POINT: COUNTERPOINT – MACE and Its Relationship to IL 12/23 Inhibition
Ken Gordon, MD
Dr. Gordon presented a counterpoint to Dr. Leonardi’s perspective regarding the association of IL 12/23 inhibition with MACE events. Dr. Gordon feels that while there is not a definitive answer at this time regarding the association it is important to discuss both sides to the story.
When a dermatologist thinks about the data that has been presented, specifically on ustekinumab, one needs to consider whether IL 12/23 has a true effect on MACE events. There are two approaches to this: 1. Observe the world, here is the observation and it needs to be explained; 2. Utilize the Scientific Method.
In the context of IL 12/23 and looking at MACE events, there has been an inductive moment, and that is the paper that Dr. Leonardi referred to, i.e., the Meta-analysis regarding IL 12/23.
In reviewing the overall conclusions that were presented in the meta-analysis of the association of IL 12/23 inhibition and MACE events, Dr. Gordon concluded that at least in the case of ustekinumab, there may not really be an answer. It is important to remember that meta-analyses have two required aspects in order for them to be effective: 1. The trials have to be similar in structure; 2. The intervention that is used has to be similar. In combining the trials using ustekinumab with those using briakinumab one can see that despite their effect on IL 12/23 these drugs are actually quite different in the dosages used. For example, Dr. Leonardi pointed out that briakinumab had issues with infections and malignancies. Briakinumab has opportunistic infections; whereas, ustekinumab does not. Additionally, there are a large number of squamous cells carcinomas developed in patients treated with briakinumab. There is a change in squamous cell carcinoma to basal cell carcinoma ratio in briakinumab versus placebo. This was not seen with ustekinumab. Therefore, putting the trials using briakinumab and ustekinumab together in a meta-analysis is difficult.
Regarding other differences, looking at the long-term data and only patients who are tolerant of medication in the ustekinumab trials is unfair. Dr. Gordon would argue that if you look at data over time, the retention rate was over 80% over five years in the ustekinumab trials. He pointed out that the high percentage rate of retention is rare in clinical trials and he doesn’t feel that eliminating a large number of patients is an issue here.
When you look at 0.11 what does that mean?
Dr. Gordon feels that the P value being 0.11, as presented by Dr. Leonardi, is very significant. If the experiment were redone 100 times, one would expect to find a difference between the intervention and the placebo 89 times out of 100. That is quite significant.
However, we have a P value and a standard for statistical significance for a reason and that is because we need to be stringent about our conclusions and when they get too close together, it can be difficult to draw any final conclusions. It doesn’t say the magnitude of the problem, only that the two populations will be different.
With meta-analyses, there is another problem, and that is when trials are combined, you can have a trial that is the “driver”, i.e., one or two of the trials out of a larger group of trials will drive the results in one direction and everything else will be inconsistent with that; therefore, the validity of the meta-analysis can be called into question. In particular, when you combine drugs that are distinct and have a driver that is utilizing one of the medications, the validity can be questionable.
(Of note, Dr. Gordon is an author on the meta-analysis with Dr. Leonardi and feels that it is the best analysis that we can have)
A recent article in JAMA (August 2011) reported that there are two studies that are the drivers of all of the IL 12/23 data. The first of which is the briakinumab trial, the second of which is the ustekinumab Phase II trials in which the drug was given in a different fashion than it was given in Phase III. In fact, the numbers in the Phase III trials do not appear to have any statistical significance in the occurrence of MACE events. This makes it very difficult to draw conclusions on ustekinumab and MACE events. In fact, ustekinumab, at the doses given during the phase III, does not seem to statistically show effects at all—observationally, less so than anything else that has been seen.
Dr. Leonardi pointed out the initial increase in risk of MACE events with the initiation of treatment and then it levels off. However, other data presented shows that there are events (randomly) throughout the course of treatment. Regarding the p40 subunit serum levels, there may not be enough information to draw conclusions based on this time course and the multiple events that are seen.
What does Dr. Gordon do for patients on ustekinumab?
Do you warn patients for potential cardiovascular risk? Yes, patients should be aware of the potential risks.
Do you put patients on aspirin? In general, Dr. Gordon does not put people on aspirin; however, if the patient is at risk and should be on aspirin, then he recommends it. If a patient does not have risk factors, is under 40 and is hypertensive, he does not use aspirin.
This is a very difficult topic and many people have many differing opinions and feelings on it. The answer regarding ustekinumab is not fully understood. Regarding briakinumab, Dr. Gordon feels that there is a definite. Only with experience and time will clinicians have the answers regarding IL 12/23 and MACE events.