From May 2012 to January of this year, shares of Repros Therapeutics Inc. (RPRX) experienced a spectacular rise of over 350%. This resurgence came about as FDA finally provided the company with regulatory clarity to allow advancement of its lead therapeutic Androxal for secondary hypogonadism through pivotal studies under a Special Protocol Assessment (SPA).
In this article I review and critically evaluate the risks related to study ZA-301, the first of two pivotal trials set to report, in an attempt to predict its most probable outcome. As a clearly differentiated therapeutic set to enter the $2 billion global testosterone replacement market, a $200mln market cap, a >20% short interest and sufficient cash through mid-2014, success in this study seems likely to result in significant stock price appreciation.
Review of recent RPRX volatility
With initial Phase 3 data expected by Q2 2013, RPRX's seemingly unstoppable share price appreciation came to a halt on January 28, 2013. On that day the company disclosed that a blinded analysis of the first of two Phase 3 studies (study ZA-301) revealed that subjects in one of its largest enrolling site (site 9) had markedly lower baseline sperm counts when compared to subjects from the rest of the trial's other 16 centers.
While subjects from this site met the strict entry criteria for the study, management felt compelled to disclose this potential 'center-effect' to the FDA and propose removal of its subjects from the efficacy analysis and their replacement with subjects enrolled at other sites. According to the company, these changes would have delayed data from trial ZA-301 by a quarter to Q3 2013 although Androxal's New Drug Application (NDA) filing could still remain on schedule for mid-2014. Shares of RPRX fell 40% on that day from $18.61 to $11.11 at the close.
On February 21, 2013 RPRX disclosed that FDA recommended analysis of all subjects in study ZA-301 per original protocol while including an additional analysis with and without subjects from site 9. FDA accepted Repros' plan to enroll additional patients in its second study, ZA-302. They also noted that Repros should revise the statistical analysis plan and sample size to reflect increased enrollment. Study ZA-302 will also be analyzed with and without patients from the high enrolling site where the patients' baseline characteristics appeared different from other sites in the study. Importantly, it was noted that the SPA for both pivotal trials remains intact.
With data expected to emerge as originally expected in early Q2 2013, shares of the company surged as much as 45% on the day rising from $10.18 to $14.80. That rally, however, was short-lived with RPRX stock price experiencing a continuous decline with shares currently trading below $10. This price action suggests that certain risks and concerns emerged in the minds of investors in relation to the ZA-301 outcome subsequent to the company's disclosures.
Study ZA-301: Odds stacked in favor of positive outcome
The ZA-301 study design includes two co-primary endpoints exploring: (1) normalization of testosterone (T) and (2) non-inferiority to placebo in the percent of subjects with a 50% reduction in sperm count. The efficacy of Androxal in normalizing T is undisputed. It would appear that Investors' fears are currently focused on the possibility that the study might miss on the latter endpoint, leading to an overall failure of the trial. Based on available data, however, I conclude that the probabilities are stacked in favor of a positive outcome on both endpoints in study ZA-301, which could emerge as early as April.
The following are my specific observations:
1) Success on ZA-301 primary efficacy endpoint is easily achievable. Androxal has been studied by Repros in hypogonadal males over a multitude of early trials demonstrating the drug's effectiveness in restoring T levels to a normal range of 300 - 1040 ng/dL in a majority of subjects. Importantly, as demonstrated below, blinded data from the ongoing ZA-301 trial provides direct evidence for success on the T efficacy endpoint by a wide margin.
2) An Androxal drug effect on sperm count reductions is unlikely. Investor concerns in relation to a potential drug-effect on sperm counts observed in study ZA-203 are inconsistent with Androxal's mechanism of action, historical observations and available blinded data from the ongoing ZA-301 trial.
