ATRS: Xyosted vs. the Alternatives

Read the article disclaimer first.


  • The FDA issued a CRL for Xyosted; citing the risk of depression, suicide, and blood pressure as reasons for rejection.
  • The issue of depression doesn’t appear to pose a serious threat to approval.
  • If the FDA chooses to approach suicidality quantitatively, Xyosted has little chance of approval.
  • Xyosted’s blood pressure data is highly concerning but it may be overcome in one of several ways.


Xyosted is a short acting testosterone ester developed by Antares Pharma (ticker: ATRS) for the treatment of hypogonadism.

In October of 2017, the FDA issued a CRL for Xyosted’s first review cycle. The main concerns were:

1. The worsening of depression, classified as a serious adverse event (SAE).

2. The instance of suicide; a cause and effect involving Xyosted could not be ruled out.

3. The clinically meaningful increase noted in blood pressure.

ATRS resubmitted (class 2 response) the NDA with a PDUFA action date of September 29, 2018.


We begin with the first point, that of depression. For now, we largely separate its discussion from that of suicide for a couple of reasons.

First, while depression may lie on the causal pathway to suicide, the exact relationship between mental states and suicide is complex and often difficult to disentangle; especially in light of numerous possible confounders found in clinical trials.

Second, we know that in trials such as Xyosted’s, a new (but not pre-existing) case of depression or worsening depression would be counted as a treatment emergent adverse event (TEAE); serious or other.

In light of this, three participants in the pivotal trial were said to have experienced a total of three SAEs: one case each of worsening depression, suicide, and vertigo. Ergo, the data implies numerically that the instance of suicide was not directly related to the separate case of worsening depression.

This does not, however, preclude the possible fact that the individual who completed suicide had a history of depression as the pivotal trial’s criteria does not explicitly specify excluding individuals currently or previously treated for depression, or those with a history of suicidality.

Since we don’t have access to important clarifying information, such as patient narratives, we have to leave a discussion of the nevertheless possible connection between depression and suicidality until later.

Instead, as the argumentation above allows us to simply focus solely on depression for now, we can ask ourselves a very important question:

Given the available data, does Xyosted lead to a greater risk* of depression when compared to marketed testosterone replacement therapies (TRTs) approved within the last decade?

The answer is unequivocally: no, Xyosted does not.

*It’s important to note that the term “risk” in this report is used to denote a mathematical correlation (proportion), not necessarily an instance of cause and effect.


Upon review of pertinent data, we find the following:

A. In Xyosted’s pivotal study, 0.67% of individuals experienced depression (1 case; n = 150). If we include the safety extension study, this percentage drops to 0.35% of participants (n = 283) and to 0.27% among all available trial data with results (n = 377). This roughly approximates out to be 447 cases of depression per 100,000 patient years.

B. Numerous pivotal and safety extension studies used to help approve 6 separate TRTs in the past decade (not including bioequivalence-only NDAs), yielded 22 cases of new or worsening depression among a total of 1,999 participants (1.10%). Where data permits, we can approximate this to be an average of about 1,188 cases of depression per 100,000 patient years.

C. If we extend B’s data range to include 5 European TRT clinical trials, as well as 7 post-marketing TRT studies, new or worsening depression was noted in 0.56% of individuals (31 cases; n = 5536). There’s not enough data to estimate an adequate rate per 100,000 patient years.

The comparison between A and B is of a higher caliber due to higher similarities between study designs; whereas comparisons to C should be taken a bit more lightly due to a relatively cruder cross-study pooling of data.

Regardless, we can still reasonably state that Xyosted is no worse than its competitors with respect to risk of depression.

Most importantly, ATRS mentioned that the individual with worsening depression had recently stopped taking antidepressants.

If depression was Xyosted’s sole concern upon resubmission of the NDA, we could fairly confidently conclude that it would be approved during the second review cycle given a similar re-analysis on ATRS’ part.


But depression is not the only concern here.

Unfortunately, a case of suicide was noted in Xyosted’s pivotal trial.

