by whether the woman lived at least another seven years or less than three. Of the 40 cancers detected at an early stage, 39 had the genetic profile predicting long-term survival. This suggests that slow-growing tumors spend more time in an early stage, and that they are the ones more likely to be detected before they spread. Yet because they are more indolent, they pose less threat, explains Duke's Andrew Berchuck, who led the study. Other scientists have found much the same in lung cancer and prostate cancer, where screening also preferentially detects more indolent cancers that are less likely to be fatal regardless of whether they're found and treated early. "This is why it's so hard for screening and early detection to make a difference," says Berchuck. "To save lives, we have to detect aggressive cancers, but early detection seems to preferentially find ones that are inherently less life-threatening. It sounds sensible that early detection will save lives, but what you tend to find with screening is the cancer equivalent of a cold virus, not HIV."

How much less threatening are cancers detected early? Last November, scientists reported that about one quarter of breast cancers detected (early) on mammograms vanish spontaneously. Yet breast-cancer survivors swear early detection saved their life. Some melanomas, kidney cancers and neuroblastomas perform a similar vanishing act, says Kramer. Will doctors' enthusiasm and patients' demand for cancer screening diminish as a result of the science? After the PSA studies came out, a scientist told clinicians he assumed so. They looked at him as if he were crazy. No matter what science says, it will be a cold day in hell before patients let go of the one slender hope they feel they have to beat cancer.