3) Disclosure of issues related to site 9 increases the probability of ZA-301 success. FDA granted Repros a very generous 20% delta for non-inferiority between the placebo and Androxal arm for the demonstration of non-inferiority on the primary safety endpoint of the trial. Under such assumptions, the lower the number of overall subjects in the trial with >50% sperm counts reduction, the lower the number of failing placebo patients required to achieve non-inferiority in the overall study. As delineated below, the inclusion of site 9 in the final analysis, as agreed upon with FDA while still under an SPA, in actuality increases the probability of a positive outcome in ZA-301.
Androxal for male secondary hypogonadism
Androxal (enclomiphene citrate) is the purified trans-isomer of clomiphene citrate (Clomid) used extensively in the treatment of ovulatory dysfunction in women attempting pregnancy. It is a weak selective estrogen receptor modulator (SERM) with relatively low binding to the estrogen receptor (ER) and low anti-estrogenic activity. The drug's mechanism of action allows for competitive binding to the ER acting on the hypothalamic-pituitary-gonadal (HPG) axis increasing testosterone via its normal synthesis pathway. Specifically, blocking the normal negative feedback mechanism of circulating estradiol on the hypothalamus prevents estrogen from limiting the production of gonadotropin releasing hormones (GnRH). The increase in GnRH level then stimulates the pituitary gland to release more follicle stimulating hormone (FSH) and luteinizing hormone (LH). In the testes, LH binds to receptors on Leydig cells, stimulating synthesis and secretion of testosterone. FSH is also critical for sperm production. It supports the function of Sertoli cells, which in turn support many aspects of sperm cell maturation.
Based on this mechanism of action, Androxal should provide an approach to hormone replacement that is more similar to the normal physiology of the HPG-axis. Exogenous testosterone replacement therapy tends to suppresses spermatogenesis and the HPG- axis and appears to cause testicular atrophy. Androxal, however, enhances the endogenous androgen synthesis pathway while maintaining natural circadian rhythm. This method of treatment is especially important in hypogonadal men who wish to preserve their fertility (see figure 1).
Figure 1. The hypothalamic-pituitary-gonadal (HPG) axis
Source: Roth MY et al., 2008
Androxal's Phase 3 efficacy studies
The development of Androxal for male secondary hypogonadism has followed a lengthy and tumultuous regulatory path at the FDA. Originally reviewed by the agency's Endocrine division, officials insisted on a demonstration of a clinically meaningful metabolic outcome other than testosterone for registrational purposes.
A recent move by the company to the Reproductive and Urologic Products division at FDA facilitated a regulatory breakthrough for Androxal with a focus on the treatment of hypogonadism while preserving semen quality as an endpoint of relevance. This was followed by the design and initiation of two Phase 3 efficacy studies under an SPA.
The two trials (ZA-301 and ZA-302), initiated sequentially, are identical studies enrolling 152 men randomized 3:1 active to placebo in a 12-week active dosing period. Basic inclusion criteria include baseline morning testosterone levels <300 ng/dL and sperm concentrations >15,000,000 sperm/mL determined on two separate days. The trials include two co-primary endpoints, one for efficacy and the other for safety:
- Primary efficacy endpoint; at least 75% of men achieve 24 hour average testosterone >300 ng/dL at week 12 (comparison to baseline for active arm only)
- All men in active arm start at 12.5 mg with up-titration to 25 mg at week 6 if morning T <300 ng/dL
- Primary efficacy to be determined on intent-to-treat population
- Primary safety endpoint; comparison of active and placebo arm with respect to the percent of men that exhibit a 50% reduction in sperm counts between two assessments at baseline and two assessments at the end of the active dosing period. The active arm should be non-inferior to placebo.
Success on the primary efficacy endpoint is easily achievable
Based on available data, success in Androxal's Phase 3 trials on the primary efficacy endpoint of restoring T levels to normal at week 12 compared to baseline in ≥75% of subjects should be easily achievable with room to spare.
Repros has studied Androxal in hypogonadal males over a multitude of early trials in addition to an ongoing open-label study (ZA-300). Data consistently demonstrated the drug's effectiveness in restoring T levels to a normal range of 300 - 1040 ng/dL in a majority of subjects. Specifically, as of the company's last update >80% of patients enrolled in study ZA-300 (N=394) had T within normal levels by week 6 compared to baseline (see figure 2).
