PJ's case highlights an emerging medical dilemma: catching cancer early and treating it aggressively may not always be the ideal strategy.

For a range of adult cancers, too - including prostate, breast and lung - some doctors are wondering whether search-and-destroy missions against the tiniest clusters of dubious cells really benefit patients, or whether too many people suffer needless upheaval, anxiety and even surgery in the fight against this most feared disease.

For decades, early detection has been a cornerstone of Australia's cancer policy, as doctors and the community accept the apparently obvious: the sooner a tumour can be found and treated, the less likely it is to kill.

Screening programs have been introduced, such as the BreastScreen X-rays and the prostate specific antigen (PSA) blood test, burgeoning in popularity despite no government promotion. By searching for traces of cancer in people with no symptoms, screening has brought diagnosis inexorably forward, turning larger cohorts of younger people into cancer patients or survivors.

But what if we have fallen too hard for the early detection mantra? What if more cancers than we ever knew lie dormant and virtually harmless for many years? What if - for some at least - the cure is worse than the disease?

Professor Ian Olver spends his life meeting cancer patients and cancer survivors. Those who have come through the disease, says the chemotherapy specialist and Cancer Council Australia chief executive, almost always are convinced the benefits of screening outweigh the negatives. "They're generalising their own experience," he says

But objectively, the value of screening depends in large part on how much harm is done by positive results that are wrong, or by people having more invasive tests such as biopsies, or being treated for cancer that left alone would not have threatened their lives.

Most of all it depends on whether early detection saves lives. Despite the range of screening tests available - and the more than $250 million Australian governments spend on them each year - that essential evidence is often ambiguous and sometimes absent.

Even when it is more clear-cut, the number of people who must be screened to save a life is typically much larger than many would expect. A Sydney University study found just two breast cancer deaths might be prevented by screening 1000 women for a decade. One bowel cancer death can be avoided by screening 1274 people for five years, according to an analysis in the United States.

Nearly a million Australian men took a prostate specific antigen blood test last year, including many who had no symptoms, checking for signs of possible prostate cancer. But the first reliable study on whether such tests prolong life will not report until late next year. In the meantime, men must punt on whether to be screened. And the growing number of men whose prostates have been removed after a PSA test cannot know their decisions were sound.

Olver says that after a prostatectomy, men often become cheerleaders for PSA screening. "Of course you'd advocate for it; [they believe] it saved their lives. They may have been over-treated, but try telling them that. It would be almost cruel to do so."

Dr Barnett Kramer is the associate director for disease prevention at the US National Institutes of Health. He says the message that early detection is always preferable has become an "oversimplified sound bite ... it's easy to believe that it must be so".

Intercepting cancer sooner is only part of the story, Kramer says. "Earlier diagnosis is not equivalent to early enough diagnosis" if the disease has spread. Even in these cases, however, screening and early detection can be made to look successful because medical statisticians usually talk in relative terms - counting the years people survive after cancer diagnosis, instead of the absolute measure of whether they are alive or dead by a certain age. Additionally, screening may exaggerate the survivor tally because it detects cancers that do not cause symptoms and are not identified by other means.

Even when some lives are clearly saved, Kramer says, that still needs to be viewed in a wider context.

Figures from the Australian Institute of Health and Welfare show cervical cancer deaths in Australia a year dropped from four to two per 100,000 women in the 20 to 69 age group, between 1991, when the the pap smear program was introduced - it detects treatable pre-cancerous abnormalities as well as cancer - and 2005. But more than 700 high-grade abnormalities are found per 100,000 women screened. Occasionally, treating these abnormalities can scar the cervix and lead to infertility or premature birth.

More and more is understood about how cancer works at a microscopic level: gene mutations that cause it, how it evades immune defences, the chemical byproducts that signal its presence. Surprisingly less is known about the ecology of cancer in the human body: how often cancer cells occur, how quickly tumours spread, why it lodges where it does.