We must ask: has the FDA ever approved a drug, not intended for the treatment of psychiatric illness, with one or more suicides in the treatment arm? Yes, on more than one occasion, and for a wide range of reasons (some of which we discuss later).

A more germane question would be:

Has the FDA ever approved a drug, not intended for the treatment of psychiatric illness, which posed a unique safety signal with respect to suicide, above and beyond its competitors?

Because this is ultimately the issue here. The risk of suicide with Xyosted appears to be far greater than with any other currently marketed TRT.

First, not a single TRT approved within the last decade recorded a case of completed suicide in the treatment arm of any phase’s trial. Meaning Xyosted poses, at minimum, a quantitatively unique safety signal.

It’s harder to argue that this is a qualitatively unique safety signal as post-marketing data (including from FAERS) makes it clear there is some, not fully elucidated, correlation between TRT and suicide.


Here’s the relevant data on this point; the same kind of data the FDA has required of sponsors in the past.

A. 0.67% of individuals completed suicide in Xyosted’s pivotal trial (1 case; n = 150). If we include the phase 3 safety extension study, this comes out to be 0.35% of participants (n = 283); and 0.27% among all available trial data with results (n = 377). Given publicly available information, this is around 447 cases of suicide per 100,000 patient years.

B. 0.00% of 1,999 participants in the pivotal trials used to approve 6 TRTs in the past decade completed suicide. Moreover, even suicide ideation and behavior was practically non-existent.

C. As before, after extending our data to include another 5 European clinical trials and 7 post-marketing studies, 0.02% of participants completed suicide (1 case; n = 5536). An adequate approximation of completed suicides per 100,000 patient years cannot be gleaned from this set of publicly available data.

D. Most men in Xyosted’s pivotal trial were 41-66 years old, White, Black, Hispanic, and Asian; observed mainly in 2015. Applying the same criteria to the U.S. population at large, we come to find that 0.03% of all such men in 2015 completed suicide (14,961 cases; n = 51,530,467). This equates to, in a manner of speaking, 29 suicides per 100,000 “patient years”.

While D establishes only an approximate comparison given trial exclusion/inclusion criteria, the FDA tends to look at this type of information even if only to get a very general understanding of trends.

Therefore, we can clearly see that the risk of suicide among the general population (0.03%) closely mirrors that of overall TRT trial data (0.02%), while the risk in either is far smaller than that of Xyosted’s (0.27%).


We can look to FAERS data for additional clues. Of course, we must keep in mind that FAERS data is heavily confounded by numerous factors but what we find thanks to FAERS helps put a few things in perspective.

A search of this system, limited to men on TRT with a known adverse event date that occurred in the last decade (01/01/2008-03/31/2018), yielded 14,816 cases of adverse events (each case could have multiple adverse events) and 676 deaths; 5 of which were suicides.

In this subset, the FAERS data was heavily skewed towards reporting serious adverse events when compared to any TRT clinical trial data. This means that the number of actual adverse event cases for TRTs in practice is probably far higher than 14,816 for the last decade; as minor events are probably less likely to be noticed and/or serve as an impetus for reporting.

Conversely, this means we can place greater confidence in the notion that the number of deaths and suicides correlated to TRT were more adequately and proportionately reported in FAERS.

Because causality is not established in FAERS, and is often difficult to establish for suicides in clinical trials as well, we make no assumptions as to which deaths were legitimately attributed to the effects of a study’s experimental intervention. Therefore, all the numbers presented below are raw.


As such:

A. FAERS data tells us that suicides constituted 0.74% of all deaths correlated to any/all TRTs out on the market (5 suicides; 676 deaths).

Suicides comprised 0.04% of adverse event cases (5 suicides; 14,816 adverse event cases).

B. Data from the pivotal, extension, and post-marketing studies on TRTs approved in the past 10 years reveals that suicides constituted 12.5% of all deaths (1 suicide; 8 deaths).

