Developments in breast cancer screening

The future of breast cancer screening will go in one of two directions," says Jean-Charles Piguet, founder of the Clinique des Grangettes in Geneva, Switzerland. "Either digital mammography will become sensitive enough to pick up the cancers we miss at the moment, or - and this is where I'd put my money - we'll be able to use biological tests to divide the population into those who're likely to develop breast cancer and those who're not, then screen these two groups accordingly."

Although he is usually to be found at the forefront of screening developments, Piguet is not the first person to consider the possibility of genetically testing for breast cancer. The New York Breast Cancer Study Group has found that women with genetic faults BRCA1 and BRCA2 have an 85% chance of developing breast malignancies during their lifetime. Gene faults TP53 and PTEN have also been identified as significantly increasing risk, along with ten lesser-known gene types, six of which are commonly found in the population.

Yet despite the apparent relevance of genetic testing, it is still a long way from being implemented on the same scale as mammographic screening. In Europe, individuals must be able to demonstrate that they are likely to develop breast cancer through their family history, and a living relative currently affected by the disease must also consent to being tested first.

There is also the question of how effective genetic screening would be in practice: only 5-10% of breast cancers are thought to be caused by a strong hereditary predisposition. The potential level of uptake is also debatable: a Cochrane review spread over 40 peer-reviewed primary clinical studies published between 1990 and May 2002 by the US National Institute of Cancer found that, on average, public interest in undergoing the tests was only 66%. There is also the labyrinth of psychological issues, potential costs and questions relating to how effective genetic tests would be in accurately predicting cancers, all of which would have to be thoroughly considered before any such programme could be embarked upon.

The key to successful screening

For now, then, the breast screening debate is firmly focused on the development of existing technologies. Of these, Piguet is adamant that digital mammography is the best device for the job.

"The problem with digital is that it only has a sensitivity rate of 75%," he says. "But it is the only one that we can apply to a healthy population, where 99% of the women screened are healthy."

For Piguet, achieving the right balance between accuracy and the level of X-ray radiation patients are exposed to is the key to a successful screening device. It is for this reason that he is particularly excited to begin testing one of the latest developments in digital mammography, the Philips MicroDose SI, a full-field digital mammography system (FFDM) that uses non-invasive spectral imaging.

"Digital mammography usually entails some kind of trade-off," he explains. "Hologic's Lorad Selenia produced good-quality images, but entailed a relatively high dose of X-ray. GE's Senographe 2000D, on the other hand, had a low X-ray dosage but also a low resolution - only 100 microns.

"The Philips is, I think, the best so far. It has two energy X-rays, one low and one high, which allows you to get a high-quality image of all breasts, even ones with a BI-RADS [Breast Imaging Reporting and Data System] score of 3 or 4, while at the same time keeping the radiation dose as low as possible."

CAD: the pros and cons

Piguet's opinion on such matters carries weight: having advised GE during its development of the world's first digital mammogram, he is extremely well respected within the radiology community. Some, however, may find it more difficult to agree with his views on computer-aided detection (CAD); although the transference from film to digital has significantly improved the speed with which radiologists can perform CAD analysis, the technique is still criticised for its tendency to generate false positives.

"It's true that CAD gives you a lot of false positives when you are looking for masses," says Piguet. "But the fact of the matter is that if you're an experienced radiologist, you're easily going to see if it's something important or not.

"Where CAD is really useful is in picking up microcalcification. It has a very high accuracy rate for that, and, to be honest, I would say that it doesn't have any false positives; it's not the job of CAD to decide whether or not something is malignant. The same goes for other imaging technologies - ultrasound, mammography, MRI - what I expect from them as a radiologist is a sensitivity to abnormality; they are not supposed to be pathological. Deciding whether or not something is dangerous is down to my eyes, not the device."

The actual usage levels of CAD vary between Europe and the US, largely a result of the differing financial regulations. In Europe, radiologists are not able to charge extra for the service, and so pay comparatively high licensing fees.

"It's a shame that some European radiologists are missing out on CAD for financial reasons, especially when you consider how much money they spend on other things like IT services," Piguet comments. "They can also perform far more mammograms with digital than they could ten or 20 years ago; you'd have thought that would compensate for the extra spending on CAD."

DBT: the front line of breast cancer screening?

Digital breast tomosynthesis (DBT) is another technique that has generated a great deal of attention lately, with its supporters claiming that it will soon replace mammography as the front line of breast cancer screening. Taking a series of images, from a variety of angles, which are then processed to produce tomographic sections through the breast, DBT allows for a much higher sensitivity than conventional digital mammography in patients with dense breasts. This is particularly important because these women are four to five times more likely to develop malignancies.

The downside, however, is that, along with presenting radiologists with a time-consuming 2GB of data to read, DBT significantly increases the X-ray dosage patients are exposed to - a result of the need to interpret the tomographic sections in the context of an additional 2D FFDM.

''Tomosynthesis can give you beautiful images, but at the end of the day, it doesn't help you," says Piguet. ''You cannot use it for screening because of the extra radiation, and it's silly to have it if you're only going to use it for special exams like spot views once or twice a day."

The technique should not be written off yet, though. Hologic has recently developed an algorithm that creates a synthesised 2D mammogram from the data acquired during tomosynthesis, eliminating the need for an additional 2D FFDM and thus cutting the radiation to which the patient is exposed.

"It is an interesting development that will probably be very useful in diminishing false positives," says Piguet. "But it will be five to ten years before the technology appears on the market; the European Reference Organisation [for Quality Assured Breast Screening and Diagnostic Services] is only now in the process of devising the necessary checks."

Ultrasound as a screening process

The other major technological debate is whether or not ultrasound should be used in conjunction with mammography. At first glance, the move makes sense: ultrasound is far more effective at picking up potentially malignant masses, and also functions well with dense breasts.

But France is currently the only country to include ultrasound as part of its screening process, and even there, it is only to be used if something suspicious is found on the mammogram, during the clinical examination, or if the patient has a BI-RADS score of 3-4.

"If you were to combine ultrasound and mammography, you would increase your sensitivity to almost 98% - much higher than the 75% of mammography alone," Piguet explains. "But ultrasound is not viable as a screening process; it is too time-consuming. A doctor will spend five to ten minutes performing one, even if it's small; whereas to read a mammogram, if you're used to it, only takes 30 seconds.

"A possible way round this is to have automated ultrasound that stores images you can look at in the second step. I trialled the Siemens ACUSON S2000 automated breast volume scanner (ABVS) for this. As a machine, it worked well, but it was far too time-consuming to read all the data it produced - 400 slices for each breast. It certainly doesn't replace human hands on the ultrasound."

There is another problem with ultrasound: the capacity for error. While digital mammograph images can be easily compared with previous images or sent to other radiologists for a second opinion, the same cannot be said of sonography. Consequently, if the radiologist is inexperienced, there could be some weaknesses in the diagnosis.

It would seem that mammography is still the only imaging technique that can tick all the boxes as a screening device: low radiation exposure, high sensitivity, time efficiency and financial viability. As Piguet says, "It is undoubtedly far from perfect. But it is the best thing that we have right now."