Many cancers, for example, never come to light in people's lifetimes. A review last year of forensic autopsies in Hamburg, Germany, found more than a quarter of the cancers identified were undiagnosed. That included 17 per cent of the apparently fatal cancers, which were found only at autopsy.

When knowledge of the normal progression of a disease is flimsy, the effects of intervening early to stop it are also unclear.

A form of breast cancer called ductal carcinoma in situ (DCIS) was rarely found before widespread mammogram X-rays.

Now, it accounts for about one in five diagnoses made by the BreastScreen program.

Certainly, some cases of DCIS - in which cancer cells are confined inside the milk ducts, and have not penetrated the membrane into the surrounding gland - progress to become invasive breast cancers. Others - the "indolent" forms - remain indefinitely harmless in the closed package of the duct. The problem is that doctors cannot determine which course a group of abnormal cells will take.

A National Breast and Ovarian Cancer Centre analysis found that about 9 per cent of breast cancers identified through screening would not progress to life-threatening status, and about three-quarters of them are DCIS. Clearly, then, a significant proportion of women who have surgery and radiation therapy for DCIS could do without it.

Dr Helen Zorbas, the centre's director, takes a different perspective. "Some people wouldn't agree there is such a thing as over-diagnosis [of DCIS]," Zorbas says. "The question is, are there some cancers we can leave untreated?

"Given our current understanding of the biology of the disease, we couldn't say to anyone with any certainty that if you left this alone you would be OK. Patients, given the choice, would rather have treatment."

An Australian audit published last year by the Royal Australasian College of Surgeons revealed too many women with DCIS were having operations on the lymph nodes in their armpits - which can result in swelling and restricted movement - without good evidence the procedures were warranted. For those women, it can be argued that screening mammograms intended to protect their health led directly to excessive surgical treatment, although others may benefit from having more minimal lumpectomy surgery instead of a full mastectomy.

Sally Crossing, the chairwoman of Breast Cancer Action Group NSW, says some women may be comfortable with a more laissez-faire approach, while for others the logical thing is to hit each cancer cell with the full armoury.

"The point is, women want control of their bodies," Crossing says. "We don't want these do-good public health people to say [screening is] harmful. When they talk about 'harm' it's not a harm that people perceive. They're studying things researchers like to study. They're not studying the things that are important to a human."

The US National Cancer Institute in 2005 analysed the contributions of better treatments and earlier detection to falling death rates among breast cancer patients since the 1970s. It credited screening mammography with nearly half the improvement.

"That's cleared the air a little bit," says Professor Jim Bishop, the chief executive of the Cancer Institute NSW and overseer of the state's screening programs. "We've got no doubt [breast] screening works." The remaining question, Bishop says, is whether the screening program is designed as well as it could be - including whether it is targeting the right women, identifying a realistic number of cancers for the number of women recalled for extra testing, and following them up appropriately.

One reason breast screening works well is that so many women - one in 11 in Australia - develop breast cancer. Community screening will identify an acceptable proportion of true cancers to false alarms only if the disease is common, and the test has a high pick-up rate.

For an uncommon malignancy, such as ovarian cancer, which affects one in 2500 women over 50, screening has to clear a higher accuracy hurdle because the chances of detecting a real case of the disease are so much less.

Professor David Bowtell, the director of research at Melbourne's Peter MacCallum Cancer Centre, is running computerised genome scans on blood from thousands of women with ovarian cancer to try to identify gene mutations more common among women with the disease. If those same patterns occurred in blood samples from healthy women, researchers could perhaps narrow the group who would benefit most from screening tests.

The well-known gene mutations BRCA1 and BRCA2, which greatly increase breast and ovarian cancer risk, are only part of the picture. Many other variants might cause smaller increases. "Now we need to look at how the genes interact with each other, if there is a compounding effect" of multiple mutations in the same woman, Bowtell says, which might explain why cancer runs in families even in the absence of high-risk genes.

The Cancer Council's Ian Olver agrees individual gene scans will one day supersede screening programs that target the entire population. And the current medical debate on the pros and cons of screening is, he says, "just an aberration in time".