Celsion Corp.'s (CLSN) parabolic price action and insider purchases have created a perception of imminent success. With topline data, from the Company's phase 3 HEAT study of Radio Frequency Ablation ((RFA)) plus ThermoDox (heat sensitive liposomal encapsulation of doxorubicin) for hepatocellular carcinoma (HCC), expected in January 2013, the stakes are high for investors. Given less than $0.64 of cash per share, failure will likely result in a share price collapse. Success, on the other hand, will ensure the company's viability, allowing it to raise the required capital to seek ThermoDox regulatory approval in the U.S. and abroad, in addition to financing its potential commercialization.
In this article, I critically evaluate available RFA + ThermoDox data in an attempt to predict possible outcomes in the phase 3 HEAT trial. This study is a large, well-designed trial with a primary endpoint of progression-free-survival ((PFS)). Well designed, however, does not suggest a positive outcome - only a conclusive one. Based on available data and a literature review I conclude that there is no support for a potential ThermoDox PFS benefit in HCC and that the study will most likely fail.
The following are my specific observations:
- Limited clinical experience with ThermoDox. The only available clinical data using ThermoDox in the HCC setting comes from a poorly designed, single-arm, subgroup analysis of a 24-patient phase 1 study, involving only 2 centers. Celsion did not conduct a robust "proof-of-concept" phase 2 (as most traditional drug developers often do) to allow for proper assessment of the drug's therapeutic potential.
- Phase 1 data fail to demonstrate a clear dose-response. Contrary to the Company's assertion of a dose-response relationship with ThermoDox, actual data is highly variable and inconsistent. Further, the available phase 1 trial did not enroll the relevant patient population explored in the ongoing phase 3 program. Specifically, only nine of the twenty-four subjects enrolled were diagnosed with HCC and of those only three had tumors larger than 3cm.
- RFA+ThermoDox responses are consistent with RFA alone. The range of responses observed in the ThermoDox phase 1 HCC subgroup is consistent with published results seen with RFA alone. This observation, along with the lack of a clear dose-response relationship for the drug, reinforces the view that ThermoDox will fail to demonstrate a PFS benefit in the HEAT trial.
The HEAT phase 3 study is a randomized, double-blind, controlled trial comparing RFA in combination with ThermoDox to RFA alone for the treatment of unresectable HCC. According to the study protocol, patients are randomized 1:1 to ThermoDox 50 mg/m2 infusion over 30 minutes, about 15 minutes before RFA begins or to a sham infusion. According to the Company, the trial has enrolled 700 patients with top-line PFS data expected in January 2013. Overall survival, a key secondary endpoint is expected to report out in 2015.
General inclusion criteria include: no more than four lesions; at least one lesion >3cm and none >7cm; and Child-Pugh Class A or Class B status (used to assess the prognosis of chronic liver disease). Under the trial guidelines, the RFA procedure can be performed percutaneously, laproscopically, or surgically. Furthermore, the trial is stratified by lesion size (3-5cm and 5-7cm) and RFA technique.
In his recent initiation of coverage report, Michael G. King of Dawson James specifies that the HEAT trial is 80% powered to detect a 33% improvement in PFS. Mr. King cites an expected 12 month median PFS assumption for the control arm, implying a required 4-months improvement with ThermoDox for study success. As I argue below, there are limited historical data on which to base assumptions about ThermoDox's potential PFS benefit. In addition, while management cites the Berber study as the basis for its overall survival assumptions in the control arm, there is a general lack of high-quality studies to serve as precedents for PFS response to RFA in the unresectable HCC setting.
Limited clinical experience with ThermDox+RFA in HCC
The only relevant clinical data associated with ThermoDox pertains to a phase 1 trial conducted at 2 clinical sites: NCI (U.S.) and Queen Mary Hospital (Hong Kong). The premise for the phase 3 study appears to be based on a post-hoc analysis of a small subset of patients from this trial.
These phase 1 data suggest potential responses of ThermoDox+RFA in HCC but not in other cancers. The study enrolled a total of 24 patients (9 with HCC and 15 with liver tumors metastatic from 9 other primary sites). Subjects were treated with a single 30-minute intravenous (IV) infusion of ThermoDox at 20, 30, 40, 50 mg/m2, starting 15 minutes before RFA. RFA was performed percutaneously or surgically on a total of 28 tumors. Half of the treated tumors were 3.8 - 6.5 cm in diameter. Based on the study's results, the company concluded a high response rate and direct dose correlation as presented below:
Table 1: Time to progression by dose cohort
Time to Progression (days)
MLC - Metastatic liver cancer
HCC - Hepatocellular carcinoma
Source: Company presentation
Data is available to provide additional clarity in relation to the breakdown of results (see Table 2) and potential prognostic factors influencing the median time to treatment failure ((TTF)) outcome. Specifically, primary liver cancer, tumor size equal or less than 3cm, open surgery RFA and a ThermoDox dose of 50mg/m2 or higher seem to correlate with better outcomes.
