On Lack of Dose Response and Study Information: A Double Whammy Against Investors
Pharma’s Cutting Edge
Vol. 3 Number 11 - November 2005
On Lack of Dose Response and Study Information: A Double Whammy Against Investors
Putting aside just how truly horribly this press release is written, I thought it might be worthwhile to discuss one particular morsel amidst this mess. That is: what does it mean when a company reports positive efficacy results from just the low-dose arm of an ascending dose, parallel-group Phase 2 study?
Let’s first consider the obvious causes of a lack of dose-response. Were the treatment groups well-matched at baseline? We can’t tell from the abstract; perhaps more information was presented at the meeting (the American College of Rheumatology). Of course, most investors don’t attend these meetings, nor do most investors have access to well-written, detailed, timely reports of such meetings. This is why all presentations involving investigational therapies under development by public companies should be made available as soon as possible after they have been presented and should be linked directly to press releases. Why the SEC does not demand as much is beyond me.
We do know from the attached press release that background therapy was heterogeneous, meaning that the possibility for important baseline mismatches among treatment groups is relatively high. We don’t know whether an attempt was made to prevent such differences (e.g. by stratifying randomization on 1 or 2 pre-defined factors) or to adjust for them statistically if they occurred (e.g. ANCOVA). This is key information, folks. Without it the results are nearly meaningless.
Another culprit could be the dose range studied. We’re not provided with the pharmacokinetic properties of this drug nor the relationship between pharmacokinetics and pharmacodynamics (e.g. decreases in immunoglobulins) in this press release, so it would be necessary to research such issues elsewhere. But let’s give HGS scientists the benefit of the doubt and assume that they chose a dose range that should, based on well-defined PK/PD relationhips, have covered the “steep part” of the dose-response curve. Could they have been fooled by Phase 1 data? Absolutely. It happens all the time, especially in diseases like RA, where the primary endpoint is subjective, and Phase 1 studies rely mostly on objective PD measures (also called biomarkers) to determine the PK/PD relationships. Clinical scientists make their best estimates of expected dose-response based on all available data–human, animal and in vitro–often using simulations for assistance, but it sometimes just isn’t good enough. So, it’s possible that HGS just got unlucky and ended up at the top of the dose-response at the lowest dose studied. They’re unlucky, because they still have an obligation to demonstrate a no-effect or minimal-effect dose and a submaximal dose to minimize the potential risk of long-term serious complications of therapy. That means either doing another Phase 2 dose-response study at lower doses or carrying at least two, preferably three, active treatment doses into a Phase 3 parallel-group study. The latter can be accomplished at modest excess cost (but longer duration) over a single dose study by allowing for dose-titration for non-responses to lower doses.
What about the possibilty of a strange dose-response curve, like a genuinely flat curve or a shallow, inverted “U”? I have read about such phenomena, but practically speaking, it rarely happens. Dose responses are usually linear. I don’t know why, but it’s a great thesis project. What does seem to occur with some frequency is a Type II error (failure to demonstrate a true dose response, in this case) due to study design, conduct or outcome factors not directly related to the primary endpoint. For instance, suppose that a drug causes a dose-related side effect that leads either to non-compliance with therapy (i.e. missed doses) or early study discontinuation. Subjects who receive lower doses would be less likely to experience the adverse effect and thus more likely to derive benefit from the therapy. The dose-response would be either flat or inverted, depending on factors including the severity of the adverse effect, the robustness of drug benefit, and the slope of the true dose-response curve. The type of analysis chosen is also relevant. We’re not told whether the primary endpoint was analyzed using interpolation for missing data (e.g. last observation carried forward). In Phase 2, study completer data (data only from subjects who completed all visits) are often used for the primary analysis. Use of study completer data would tend to reduce an apparent treatment effect of doses associated with early study discontinuations.
Note that just being told how many subjects contributed to the primary endpoint inference (i.e. the “N” of each group) would have helped us to deduce whether there were tolerability or safety issues associated with dose. It wouldn’t have been all we needed to determine which of the above factors contributed most to the lack of dose-response, but it would have helped. Generally speaking, I believe that investors should seek maximal transparency from such press releases, which means demanding the type of information I’ve discussed briefly here. Public companies should not be allowed to use the power of widely available information to tout the promise of investigational therapies without also providing a readily available means to check their claims.
