The ShortHER trial is an investigator driven trial which was founded by the Italian Drug Agency with the aim to explore if a shorter duration of trastuzumab administration for the adjuvant treatment of HER2 positive breast cancer patients could produce comparable efficacy with lower toxicities and certainly lower costs in comparison to the standard one year duration of trastuzumab.
More than 1250 patients were entered into the trial from 82 Italian hospitals. These patients had node negative disease in about 53% of the cases; 47% of the patients had node positivity and among these there were about 15% of patients with four or more positive lymph nodes. The vast majority of patients had hormone receptor positive disease, 68% of the entire patient population. The two treatment arms were a standard arm consisting of eight courses of chemotherapy, four courses of an anthracycline cyclophosphamide followed by four courses of docetaxel plus trastuzumab starting concomitantly with the first course of docetaxel and given three weekly for eighteen doses up to one year. The experimental arm consisted of three courses of docetaxel followed by three courses of FEC with trastuzumab starting at a weekly dose with the first administration of docetaxel and given for nine weekly doses only. So one year trastuzumab versus nine weeks or more or less two months of trastuzumab, so a big difference in terms of the total amount of trastuzumab and also differences in the total amount of chemotherapy because there were eight courses in the long treatment arm, six in the short. There were also lower doses of epirubicin in the short in comparison to the long treatment arm. Then, of course, all the patients with hormone receptor positive disease received endocrine therapy for five years.
The primary endpoint of the study was disease free survival and overall survival was the co-primary endpoint. This is a non-inferiority trial and we planned two types of analysis to demonstrate non-inferiority: one is the classic frequentist analysis but then we also planned the variation analysis. Differently from the frequentist analysis, the variation analysis is based on the observed data and allows us to estimate on the basis of observed data what is the probability that the short or experimental treatment is not inferior to the long one. The statistical assumption was that the short had to be considered not inferior if the hazard ratio in comparison to long was less than 1.29.
At the median follow-up of 5.2 years we observed 189 events, 89 in the long treatment arm, 100 in the short arm. The hazard ratio was 1.15, so below the limits of non-inferiority, however the upper margin of the 90% confidence limit crosses the 1.29 boundary for non-inferiority. So from the frequentist analysis we cannot claim non-inferiority but by doing the pre-planned Bayesian analysis the probability that the short treatment is not inferior to the long one is 78%.
We have moreover done a pre-planned subgroup analysis and we found that there were two subgroups, patients with stage 3 disease or patients with four or more positive lymph nodes, and these patients represent around 15% of the entire patient population, had a significant difference in disease free in favour of a long treatment arm. Essentially the small difference between short and long is completely attributable to the 15% of the patients with four or more lymph nodes or stage 3 disease. Again, in 85% of the patients there was no difference at all.
Another important outcome of the study was the cardiac safety which was a secondary endpoint of the study. In terms of cardiac safety we have seen a significant difference in the decline of the [?? 5:15] fraction in favour of the short treatment arm compared to the long one with a p-value of 0.02. Also the numbers of cardiac events of grade 2 or more and time to incidence of cardiac event were in favour of a short treatment arm with a hazard ratio of 0.3 and a p-value of less than 0.0001.
So our conclusions are that the ShortHER trial has shown that nine weeks of trastuzumab versus one year of trastuzumab produces a hazard ratio of 1.15 which is within the margin of non-inferiority. However, because the upper limit of non-inferiority crosses the 1.29 boundaries for non-inferiority, on the basis of the frequentist analysis we cannot claim non-inferiority. However again, the Bayesian analysis allows us to say that the probability that the short is not inferior to long is 78%. Moreover, subgroup analysis suggests that the difference between the two treatment arms is null for patients with node negative or up to three positive lymph nodes while there is a benefit in favour of a long treatment for those patients with four or more positive lymph nodes. Short treatment is significantly more tolerable from the cardiac point of view with a significant reduction in the number and time of cardiac events. We do acknowledge that one year trastuzumab should still be considered the standard treatment, however, the data from the ShortHER trial provides useful information to clinicians because it can reassure the clinicians that if a patient has to stop trastuzumab because of the occurrence of a cardiac event it can be done without compromising the efficacy of the treatment. Then we can also now start the treatment with trastuzumab even to those patients at risk of cardiac toxicity, patients with pre-existing cardiac comorbidities. Finally, and this is not an eligible point, it might expand the access to trastuzumab in countries where because of financial problems the cost of the treatment cannot be afforded by most of the patients.
Could it not also be read that for patients with less than four positive nodes a shorter course would be an option, if you said that that was what was [?? 8:18]?
Certainly. As I said, we have to recognise because this is the approved dose of trastuzumab this is the standard. However, if I have a patient with a very low risk of relapse and nowadays we treat patients with one year of trastuzumab with very small tumours, with node negative disease, for these patients we might consider the use of only nine weeks trastuzumab which is equally effective and less toxic.
Are there any plans for a further trial of, say, twelve weeks of trastuzumab to find what is the threshold to reaching non-inferiority?
As you can imagine, it’s not easy to run these kinds of trials because you have to get funded by public agencies, there is no interest from pharma companies to support this kind of trial. Moreover, non-inferiority trials are not so popular so it would be very difficult to repeat another study to see if you can by prolonging by maybe a few more weeks or a few more months the duration of the treatment you can really equalise the efficacy of the therapies, one year versus something less. What we can say is that there are other trials comparing different durations of trastuzumab administration. There has been a French trial which has already been published comparing six months versus one year. There will be another British trial which has completed recruitment and is not yet being presented again comparing six months versus one year. Another Finnish trial has compared nine weeks versus one year. We are planning altogether to have an individual patient meta-analysis of all these trials in order to reinforce the concept that there are groups of patients who deserve a shorter treatment with less toxicity and certainly less cost.
I wonder if there might be funding available for, say, nine or ten weeks of a trastuzumab biosimilar?
Yes, that is also a possibility of course. We have now the biosimilars coming out, they have very similar efficacy in comparison to the patent drug. That might be a good idea to see whether a company producing biosimilars might be interested to explore a shorter duration which, of course, would be far less expensive.