Table 2: Prognostic factors from phase 1 ThermoDox trial
Source: Poon et al., 2009
Phase 1 data fail to demonstrate a clear dose response
Looking at TTF we see a striking difference between liver cancer patients and others. Specifically while TTF in HCC patients was 355 days, it was only 64 for patients with other primary cancers which metastasized to the liver. The data clearly suggests that only HCC patients derive a benefit from ThermoDox therapy, while other liver metastases do not. This observation reinforces the choice of the HCC setting for the Company's phase 3 trial. It does, however, raise concerns about the relevance of the overall dataset and the claim of dose-response correlation in the overall study.
Can the efficacy measured in the phase 1 trial predict a required PFS benefit to ensure the HEAT trial is a success? Unfortunately, the phase 1 HCC subset leaves more questions than answers. In exploring the trial's probability of success one needs to consider the following numbers from phase 1:
i) NINE: the number of HCC patients treated with Thermodox + RFA
ii) THREE: the number of HCC patients treated at or above 50mg/m2 dose used in the phase 3
iii) ONE: the number of HCC patients with lesions that are >3cm treated with the 50mg/m2 dose.
The Company presented individual patient data from the HCC subset of the phase 1 trial (see table 3).
Table 3: Individual phase 1, HCC patient data
Tumor size (cm )
P - Percutaneous; OS - Open surgery; NR- not recorded
Source: Company presentation
Looking through this dataset, no clear meaningful conclusions can be drawn about dose-response relationships - let alone efficacy relative to RFA. I make the following observations:
i) The relevance of this small subgroup of HCC patients to the one enrolled in the ongoing phase 3 study is limited. Only 3 patients out of 9 (or 33%) in this subgroup analysis would meet the inclusion criterion of the ongoing HEAT study of having tumors >3cm in size. Of those, only one patient was exposed to the 50mg/m2 ThermoDox dose being explored in the phase 3 treatment arm.
ii) Potential prognostic factors such as tumor size, RFA type and ThermoDox dose do not clearly correlate with TTF. Looking through table 2 we see that patients with tumors <3cm undergoing open surgical RFA at doses =>50mg/m2 fair best. Judging, however, by the data presented in the HCC subgroup we see striking variability and inconsistencies. Examples include:
a) We observe 2 open surgical RFA patients with tumor size of 1.7cm. One is in the 30mg/m2 dose group and the other in the 40mg/m2. Yet the TTF of the patient treated with the lower dose is vastly superior at 188 days versus 85 days.
b) The patient who performed best in the trial, exhibiting TTF at 427 days, received 40mg/m2 ThermoDox, had open surgical RFA and a tumor size of 2.1cm. Yet a patient at 60mg/m2 with a 2.5cm tumor exhibited a TTF of only 337 days (a 90 day difference).
c) The one patient in the trial treated at the relevant 50mg/m2 dose had the largest tumor at 6.5cm and percutaneous RFA. That subject exhibited an impressive 374 days TTF despite having the worst prognostic factors. At 60mg/m2, however, we find another patient treated with percutaneous RFA, although the tumor size was not recorded. Based on tumor size distribution in this HCC subgroup it would be safe to assume that this subject had comparable and more likely a smaller tumor size. Yet, despite a higher dose of ThermoDox and better prognostic factors this patient exhibited a TTF of 122 days, a striking 252 day difference between the two.
iii) Potential bias in TTF endpoint is amplified in single arm studies. In the context of the open-label nature of Celsion's limited phase 1 trial, disease progression data should be viewed with caution. Treatment failure was defined as objective disease progression and/or initiation of an alternative anti-cancer therapy. There are several concerns regarding the objectivity of disease progression measures unlike survival. Exact times of progression are unknown, could be over- or under-estimated, introducing potential bias and variability. To counter such challenges independent review, patient randomization, blinding, precise definitions and protocols are a general requirement in quality trials. Most of those (if not all) are lacking in Celsion's phase 1 study.
RFA+ThermoDox TTF responses consistent with RFA alone
In a 2011 publication by Wang et el., the authors explored the addition of intravenous pegylated liposomal doxorubicin to RFA versus RFA alone. This study, while not without limitations, found no difference between the treatment arms on local or total recurrence rates, let alone survival. Median total recurrence rate in the study amongst the 24 patients enrolled was at well over 1 year, i.e. over 365 days. As seen in table 3, the range of TTF responses in Celsion's phase 1 ThermoDox trial was 80-427 days, consistent with published data for RFA alone.
This observation, along with the lack of consistent dose-response relationship, reinforces the view that the addition of ThermoDox will fail to show a benefit in the HEAT trial.
Celsion, as many cash-strapped, small-cap biotechnology companies often do, elected to leap from a limited phase 1 trial to a registrational phase 3 study. In this article I delineated how the Company failed to properly assess all variables that might affect ThermoDox's efficacy in the HEAT study population, thus significantly increasing the risk of failure.
Recent insider purchases appear to support a potential positive outcome. I do not believe, however, that insiders could come to that conclusion based on publicly available data. Further, it is hard to imagine that they would choose to invest immediately prior to the release of topline results were they in possession of material non-public information. I would therefore advise caution in over-interpreting their actions.
Based on the above, demonstrating a statistically significant PFS benefit for RFA+ThermoDox in the phase 3 trial is far from certain and success is indeed doubtful. Betting on a positive outcome on the basis of available data would amount to a pure gamble and should be reserved for those who can afford to lose most of their investment.
Disclosure: I am short CLSN.