0.05% of all adversely affected individuals (n = 2,012) completed suicide. Of note: this latter percentage (0.05%) appears to be consistent with approximately equivalent FAERS data (0.04%).

C. Available phase 1, 2, and 3 data (n = 377) from Xyosted’s trials show us 100% of deaths were as a result of suicide (1 suicide; 1 death).

0.46% of individuals experiencing an adverse effect (n = 218) completed suicide.

In summation, what most of this data tends to show is that suicide (quantitatively speaking) poses a serious and unprecedented safety signal for Xyosted with respect to its drug class.

It appears that, among individuals with adverse events, the risk of suicide is roughly 10 times greater for Xyosted than for any other TRT currently in use.

The one flipside is that the raw proportion of suicides to deaths in all TRT trial data (2 suicides; 9 deaths; 22.22%) is very similar to the same proportion in similar men among the general population.

That is to say: in 2015 suicides accounted for 27.75% of all deaths in White, Asian, Black, and Hispanic men aged 41-66 (14,961 suicides; 53,914 deaths). Again, this is a rough comparison given numerous limitations.


Having understood the quantitative significance herein we can now ask again:

Has the FDA ever approved a drug, not intended for the treatment of psychiatric illness, which posed a unique safety signal with respect to suicide, above and beyond its competitors?

It may surprise us, but the answer is actually yes. The most famous recent example of this was Siliq, a biologic psoriasis drug with no orphan or expedited designations no less.

But a comparison to Siliq, numerically speaking, would be remiss without important qualifications.

First, while Siliq posed a far greater suicide risk than any other marketed psoriasis drug ever approved, so much so that its clinical development was partially halted, its risk was nonetheless very small when compared to Xyosted.

Among all available clinical trial data, Xyosted’s risk of suicide was 0.27% (1 suicide; n = 377); or 447 cases of suicide per 100,000 patient years.

Compare this to Siliq’s unprecedented risk of suicide among its own drug class of “only” 0.10% (6 cases; n = 6243); which was equal to 58 suicides per 100,000 patient years.

Quantitatively speaking, it’s quite evident that Xyosted correlates to a greater threat of suicide than even Siliq. This may be grounds for rejection as Siliq just barely got approved for reasons we discuss below.


How then, might Xyosted be approved? Well, luckily for ATRS, the FDA often qualifies quantitative information in ways that might rescue this NDA.

For instance, the FDA brought Siliq before an advisory committee because its risk of suicide was higher than for even many psychiatric drug trials with higher background risks for depression and suicide.

Numerous divisions and reviewers across clinical disciplines recommended it not be approved as a result of this. Things seemed bleak before, in so many words, the lead reviewers and director overruled the initial reviewers to get the drug approved.

Their justification was in part that, despite the issue of suicide, this new biologic promised an important alternative in the treatment of psoriasis for patients who had no other choice.

This was an important qualitative rationale used to overrule a serious quantitative risk: Siliq was a unique biological molecule that could help someone better than marketed alternatives.

The important question at this point is:

Does Xyosted provide a similarly important qualitative alternative to existing TRTs?

While the dosage form is unique, the active pharmaceutical ingredient is not (it’s an alternative, shorter acting, testosterone ester – no more).

So we must assume Siliq’s strong qualitative justification for approval despite its risk of suicide won’t work in Xyosted’s case; unless there’s evidence to suggest men eventually stop responding to various testosterone esters or that subcutaneous TRT solves an unmet medical need.

Luckily for ATRS, the FDA has approved effective drugs potentially linked to suicide during pivotal trials, even when they offered no significant advantages over alternatives, for numerous other qualitative reasons.


ATRS could argue that the suicide in its pivotal trial provides no basis for establishing cause and effect. In Xyosted’s case, it’s clear that there are several ways by which this can be argued to increase the chances of approval:

1. Pre-existing confounders. ATRS could show the individual who completed suicide had a history of depression, suicidality, or was no longer taking anti-depressants if he should’ve been. A life-event, even after a long period of uneventful treatment with Xyosted, may have triggered the completed suicide.

