Mammography – The Evolving Story of a Cancer Diagnostic Tool
The use of low-dose radiation to detect occult breast malignancies can be traced to work done at the MD Anderson Cancer Center in the 1950s. Early published studies conducted by the “Egan technique,” correctly identified the majority of palpable cancers subsequently proven malignant at the time of surgery. As a diagnostic tool, mammography is an effective means of confirming the presence, and defining the extent, of breast pathology in woman at high risk for cancer, or who note a palpable lump. No one is arguing the diagnostic use of this technique. Where the controversy has arisen over the last years is the use of mammography as a screening technique.
To clarify the use of terminology, screening techniques are applied to the general population to identify unrecognized disease. The popularity of mammography as a screening technique led to the recommendation that every woman over 40 should have an annual mammogram. The problem with screening techniques is that they apply a diagnostic tool to a population at low risk. This burdens the technology with numerous false positives, engendering costs, risks, and toxicity for those who undergo unnecessary biopsies and surgery. The entire discussion came into sharper focus in the past week with the publication of a large Canadian study that examined the impact of mammographic screening over a 25-year follow up in women ages 40-59.
In this study, launched in 1980, more than 89,000 women were divided into two groups. One group underwent routine physical examination and the second group had routine physical examination combined with mammogram. There were 3,250 diagnoses of cancer in the 44,925 women who underwent mammography and 3,133 cancers diagnosed in the 44,910 women who underwent physical examination alone. Five hundred patients in the mammography group and 505 women in the control group died of their disease. While women who had mammograms were more likely to be diagnosed with breast cancer, this did not have an impact on their risk of dying from the disease. Furthermore, 22 percent of women with positive mammograms did not have cancer at definitive workup. The conclusion of paper and the accompanying editorial by Mette Kalager from Oslo, Norway, was that ”the rationale for screening by mammography needed to be urgently reassessed.”
What are the shortcomings of the study? Mammographers have claimed that the equipment used was suboptimal, leading to less sensitive detection that might have occurred with modern, high-quality digital equipment. There was also no group over 60. It is also theoretically possible that some patients obtained mammography after concluding the study, or had mammograms done during the study, contaminating the final results. Nonetheless, this is a high quality randomized study in a large population that fails to provide an impact upon survival for a widely used technique.
Prior meta-analyses conducted between the 1960s and 1980s revealed a reduction in deaths in breast cancer between 15 percent and 25 percent in the population of women age 50 to 69. Explanations for the disparity between the current study and those older studies may include the relative lack of sophistication of the population during the 1960s through 1980s, who might fail to evaluate a breast lump, thus, earlier detection would have a significant impact on those not responding to even physical evidence of disease.
A second confounding variable is the broad use of Tamoxifen, which has so profoundly influenced the natural history of breast cancer, that the earlier detection of breast cancer may be less important than the potent efficacy of anti-hormonal agents. This is an interesting wrinkle in the story, as it is contrary to most contemporary thinking that holds that early detection, not treatment is the principal influence upon better outcomes today.
So where does this study leave us? There are several points that must be considered. The first is that mammography is a test not a treatment. Tests perform according to their performance characteristics, described as “sensitivity and specificity.” Within this framework mammograms are sensitive and specific enough to provide immense value ….in the right patient population, e.g. those at some risk for the disease in question. How you define that “risk” is the rub.
Mammograms identify the disease; they do not influence its biology. While some may demand that more sensitive equipment for the detection of disease be implemented, a different principle may underlie the findings. This would be that cancer, at virtually any stage of diagnosis, is a systemic disease with its own trajectory. Under this scenario, mammograms in an unselected population provide little more than a lead-time bias. This term is applied when a test identifies an event earlier than it might have been found, but has no impact on the ultimate outcome. Lead-time bias is a common phenomenon in screening techniques and has been the rallying cry for those who argue against PSA screening for men. Once again, the number of patients diagnosed versus the number of patients requiring intervention is the overarching dilemma.
While we seek to decipher the genetic basis of cancer using increasingly sophisticated genomic techniques, we recognize that cancer is common and that a substantial percentage of patients may not die of their disease. Cancer results from stresses that force cells to either die or seek novel mechanisms to survive. Deprived of estrogen, testosterone, nutrients, oxygen or growth factors, cells within the aging human body discover novel ways to stay alive, albeit to the detriment of the organism as a whole. However humbling, it can be argued, that it is pathways that aberrant cells pursue that guides the trajectory of the disease, largely independent of our roles as diagnosticians and treating physicians.