We are short Prothena Corporation PLC (PRTA). The publicly-available data, in our opinion, does not show that NEOD001 is efficacious. Our impression appears to run contrary to that of the sell-side. We believe that perceived positive responses to the drug are quite possibly – if not likely – due to previous plasma cell directed (“PCD”) therapy, and also to the manner in which some early trial data has been presented. Further, NEOD001’s proposed mechanism of action is not proven.
sider linking to these articles
AL amyloidosis is a horrible disease. We would be happy to be wrong about our reading of the data. However, even if NEOD001 were to be approved, we see the Street’s peak annual sales estimates generally of over $1.5 billion as unrealistic. The patient population is small, and we believe most payers are unlikely to reimburse for this drug if it were priced at a level sufficient to put the peak annual sales estimates within reach.
We note that insiders hold virtually no shares, immediately sell substantially all stock received through option exercises; and, that Prothena Corporation raised an additional $155.3 million gross in an equity offering in March 2017. A recent investor conversation with Prothena Corporation’s CFO led him to take a tone described as defensive when asked pointed questions about the trial data. These observations lead us to believe that management might share our concern about the efficacy of NEOD001. We also find it noteworthy that PRTA presently plans not to release interim data from its Phase 3.
We Conclude Publicly Available Trial Data Does Not Support that NEOD001 is Efficacious
The publicly-available data from the Phase 1/2 trial conducted on NEOD001 do not, in our view, support a conclusion that NEOD001 is efficacious. The main issue we see is that patients’ improvements in biomarker levels are quite possibly – if not likely – to have been caused by previous plasma cell directed (“PCD”) treatments. In contrast to the conclusions Liedtke et al presented at American Society of Hematology (“ASH”) 2016, we see the timing of patients’ organ responses as consistent with responses from PCD treatments, which often consist of chemotherapeutic agents.3 We see in the data a lack of correlation between cardiac responses and the depths of hematologic responses, which casts doubt on NEOD001’s efficacy. The NEOD001 “Best Response Analysis” is misleading in our view as to efficacy, and we further believe it has led to undue optimism about the drug.
Between 2013 and 2016, Prothena Corporation conducted a Phase 1/2 trial on 69 patients with AL Amyloidosis. The first 27 patients were part of the dose escalation and 42 additional patients were added to the expansion cohort. Patient discontinuation in the dose escalation part of the study was substantial, 10 patients did not complete the study and one patient died.4 All patients had previously undergone PCD treatment, generally consisting of various combination of chemotherapy alkylating agent, protesome inhibitor, steroid and some patients had autologous transplantation. The trial was not placebo controlled, which of itself makes the data of questionable value without a control arm. (The current Phase 2B and Phase 3 trials are placebo-controlled.)
The publicly-available data trail shows biomarker responses that we believe are consistent with improvements expected from PCD therapy. The current treatment standard for AL amyloidosis seeks to improve end-organ function – i.e., achieve a very good partial to complete organ response – through PCD therapy that eliminates the precursor protein of the amyloid that is depositing in the body. Research shows better hematologic responses improves the chances of organ responses, and significantly impact overall survival.
There are a number of PCD treatments that are effective in generating hematologic responses.5 These treatments include combinations of dexamethasone, oral melphalan, oral cyclophosphamide and the proteasome-inhibitor bortezomib (Velcade). Many of these medications are authorized for multiple myeloma, and are often used off label for the treatment of AL amyloidosis. Two independent retrospective studies of hematologic response rates show the combination therapy of cyclophosphamide-bortezomib-dexamethasone (CyBorD) is highly efficacious. One study reported a hematologic response rate of 94% and a complete hematologic response rate of 71%, while the other reported a 81% hematologic response rate and 42% complete hematologic response rate.
Timing of Organ Responses versus Hematologic Responses
Liedtke et al presented slides at ASH that appear to suggest that NEOD001 trial patients’ organ responses were likely due to the drug. We do not agree with that interpretation. One of the key slides is below. The graph on the left, showing the cumulative probability of an organ response following nFLCr (normalization of the free light chain ratio, a surrogate for hematologic complete response) following chemotherapy, states that the median time to organ response is only 2.1 months. (This data is drawn from Kaufman et al8) Because Liedtke et al point out that the median time in the NEOD001 study from the last PCD treatment was 5.8 months, the implication is that the organ responses observed are likely due to NEOD001 because PCD treatments may have already manifested any organ responses by the commencement of NEOD001 treatment. However, closer examination of the Kaufman et al data belies this conclusion, and emphasizes that there was no wash out period in these trials to control for lingering effects of PCD treatments.
Background: If Organ Response Occurs After Hematologic Response (HR), It Is Generally Rapid
A close look at the cumulative probability of an organ response yields makes a contrary point (red lines added to Kaufman et al graph):
When we add lines to the graph at approximately zero, 10, 20, 30, and 40 months, we see that organ responses have been observed up to approximately 40 months after hematologic responses. Note also that close to 30% of the patients had an organ response before a hematologic response. The organ response preceding hematologic response would seem to impact the 2.1 month-median response time that Liedtke et al cite.
Kaufman et al shows the below rates of eventual organ responders achieving an organ response within six and 12 months from nFLCr, which was the surrogate for a complete hematologic resposne:
The above table shows that shows that if an eventual organ responder did not achieve an organ response within six months from nFLCr, he had an approximately 50% chance of achieving it within next six months without NEOD001.
We calculated the months 6-12 response rates from the Kaufman et al data in the following manner: (12-month rate – six month rate) / (1 – six month rate):
The NEOD001 studies only enrolled patients who had already had a HR.10,11 Based on the foregoing, we therefore conclude the NEOD001 patients’ median time since the last PCD treatment of 5.8 months to be unpersuasive with respect to the drug’s efficacy. We suspect that the improvements in biomarker levels could be attributable to PCD therapy, rather than NEOD001.
We urge Prothena Corporation to release data by individual patient so that investors can better analyze the timing of organ responses, and have a better basis to judge efficacy.
Cardiac Responses versus Depths of Hematologic Responses
We see in the data a lack of correlation between cardiac responses and the depth of hematologic responses, which casts doubt on NEOD001’s efficacy. Kaufman et al’s analysis establishes that there should be a relationship between the depth of hematologic responses and organ responses. The Liedtke et al slides below show the cardiac and renal response rates on the Y axes, versus the depths of the hematologic responses on the X axes (best and last, respectively). The hematologic responses are categorized as CR (complete response), VGPR (very good partial response), and PR (partial response). The renal response tables show the trend one would expect: A deeper hematologic response seems to lead to a greater likelihood of organ response. However, the cardiac response breaks from that pattern – we inserted red trend lines based on the VGPR and PR to illustrate the seeming underperformance of the CR category. The criticism of this comparison is of course the small sample size; however, small sample size is an inherent problem in the NEOD001 trial data to date.
Article by MuddyWaters
See the full PDF below.