These slow-growing - and sometimes non-growing - tumours look the same as life-threatening ones under the microscope; they only behave differently in the body. The latest evidence suggests that regular screening is most likely to pick up these slow-growing, non-lethal cancers and lead to overtreatment that may actually increase death rates rather than reduce it (Arch Intern Med, 2000; 160: 1109-15; Lancet, 2001; 385: 1340-2, 1284-5).

Knowledge after death Evidence from autopsies has taught us some amazing facts about cancer. These examinations regularly turn up otherwise undetected cancers that were not the cause of death. Undetected cancers, of course, make a mockery of the official cancer registries since, clearly, a sizeable proportion of cancers are never diagnosed. They tell us that the incidence of cancer is much higher than we believe it to be. They also tell us that cancer is not always a killer.

These undiagnosed cancers are referred to in the medical dialect as ‘disease reservoirs’. When Swedish scientists spent a year of concentrated effort in an attempt to find all the lung cancer cases in the country, they discovered that the true rate of lung cancer in Sweden was 40-50 per 100,000, and not the 30 per 100,000 they thought it was. That’s a significant 30-60 per cent ‘reservoir’ of undetected cancer (Lung Cancer, 2002; 37: 137-42).

Other studies have shown high rates of lung cancer only detected after death (JAMA, 1987; 258: 331-8; Chest, 1986; 90: 520-3). Reasons for the lack of diagnosis were in part because some elderly patients were simply too sick to undergo diagnostic testing for troublesome symptoms. But another reason was that patients showed no symptoms that betrayed the presence of cancer - and most were non-smokers, a group unlikely to be referred for lung cancer investigations in the first place.

Another Swedish study found that as many as 15 per cent of major cancers were not diagnosed before death, and around half of these were of a type normally considered fatal (Hum Pathol, 1994; 25: 140-5). In this study, the discrepancy between medical diagnosis in life and autopsy findings after death was higher in elderly patients, a finding that echoed an earlier Swedish study which concluded that undetected cancer in this older age group may be the result of undifferentiated symptoms such as weakness and fatigue as well as the type of tumours detected, which were often small and slow-growing (Nord Med, 1989; 104: 23-4, 29).

However, as some researchers have also discovered, cancer in older people is generally less aggressive than those in younger people - though no one is sure why (Int J Radiat Oncol Phys, 1982; 8: 1471-80; Cancer J, 1994; 7: 212-3; McKay FW et al., Cancer Mortality in the US, 1950-1977, NIH Publ No. 82-2435, 1982).

Other studies have found high rates of undetected colorectal cancer after death (Gastroenterol J, 1989; 49: 26-8) and, when US researchers reviewed autopsy studies of women not known to have breast cancer and who died from other causes, they found that 1.3 per cent of the women had occult (hidden) invasive breast cancer and 8.9 per cent had DCIS (Ann Intern Med, 1997; 127: 1023-8). Taken as a whole, such findings clearly have implications for what it really means to have cancer.

Why does cancer go away? The mechanisms of spontaneous remission are by no means fully understood. The most popular theory is that some form of immunological reaction occurs, though this is still unproven (Onkologie, 1995; 18: 388-92).