Friday, February 20, 2009

Cancer Screening

When the CT scanner first appeared people dreamed of everyone having a scan at regular intervals to detect early cancer. We have become used to the mantra that in cancer "Early diagnosis means early treatment means more cures and fewer cancer deaths". Screening has been so heavily backed by government information that we have all bought into it. In today's BMJ is a paper from Peter Gotzsche and his colleagues in Denmark which questions the value of mammography. They produce the following figures: If 2000 women are screened regularly for 10 years, one will avoid dying from breast cancer, but 10 healthy women will as a consequence become cancer patients and be treated unnecessarily. These women will have whole or part of their breast removed and some will also receive unnecessary radiotherapy or chemotherapy. Some will develop a secondary leukemia because of the therapy. Furthermore another 200 healthy women will experience a false alarm and suffer psychological trauma.

All this sounds counter-intuitive. Surely it is better to know what is going on?

The problem with breast screening is not that cancers are missed, but that they are over-diagnosed. It is becoming clear that cancer may be diagnosed when the tumor is very small and despite its grim reputation, many cancers do not progress during the lifetime of the individual. Those of us who work with CLL are well aware of this fact: the commonest treatment applied is watch and wait. I am reminded of an obituary of a man from Oklahoma who died in his eighties, 52 years after his untreated CLL was first diagnosed. If it is true for CLL, why would it not be true for cancers of other tissues. Since CT scanning has become so sensitive we have been recognizing very small lumps in the lungs. Do these represent lung cancer? Or perhaps marginal zone lymphomas? Biopsy is the only way of finding out, but isn't that a bit invasive for what may be a false alarm? So we tend to watch and wait there too.

With mammogram results it is relatively easy to do a biopsy and this leads to the possibility of over diagnosis. For a start there is carcinoma-in-situ, which constitutes 20% of the diagnoses made. We know that fewer than half such cases lead to invasive cancer, but 30% are treated with mastectomy. Then there are patients who really do have cancer, but such an indolent cancer that it would never have become noticeable in the patient's lifetime.

Screening does not lead to fewer mastectomies; indeed in randomized trials 20% more mastectomies are performed in screened patients. You would think that this would be offset by a reduced number of mastectomies in older women whose late-occurring cancer had been forestalled. Unfortunately, this is not so. Radiotherapy is applied to some women whose cancer would not have progressed. It is known that radiotherapy doubles teh rate of mortality from lung cancer and heart disease.

Breast cancer rates are apparently increasing because mammography finds more cases. The cure rate is also improving, but the absolute number of women dying from breast cancer has not changed.

Similar results are available for PSA screening for prostate cancer. The disease is not 10 times more common than it was in the 1990s, but the number of people dying from prostate cancer every year has not changed - it remains the same in countries which adopt both a restrictive and a liberal policy on the use of PSA as a screening test.

I speak with some feeling as someone whose screening colonoscopy has led to two further colonoscopies, and octreatide scan and two CT scans, and still no diagnosis.

To explain the problem it helps to look at the maths. Suppose that a screening test is almost completely accurate; that it misses no positive cases and is 99.99% accurate. That means that one test in 10,000 will be a false positive. Not many tests are as accurate as this, but suppose you are screening for a rare disease with an incidence of 4 in 100,000 in the general population. That means that for every 10 positive tests, 6 will be false positives. (4 in 100,000 is the approximate frequency of CLL).

We underestimate the harm done by worrying patients. Few there are who face impending doom with equanimity.


Anonymous said...

Dear Dr.Hamblin, thank you for this post. I agree with you - today we just go too far with all these tests and screeings. I wish I could postpone my "knowledge" of having CLL. I feel just fine, but the diagnosis itself nearly killed me. Only now, 10 months later I gradually become adjusted to this wait and watch thing. All the prognostic tests did not help - I have discordant markers, my genes are unmutated, but CD38 is 15% and ZAP is 1%. All I learned and understood is that CLL is different in everybody and unpredictable. But so is life!
Thank you for your posts and may God bless you and keep you healthy and happy!

Marcia said...

Thank you for the report on the BMJ article. In my case I think the hardest hit is my 92-year old mother. Two of her three children have been treated for cancer (prostrate for my brother, CLL and breast cancer for me) and the youngest has had 5 stents installed in his heart arteries (I read an article the other day questioning that procedure). Poor Mama is so afraid she will outlive us. She has been incredibly healthy, but does not handle worry very well as she ages.
My breast cancer was also found in one lymph node, so it was diagnosed as invasive. I will be having radiation therapy because I had a lumpectomy rather than mastectomy. Will the increased risk of lung cancer and heart disease make them more dangerous than the recurrence of breast cancer if I do not have radiation? How is one to choose?

Terry Hamblin said...

This highlights another risk of statistics. Supposing one person in a million gets a certain disease - in other words 999,999 dont. Something happens so that 2 people in a million get it - now 999,998 don't. The risk has doubled, or even worse gone up by 200%! But it is still an infinitessimal risk! Beware of how statistics are presented. The risks of radiotherapy are still very small.

Marcia said...

Thanks for that reminder of why statistics can be misleading. I guess it would help if the actual amount of risk is included with such statements. But thanks for relieving my mind about going ahead with the radiation.