2. Abuse and/or withdrawal. If the subject abused Xyosted and/or another TRT then this can lead to an increased risk of mood disorders and suicide. Similarly, withdrawal due to abrupt cessation of abused anabolic steroids can do the same. Labeling and REMS would be used to help mitigate such an issue.

3. Time-frame. Although complex interrelationships are possible, if the suicide occurred weeks after Xyosted was last taken, proving a causal link is extremely difficult.

4. PK data. If ATRS has the relevant PK data, it may be able to show that the data from the subject who completed suicide is unremarkable. In other words ATRS could argue, from a clinical pharmacology perspective, that Xyosted couldn’t be clinically linked to the suicide even if some form of biological plausibility for it existed.

5. Hypogonadal men may be at higher risk of depression and/or suicidality and numerous independent studies have shown improvements in mood thanks to TRT. If ATRS conducted prospective assessments of mental health in its studies, it could show that the risk of depression remained stable or improved compared to baseline, making it harder to argue the drug’s effects were part of the causal pathway to suicide.

This is assuming, of course, the study’s baseline was similar to the background prevalence of depression/suicidality in the hypogonadal population.

In any case, it may even be feasible to argue that drug-induced depression doesn’t pose the same risk for suicide as would a case of major depression.

6. Unfair comparisons. There is another important thing ATRS can do to improve its chances of approval on this point. Assuming no omission in public data was made, Xyosted’s pivotal trial and safety extension study makes no explicit mention of excluding patients with a history of mental illness.

The same cannot be said for many of the other TRT clinical trials. Some of them specifically excluded some individuals such as these. This, then, would artificially inflate Xyosted’s risk of suicide when compared to the “less risky” population pools used in other TRT clinical trials.

As such, claiming that Xyosted is riskier than other TRTs becomes that much more difficult.


However, ATRS did state that it couldn’t rule out that the suicide was related to Xyosted.

This is a bit worrisome as it could signal that it can’t use any of the arguments above to avoid a cause and effect relationship from being established. Or, it could simply mean that it’s hard to be sure either way as is simply most often the case. After all, even trial investigators and FDA reviewers disagree on when a drug is linked to an adverse event or not.

As a result, completed suicides may be seen as suggestions, not proof, of cause and effect; meriting further post-marketing investigation and risk mitigation but not necessarily rejection of a drug.

We also have to make the cold statistical assessment here that, given enough TRT trials, a completed suicide utterly unrelated to the study drug is bound to happen and that it may have inevitably done so in Xyosted’s trial due to chance alone.

Even if ATRS cannot adequately disprove a causal link between Xyosted and suicide, all is not lost; with some important restrictions in mind.


In cases where the FDA was unsure about the true link between suicide and a drug, or wanted to be cautious, it required any combination of the following upon approval:

  1. Clear labeling on the increased risk of depression and/or suicide.
  2. Very strong REMS with ETASU (Elements to Ensure Safe Use).
  3. Post-marketing (observational) studies to assess the risk of suicide.

This means that, if Xyosted cannot disprove a causal link to suicide, its patient base may be significantly curtailed. May is the key word here. The FDA is unlikely to completely restrict the use of this drug for individuals with no history of depression or suicidality.

That’s because such a strategy will force some patients to fail to disclose such medical histories in order to gain access to Xyosted. By extension, this means the patient and physician will not be fully engaged in the patient’s medical care.

The worst-case scenario, of course, is a CRL. In which case, we might expect the FDA to require a better clinical (interventional) study to assess the risk of suicide. This seems like the least likely scenario, however. Such a study would have to be very large, quite long, highly difficult to design, and potentially unilluminating given the need to capture extremely rare events (completed suicides).

Because ATRS mentioned there was no need for a study after its type A meeting post-CRL, we have to assume such a study is extremely unlikely to be required.

Not to mention: such a study requirement poses ethical questions.


On to the issue of blood pressure.