Figure 2. Percent of men with T in normal range in study ZA-300
Source: Company presentations
Further, and most importantly, blinded data from the ongoing ZA-301 trial provides direct evidence for success on the T efficacy endpoint. Specifically, as of the last update on January 28, 2013 the blinded blended (treatment+placebo) percent of subjects achieving T>300ng/dL was disclosed by management for 105 out of the 152 total patients in the trial (see figure 3). Based on this data it appears that approximately 69% of subjects in ZA-301 achieve T levels within the normal range.
Figure 3. Blinded analysis of 105 subjects within normal T range in ZA-301
Source: Company presentation January 28, 2013
Assuming the 69% blinded rate holds at full enrollment and using the trial's 3:1 randomization, one can explore possible outcomes. In order to hit the primary efficacy endpoint of >75% of men achieving 24 hour average testosterone >300 ng/dL, as many as 51% of placebo subjects could be within the normal range to account for the blinded observation. Such a high placebo rate is highly improbable and well above any expectations based on prior data.
69% = (3/4 x 75%) + (1/4 x X)
X = (69% - 3/4 x 75%) x 4 = 51%
Where X = maximum percent of placebo patients exhibiting T within the normal >300ng/dL range for the Androxal arm to have 75% of subjects hit on the primary endpoint.
Investor concerns relate to the safety co-primary endpoint
Recent investor concerns have focused on the safety co-primary endpoint of reduction in sperm count with regards to study ZA-301's ability to demonstrate non-inferiority to placebo. Based on Androxal's demonstrable clinical effects on the preservation of fertility and increases in FSH (see figure 4), which should promote spermatogenesis, one would expect improved fertility. So the question arises: what exactly lies at the root of investor anxiety?
Figure 4. Effect of treatment on mean sperm counts and median FSH in study ZA-203
Source: Company presentations
Investors concerns most likely relate to the following key points:
i) In Repros' Phase 2 trial (ZA-203) sperm concentrations appeared to decline in response to drug while rising on the placebo arm (see figure 5).
ii) The absence of "failures" in ZA-301 site 9, due to low baseline sperm counts, might adversely affect the study's powering since it represents 26% of the overall study sample.
Figure 5. Effect of Androxal on sperm concentration in study ZA-203
Source: Company presentations
Addressing investor concerns
I address investor concerns with respect to the safety co-primary endpoint in two ways: (1) probability of a drug effect on sperm count, and (2) impact of site 9 on the outcome.
1) An Androxal drug-effect on sperm reductions is unlikely
As highlighted above, Androxal's mechanism of action should preserve or even enhance fertility. Yet, we are faced with data pointing to the possibility of sperm-count reductions on therapy. Some have hypothesized that this counter-intuitive observation is a function of increased male libido due to higher testosterone levels. This in turn raises the probability of increased frequency of intercourse or masturbation throughout the study potentially reducing sperm counts - but only in the treatment arm.
While this is a reasonable hypothesis to consider, I would make the following observations:
1. No evidence of dose effect in ZA-203. Median sperm concentration reductions in the Androxal arm of study ZA-203 do not exhibit a dose effect. Instead, they seem to be a function of variable median baseline measures. Specifically median reductions on Androxal were 37% and 25% in the 12.5 and 25 mg doses respectively. Median baseline measures were approximately 95 million and 65 million in the 12.5 and 25 mg doses respectively (see figure 5). Arguably, there are two outlying sperm-counts in figure 5: a high baseline in the 12.5mg arm and a low end-of-study (EOS) measurement in the Testim arm. The other six bars demonstrate normal variation around 55mln sperm/ml.
2. No issue in ZA-203 using identical endpoint. A post-hoc analysis of study ZA-203 using the Phase 3 SPA-defined safety primary endpoint of >50% reduction in sperm counts on a last observation carried forward (LOCF) basis showed "failure" rates in both placebo and treatment arms (see figure 6). Those rates were 15%, 8% and 21% in the placebo, 12.5 and 25 mg Androxal respectively.
