A quick look at herceptin for breast cancer

Bandolier has tended to avoid cancer studies in looking at evidence because it is a different world compared with most clinical situations. Cancer trials may be randomised, but they never compare a new treatment with no treatment; typically they compare the usual best care with usual best care plus the new treatment. Progress is measured in increments rather than great leaps forward.

Then there are the outcomes, of which there are many. Not dying comes to mind first of all. But there is also keeping the cancer at bay. Put these together and you have a new outcome, that of disease-free survival. And there are more like that, some of real importance to patients, others of great importance to oncologists.

Not unimportant is the duration of trials. This is cancer, after all, and we want to know what happens not after months, but years. So we often learn first about results after one year, then two, then five, and perhaps 10 at some later stage. That is good, because it maximises knowledge as soon as it is available, but initially we can only extrapolate about whether benefit at one year is maintained over later years, or increases, or evaporates.

And finally there is the issue of combinability. Cancer of even one organ is usually of many different types, and is treated differently at different stages of the disease. If we are lucky, we have a trial. We almost never have the luxury of meta-analysis.

And yet oncologists continue to improve cancer treatments, achieving not just longer survival or disease-free survival, but better quality of life. Bandolier is reminded of the friend with serious liver secondaries who boasted of living for four years with six months to live.

The point is that judging cancer trials and evidence isn't easy. It requires time, hard work, and not a little insight. Bandolier has seen involved professionals ecstatic over trial results that leave purchasers of care wondering what the fuss is about.

In the full and certain knowledge that whatever we write, someone will take exception to some part of it, here is a quick look at trastuzumab (herceptin) for breast cancer, as readers have asked. Only published randomised trials in breast cancer are examined.

Trastuzumab is a monoclonal antibody to the extracellular domain of one of several receptor tyrosine kinases known as HER-2 that mediate growth, differentiation, and survival of cells. Over expression or amplification of the gene occurs in about 1 in 5 breast cancers, and is associated with aggressive progression. The idea is that blocking it should be a useful treatment for this form of breast cancer.

Trials of trastuzumab in FER-2 positive breast cancer have been reported both soon after initial diagnosis and treatment, and when the cancer has metastasised. These two circumstances will be examined separately. An important adverse event has been the development of congestive heart failure, which can be categorised as changes in cardiac function, symptomatic heart failure, or severe heart failure.

Three trials have been reported recently in two publications (Table 1), using treatment with different dose regimens of trastuzumab over one year, and with events reported at a median of one and two years. Although all the information for some trial arms has yet to be published, we have results on 5,060 women given trastuzumab as part of their treatment, and 2,372 women not given it.

The primary outcome was disease free survival, defined as absence of recurrence, a second primary event, or death without recurrence. Combining the information, one of these events occurred in 14% of women not given trastuzumab, compared with 8% of those who did receive it for a year. The relative risk was about 0.55 (95% confidence interval 0.5 to 0.6), and the number needed to treat 15 (13 to 19).

This combining of one and two year outcomes probably understates the benefit, though. In women who were observed for longer periods (up to three or four years), the benefits appeared to increase over time, with less disease recurrence or death compared with control. Based on very limited data and events, equivalent three and four year NNTs could be around 10 and 5. The benefits seemed to be similar in all subgroups or stratifications of women.

Set against the benefits was an increased rate of congestive heart failure. Symptomatic disease occurred in 2.4% of women on trastuzumab and 0.1% of controls, a relative risk of 16, and number needed to harm of about 45 (36 to 59). Only a small proportion of this was severe, though there were some cardiac deaths.

One other small trial used chemotherapy plus trastuzumab for six months before surgery for breast cancer, and found a complete clinical response in 21/23 women given the combination.

For metastatic breast cancer there were four trials, one of which compared two regimens of trastuzumab in a kinetic plus clinical study. Typically, the trials involved women who had been treated initially with surgery, adjuvant chemotherapy, radiotherapy and hormones, but who had then developed a secondary metastasis about two years later. The trials compared one form of treatment (usually chemotherapy) with the same treatment plus trastuzumab, with treatment continuing until the disease progressed.

Complete response (disappearance of the secondary) occurred rarely, in fewer than 10% of women. Complete or partial response (shrinkage of tumour by a large amount) occurred in about 35%. Most women had stable disease, where the tumour neither grew nor shrank. Overall, the addition of trastuzumab increased the time to disease progression by three or four months, and overall survival by three to six months.

There were many adverse events associated with use of trastuzumab, both in early and metastatic breast cancer. Table 1 largely concentrates adverse event attention on symptomatic congestive heart failure, occurring in 2-3%.

The pattern of reporting for trastuzumab is not untypical of any new cancer treatment. Because cancer treatments are often toxic, they tend to be tried first in advanced breast cancer, then less advanced metastatic disease, and then only later as first line treatments in early breast cancer. That was the pattern with aromatase inhibitors, for instance.

In trials in cancer, time is the key. What we want to know about, namely improvements in survival, and especially disease free survival, can only be fully measured when patients have been followed up for many years. A trial with a five year follow up might take 10 years between design and reporting, not least because recruiting sufficient numbers itself takes time. And the better we get, the longer it takes to show big improvements.

So if there is nothing different between the trastuzumab reporting and what we have seen before, what about the results? In terms of the size of the benefits, they look quite good. Demonstrating an additional three months before progression in women with metastatic cancer is good. The results in early breast cancer, with a 50% reduction in bad events over one or two years are also rather good.

Lest anyone wonders at the disparity between these results, looked at with a cold and fishy eye or headlines in newspapers, it all depends on your point of view. In oncology terms these are very good results, and if the medicine were cheap everyone would be cheering. Because it is expensive, it presents real problems.

In metastatic breast cancer, a cost effectiveness analysis [1] estimates the total cost of treatment to be about ¤44,000 (£29,000). The cost per life year saved is between ¤63,000 and ¤162,000 (£42,000 and £108,000), calculated from survival improvements from trials. Reducing these figures would require large reductions in the cost of the drug, or larger improvements in survival. Costs of adverse events were not included. Costs in early breast cancer would be only a guess, and cost effectiveness is bound to be much less than TNF-alpha antibodies in early rheumatoid arthritis, for instance.

Of course, only 1 in 5 women with breast cancer would be HER-2 positive, and not all of those would be eligible for treatment, but even so this would be large burden, and use of trastuzumab would probably be bigger than what is usually regarded as affordable.

So great science produces good results that we cannot afford. We need to think outside the box. The patent-protected life of a medicine is 20 years or so, with much of that time taken up by laboratory and clinical research before any sales can take place. The result is a short time to recoup development costs, and the high price we pay for drugs.

Perhaps someone ought to consider whether changing patent laws might smooth the way to our ability to develop and use new therapies. If patent life were shortened, development costs would not be recouped, and research and development would end. Longer patent life may be better. Our clever academic economists should so some sums.

donate to Bandolier!

• Knowledge Library
• Healthy Living Zone
• Extended Essays
• Learning Zone
• EBM Glossary

advanced search