It’s known that hypogonadism may be associated with hypertension itself yet, through a couple of important mechanisms, TRT can result in much of the same despite its simultaneous potential for vasodilatory action.

And so, Xyosted’s pivotal trial showed a clear tendency towards a high risk of hypertension.


12.7% (19 cases; n = 150) of individuals in the pivotal trial and 2.26% of individuals (3 cases; n = 133) in the safety extension study experienced hypertension; 7.77% between the two trials. This is about 10,161 cases of hypertension per 100,000 patient years.

Compare this to all of the other phase 3 data collected for 6 different TRTs approved in the last ten years. Among all trials, 81 or 4.05% of individuals experienced such an adverse event (n = 1,999). Where possible, we can approximate this to be 3,564 cases of hypertension per 100,000 patient years among such trials.

If we include the European clinical trials and post-marketing studies, the risk of hypertension is 1.88% (104 cases; n = 5536). An adequate per 100,000 patient years comparison is not possible given the available data.

Regardless it appears that Xyosted poses, at minimum, a threefold increase in the risk of hypertension when compared to TRT alternatives; helping to explain this last part of the CRL. This is clearly a quantitatively important safety signal.


Within the last 3.5 years, the FDA has become concerned about the cardiovascular risks TRTs may pose (stroke and myocardial infarction). This is especially so for men treated with TRTs in a chronic and off-label manner; as these are the individuals who are most likely ones to have surrogate risk factors, like hypertension.

Consequently, the FDA has increased the barrier to entry by requesting that sponsors conduct ambulatory blood pressure studies in order to assess for this risk. ATRS has stated they had conducted such a study pre-CRL.

Furthermore, based on minutes from a type A meeting with the FDA, ATRS believed that they didn’t need to conduct another study for its re-submission. Instead, a re-analysis of existing data appeared to be enough.


This means that ATRS must prove, via its re-analysis, as many of the following in order to increase its chances of approval:

1. There were no clinical sequelae as a result of the increases in blood pressure: the increases were minor (no more than 1-2 mmHg and/or below any pre-defined threshold), temporary, and resolution was spontaneous. Theoretically, there may have even been a coding issue, where “hypertension” was used somewhat inappropriately to classify such minor increases.

2. The increases were not correlated to dose or duration of treatment, nor concentration of serum testosterone.

3. The same staff and equipment were used to acquire the data on each visit.

4. The safety extension trial showed a much smaller risk for hypertension.

5. A large proportion of trial participants already had hypertension, more so than alternative TRT trials. Hypertension, after all, is the most common concomitant diagnosis among men with hypogonadism. 15% of such men have hypertension. Such a high baseline risk, given the known trial exclusion criteria, of hypertension clearly makes Xyosted’s trial results appear worse than necessary.


Assuming ATRS’ re-analysis can prove as many of the aforementioned, a clinical trial may indeed be unnecessary as claimed. Yet if ATRS’ re-analysis is inadequate, it won’t take much for the FDA to issue a CRL. After all:

1. No active comparator was used in the pivotal trial nor, as far as we know, the important ambulatory blood pressure study. The new trend is to use such a comparator, namely with ambulatory blood pressure readings, as it’s difficult to make an accurate assessment of risk without a comparator arm (even if placebo).

2. There was a relatively high drop-out rate in the pivotal trial, making it more difficult to put things into proper perspective.

3. The FDA has made it very clear it’s concerned about the long term cardiovascular consequences of testosterone therapy, especially in age-related hypogonadism. So if ATRS cannot prove long-term safety, it may be in trouble. The high drop-out rate in the pivotal trial makes proving this difficult.

4. The ambulatory blood pressure data comes from the supplemental safety study, not from the pivotal trial where we saw the issue of hypertension. The FDA could question its relevance.

5. On the note of the safety trial, where the apparent risk of hypertension was far lower than the pivotal trial’s: if the titration regimens between these trials were different, this may be a serious confounder the FDA won’t be able to look past.