3. No evidence of libido effect. In prior double-blind placebo-controlled trials of Androxal in hypogonadal males, the drug failed to demonstrate an effect on libido as determined using the DISF-SRII scale.
4. No evidence of dose effect in blinded ZA-301 data. Per recent management disclosures, no correlation has been observed to date in study ZA-301 between the change in T and change in sperm counts. Further, an outlier failure site has been identified accounting for roughly half of the sperm counts failures observed to date (out of the total of 17 sites). Both these observations are inconsistent with a sperm-reduction drug effect.
Figure 6. Sperm count reductions in study ZA-203 based on Phase 3 SPA definition
Source: Company presentations
2) Disclosure of issues related to site 9 increases the probability of ZA-301 success
Repros' management disclosures with regards to the outlier site 9 on January 28, 2013 introduced a number of risk variables in relation to the trial's timing, conduct and potential regulatory outcome. FDA's prompt response to the company's query on February 21st addressed the timing and regulatory risks. It would seem, however, that lingering concerns remain in relation to the study's conduct and potential outcome in the context of a large outlier site.
As a reminder, site 9 enrolled 40 subjects out of the total 152 (or 26% of the total ZA-301 sample). None of these subjects experienced a decline of >50% in sperm counts, as they were referred from a fertility clinic. Consequently, their mean baseline sperm count was 17.6±1.7 million compared with 80.7±74 million at the other 16 sites. FDA told Repros to proceed with the analysis of ZA-301 as previously planned, and that data should be evaluated both with and without patients from site 9.
In an attempt to understand the implications of site 9, the following sensitivity table (see table 1) was constructed to help gauge potential outcomes for study ZA-301 based on the 20% delta for non-inferiority granted by FDA. As shown below, success on the study's safety primary endpoint is set at a very low bar:
Table 1. Non-inferiority sensitivity table for study ZA-301
As demonstrated in Table 1, the lower the number of overall subjects in the trial with >50% sperm counts reduction, the lower the number of failing placebo patients required to achieve non-inferiority in the overall study. Consequently, the inclusion of site 9 in the final analysis, as agreed upon with FDA under SPA, increases the probability of a positive outcome in ZA-301.
I will use the following analysis as an example. Based on management's disclosures, we are aware of 5 out of 30 subjects (17%) with >50% sperm count reductions among the 16 sites excluding site 9 (for which we know failures = 0). If this rate were to hold, one would expect a total of 19 failures overall in study ZA-301 (based on [152-40] X 0.17 = 19). By implication, success would require only 2 placebo patients out of the total of 38 to fail (5.3%). Based on prior observations in study ZA-203 (see figure 6) in which 3 out of 20 placebo subjects (15%) exhibited >50% sperm counts reductions, success in ZA-301 should be easily achievable.
Had the overall sample of 152 subjects had a 17% failure rate on the primary safety endpoint, a total of 26 subjects would be considered failures. Under this scenario a minimum of 4 placebo patients would need to be considered failures (10.5%) for non-inferiority to be achieved.
Exclusion of site 9 from the analysis is still likely to result in a positive outcome based on prior experience with Androxal as seen in study ZA-203. Admittedly, if one excludes site 9 from the analysis entirely, ZA-301 is less well powered to exclude a 20% reduction in sperm-count failures. That said, such an analysis, as requested by FDA, is likely to be exploratory rather than restrictive from a regulatory standpoint.
With data expected in early Q2 2013, RPRX presents a compelling investment opportunity. As discussed above, based on the evidence presented, the probability of success in the Androxal ZA-301 study co-primary endpoints has in fact increased over the recent period of uncertainty.
The volatility in RPRX shares has come about due to potential uncertainties investors faced subsequent to management's disclosure of detailed issues regarding the conduct of Androxal's first pivotal trial ZA-301. At the same time, investors have been offered a unique glimpse into blinded data from the ongoing study, facilitating an in-depth analysis of its risk-reward.