A recent article in the JAMA has received a lot of coverage in the lay press. It analyzes screening for breast and prostate cancer. Critics of both screening tests (including me) have, over many years, pointed out the problems inherent in screening for any disease, but most specifically these two. We mostly have been ignored.
The rationale driving screening for these two cancers (or indeed any cancer) is that the earlier the tumor is diagnosed the better the likelihood for cure. Screening for these two cancers has had limited or no effect on treatment outcomes. First we need to distinguish survival time from mortality. Assume a patient has a cancer which will kill him in 10 years irrespective of treatment. If the diagnosis is made nine years after the onset of the cancer survival time is one year. If the diagnosis is made one year after onset survival is nine years. But the patient dies at exactly the same time. Survival has increased without affecting mortality. It took decades to make surgeons and oncologists admit to this bias of the early diagnosis of a tumor that lacked effective treatment. They have, in the main, refocused on age adjusted mortality which gives a better picture of where we are regarding diagnosis and treatment of cancer. But an improvement in age adjusted mortality cannot distinguish a salutary effect secondary to early diagnosis from that due to new and effective treatment.
Next assume another scenario. The patient has a tumor of such low grade malignancy that he will die of something else before the “cancer” kills him. Here early diagnosis and treatment seem to have cured the patient. He’s diagnosed, treated, and seemingly “cured”. But there’s no other possible scenario. He was never going to die from his tumor no matter how much or how little was done to or for him.
Both these scenarios offer the patient nothing but side effects without any therapeutic benefit. In both of them the patient would have been better off never to have been screened. The figure below from the JAMA paper shows four different screening outcomes. Cancer progression is plotted against time.
Tumor A never gets large enough to cause any problem. If it were not detected by a very sensitive screening technique the patient would live his life in blissful ignorance of its presence. Tumor B gets big enough to be detected by imaging or by a biopsy, but it too never causes a threat to life or health. Tumor C will eventually metastasize and kill the patient if not detected early and treated. Tumor D is so malignant that it likely has spread before it can be detected and treated. Note that screening is of no benefit in three of these four tumors.
There is another possible scenario that is typically overlooked by physicians in this field. In this one the tumor spreads as soon as it appears. It doesn’t have to go through the stages on the figure. Here again screening will be useless.
Perhaps as high as 90% of prostate cancers will cause no harm. Treating them, however, causes significant morbidity. “Even in breast cancer, for which there is evidence and agreement that screening saves lives, …for every breast cancer averted, even in the age group for which screening is least controversial (age 50-70 years), 838 women must undergo screening for 6 years, generating thousands of screens, hundreds of biopsies, and many cancers treated as if they were life threatening when they are not.” [The quotation is from the JAMA paper linked above.]
Managing these diseases would be vastly improved if they were like colon cancer. Here we can identify benign polyps that if not removed would become malignant. Thus we can prevent colon cancer rather than having to treat it. Screening in this instance makes a lot of sense.
Our problem is that we have difficulty telling the tumors that will benefit from screening and early diagnosis from those that won’t, though we are getting better at differentiating them. The authors of this study make several recommendations, all reasonable.
1. Develop and validate biomarkers to differentiate significant and minimal risk cancers.
2. Reduce treatment burden for minimal risk disease. Rightly believing that a lot is in a name, they advise not calling minimal risk disease cancer. The word is too loaded to use casually.
3. Develop tools to support informed decisions. Yet another plea for doctors to talk to their patients about the implications of screening before they do it.
4. Focus on prevention for the highest risk patients. This is easier said than done, but we are making progress here and should do better in the near future.
We’ve been trying to get both physicians and patients to face up to the complexity of screening for years. This holds true for screening for any disease. This paper should help a bit.