6. On the note of a potentially high baseline risk of hypertension making Xyosted appear worse than it actually is.

On the one hand, we might actually expect a quasi-regression to the mean in such cases; where we should see a consequent drop in blood pressure with treatment as a result. This obviously didn’t happen.

On the other hand, the pivotal trial’s results may show us that those with pre-existing hypertension are at an increased risk of worsening their hypertension while taking Xyosted; which may explain Xyosted’s comparatively poor results.

A high baseline risk of hypertension clearly makes Xyosted’s trial results appear worse than necessary. One would think this would afford leniency on the FDA’s part. Yet we know that in similar situations in the past the FDA actually requested a new trial be conducted; where this exact baseline risk was minimized instead.

7. And the final caveat comes to us from the American College of Cardiology and American Heart Association. In November of 2017, after Xyosted’s CRL, the blood pressure classification scheme changed. For Xyosted, it changed for the worse.

Given the new definitions of hypertension under such guidelines, we can reasonably expect (all else equal), that Xyosted increases the risk of hypertension by far more than just 12.7%. If the FDA applies these new guidelines in its re-analysis, then this may be a hurdle too far to overcome.

And the FDA has every right to do so, given the significance of this change.

Of course, one potential workaround for many of these stumbling blocks is for ATRS to acquiesce to more stringent labeling and REMS given the highly evident potential for off-label use.

In other words, if Xyosted is strictly limited to use in younger men, with classical hypogonadism, then the risk/benefit ratio will improve markedly and approval will become that much more likely.


How can we conclude this lengthy look at Xyosted’s chances of approval?

On the note of depression alone, it looks like nothing stands in its way of FDA approval.

With respect to suicide, if the FDA chooses to focus solely on the numbers, there is very little hope of approval. However, the FDA appears to take a qualitative look at this issue more often than not, which makes approval far more likely.

And when it comes to blood pressure, this is actually the biggest question mark. If we are to believe that ATRS can re-analyze its data in a way that proves the increases in blood pressure aren’t as bad as they appear, and agree to stricter labeling and REMS, then there is no reason for rejection.

However, recent history is not on ATRS’ side with respect to this concern. It won’t be surprising that if a CRL is issued, a new ambulatory blood pressure study—this time with an active comparator—will be required.

What do you think? Comment below!

Published: 6/20/2018

Author: Artem Cheprasov

This article’s author did not receive, nor was promised, any form of compensation for researching, writing, or publishing this article. While researching for, writing, and publishing this article: the author did not own any shares of any security mentioned herein.

The following were used as sources of information for this report:

  1. NDAs: 202763, 203098, Androgel, Austedo, Aveed, Axiron, Benlysta, Chantix, Daxas, Delatestryl, Fortesta, Latuda, Natesto, Propecia, Rytary, Siliq, Taltz, Vogelxo
  2. Advisory committee meetings on the following: Aveed, Daxas, Jatenzo, Siliq, Testoterone replacement therapy, Tlando
  3. Published literature (PMID): 10522982, 12523867, 14718489, 17701659, 19440073, 19653976, 20520285, 20524974, 22574772, 24501728, 28673432, 28812471
  4. Xyosted Data: Phase 1, Phase 1/2, Phase 2, Pivotal, Safety Extension
  5. FAERS
  7. SEC: ATRS 10-Ks; 2014-17
  8. Misc: ATRS confirms AMBP data, CRL PR, new BP guidelines, sponsor’s resubmission strategy, FDA TRT caution.

This article’s comment section is intended for the generation of (counter)-arguments based on reasoning or evidence, helpful reminders, or novel thoughts. Questions based on any of the aforementioned are encouraged.  Sources are always appreciated.

Please understand that in order to sustain a more substantive, intelligent, and insightful discourse for all: comments containing foul language or ad hominem-like attacks against any entity will not be admitted.



  1. Romulo August 9, 2018 at 11:35 am - Reply

    Fda doesnt worry about someone’s Health . They might be in pressure because they need to sell nebido and andogel.

  2. Anonymous June 20, 2018 at 10:30 am - