This is an analysis of Proteon's PRT-201 (vonapanitase) that was previously completed.
Introduction
This is a 26 kDa preparation of elastase enzyme for use at the time of arteriovenous graft (AVG) or fistula (AVF) creation, with the company concentrating initially on AVF. For patients undergoing hemodialysis, AVF, AVG or insertion of a temporary dialysis catheter via the internal jugular / subclavian vein / femoral vein are the choices for allowing vascular access. Of these, AVFs are the generally preferred method.
One of the major complications for AVFs is loss of patency, general short-hand for the reduction of blood flow via the access site. Loss of patency can be due to multiple factors, but most relevant to this drug candidate is adverse remodeling of the vessels involved in the AVF creation.
This figure demonstrates the general pathways of AVF dysfunction as they relate to PRT-201. Following formation of the AVF, vessel diametre of the outflow vein is of primary importance, but can be encroached by adverse remodeling. In essence, remodeling that reduces the diametre of the vein via neointima formation or negative remodeling all challenge AVF function and are causes for primary loss of patency. Proteon believes dropwise administration of PRT-201 to the outside of the vessel for 10 minutes during AVF formation will improve the function of the fistula and reduce primary patency loss.
Throughout multiple papers, Proteon believes PRT-201 will preserve AVF function by various mechanisms:
1) Degradation of elastase along the adventitia layer will increase compliance (1/stiffness) of the vessel, allowing greater lumen diametre immediately at the time of AVF creation, thereby allowing the vessel to achieve higher blood flows (versus placebo). These higher blood flows would presumably require more extensive neointima formation prior to occlusion, thereby allowing primary patency to be maintained for a longer duration.
Best surrogate: increased vein lumen diametre immediately after administration of PRT-201
2) Proteolysis of elastin fibres within the adventitia by elastase will create small elastin fragments that serve as known chemoattractants for smooth muscle cells and monocytes. As the elastin fragments are generated in the outside layers of the adventitia, they will serve as decoys prompting smooth muscle cells to move to the adventitial layer to promote positive outward remodeling rather than towards the lumen to increase neointima formation.
Best surrogate: this mechanism reflects on the primary patency endpoint, but vein lumen diametre at 6 weeks and 3 months or time to hemodynamically significant lumen stenosis (HSS) would be representative.
In effect, these two mechanisms should lead to increased vessel dilation at time of surgery, increased blood flow and decreased HSS. Achievement of all three would rationally allow the treatment to have a chance at increasing primary patency duration, which is the endpoint of the current PATENCY-1 trial.
Synopsis
Proteon has completed a phase I trial (Peden et al, 2013) and a randomized phase II trial (Hye et al, 2014) in AVF. Although the primary endpoint of primary patency was not met in the phase II trial, a subgroup was identified wherein patients receiving radiocephalic fistula (RCF) benefitted more from PRT-201 than patients receiving brachiocephalic fistula (BCF). The follow on phase III PATENCY-1 trial is therefore a double-blind, placebo controlled trial examining the effect of 30 ug PRT-201 on the primary patency of RCF AVF.
This is a classic example of a biotech running a phase III program from a post-hoc subgroup obtained from their phase II trial. There are multiple reasons to be skeptical about the outcome of the PATENCY-1 trial. To that end, we can focus on the proposed mechanism of action and determine if there is indication of PRT-201 activity in the phase I and II trials completed for this patient population.
Phase I Trial
Red Flags
1. In the phase I trial, Peden et al showed an inverse dose-response relationship wherein the primary patency trend favoured the low dose PRT-201 arm (3.3, 10 and 33 ug) whereas the medium (100, 330 and 1000 ug) and high (3, 6, 9 mg) dose arms regressed and where indistinguishable from placebo.
There is no biological reason to favour low dose over high dose, as the high dose has up to 270x more active enzyme. That should translate to significant vessel dilation at the time of surgery and significant increase in blood flow. However, the paper expressly notes that *none* of the participants showed an increase in outflow diametre of 25%. Further, the percent change in vein diametre pre- and post-surgery also showed an inverse dose-response, with high dose (3.2%) being lower than low (6.1%) and medium dose (5.7%). This significantly raises questions as to whether the 10 minute treatment is doing anything at all.
2. Although the vein diametre increase in the low and medium dose groups where higher than that for placebo (2.1%), this finding likely reflects noise. From personal experience, digestion of veins and arteries by enzyme preparations is strictly dose and time related. For example, when vessels are taken into culture and enzymatically digested to release the smooth muscle cell population, the procedure is necessarily time and activity related. Digestion of increasing amounts of elastin would simply compromise the structural integrity of the adventitial layer, increasing external load on the smooth muscle cells, which would in turn necessarily increase vessel diametre. That this does not happen in the measurements pre- and post-surgery suggests the mechanism is not progressing as intended.
3. Further, when blood flow was measured immediately pre- and post-treatment, the percent change in blood flow was relatively superior for the placebo group (41%) versus the medium (17%) and high doses (19%), and very similar to the low dose group (43%). This also suggests, along with the vein diametre data, that the enzymatic digestions at the adventitial layer is not having the intended effect.
4. Finally, the investigators report the time to HSS, which was 59 days in the placebo group versus 209, 49 and 43 days in the low, medium and high dose groups, respectively. Again, this suggests an anomalous outcome. Although there is likely to be an optimal treatment time to produce chemoattractant elastin fragments, that would predict an optimal dose that would produce peak results followed by the effect waning off as dose was increased. In this respect, the medium dose group would have been expected to be in between the results of the low dose and the high dose group.
5. Furthermore, since each dose group has 3 dose levels, there is no indication which of the 3.3, 10 or 30 ug doses in the low dose group most contributed to the increase in HSS. If there was a clear indication for 30 ug (the Phase III dose), it would have been noted as it would strengthen the method of action and rationale for the phase II.
In sum, this results from the phase I trial support safety of PRT-201 but provide no meaningful efficacy signal.
Phase II Trial
Red Flags
1. This randomized, placebo controlled trial failed to meet the primary endpoint of primary patency for either the 10 or 30 ug dose group versus placebo. Moreover, the endpoint was missed despite the investigators using a per-protocol (PP) analysis rather than an intent-to-treat (ITT) analysis. Of the 169 participants randomized, data from 151 were included in the analysis. Analyses that are PP favour the sponsor, and the trial’s inability to hit on this lessen hope for a subsequent phase III trial wherein ITT analyses will be necessary.
2. The investigators begin data-mining in order to find a benefit, and determine that patients receiving RCF AVF benefit from treatment whereas BCF AVFs do no benefit at all. This is a dubious finding, mostly because PRT-201 is a pure enzyme preparation and has little reason to act differentially in the adventitia of veins involved in the BCF procedure versus the RCF procedure.
Note from Phase I: When the investigators examined the data in the phase I trial via a Cox proportional hazard model, only treatment with low dose, white race and age <65 years were associated with decreased risk of patency loss. If the RCF finding in the phase II was real and robust, it would have also appeared in the exploratory analyses of the phase I trial. It did not.
3. The RCF subgroup analysis is very likely unbalanced between arms. For the trial as a whole, the investigators note that multiple prognostic indicators including exposed vein length, shorter arteriotomy length, white race, predialysis status were all associated with decreased primary patency loss. There is no indication that the investigators examined the placebo, 10 ug and 30 ug groups from the RCF subgroups to determine if they were balanced in this regard. But given the low numbers (n < 25 in each group), it is very unlikely that the groups were balanced.
4. For the trial as a whole, HSS at either 6 weeks or 3 months was not significantly impacted. At 6 weeks, HSS was 51, 30 and 39% for placebo, low and high dose, and this changed to 40, 41 and 35% at 3 months. Only the comparison of HSS for placebo versus 10 ug at 6 weeks was significant; this was not maintained at 3 months, likely suggesting it was a chance finding. This is especially likely given that HSS in the placebo group appeared to improve over time; this may reflect a change in the number of evaluable participants, but nonetheless supports a lack of differentiation between placebo and treatment groups.
Interestingly, the HSS data are *not* broken down further into RCF versus BCF. A reliable rule for interpreting peer reviewed papers (and biotech company PR) is to assume that analyses not presented were not statistically significant. In this regard, given the authors’ proposal that the RCF versus BCF analysis is meaningful for primary patency, it would have been consistent if the HSS surrogate was significant in RCF versus the BCF group.
5. Although vein lumen diametre and flow rate data are provided by the investigators, none of the comparisons across groups or over time points are noted to be statistically significant, nor do they appear to adhere to a meaningfully different trend in the treatment versus placebo groups. Once again, this works against one of their proposed mechanisms of action, and provides little evidence that PRT-201 has short term or long term impact on remodeling of the outflow veins or flow characteristics.
Conclusion
This appears to be a classic case of subgroup data-mining in phase II leading to a phase III program. There is little to no internal consistency between the phase I and II data. Further, there is no mechanistic consistency between their proposed biological mechanism of action and the surrogate measurements that would reflect the action of the drug. Vein diametre, blood flow and HSS data are not different between placebo and PRT-201 dose groups, providing little confidence that primary patency would be maintained for longer durations. Further, the separation of RCF versus BCF subgroups has little biological basis with respect to PRT-201’s mechanism of action, and the purported efficacy in the RCF subgroup is likely due to imbalances in patient population rather than a mechanism driven discovery.
Miscellaneous Technical Aspects
- Preparation of the enzyme is relatively trivial and will not be a limiting factor. Enzymes are supplied as powders in a vial and need to be reconstituted by the team at the time of surgery. Technically this poses a risk for some loss of enzyme activity, as enzymes tend to lose activity over time if not properly maintained.
- Administration of protein always raises the risk of antibody generation, or pre-existing antibody titre. However, there is no retreatment in this protocol, and evidence of very low / non-existant antibody titre is provided.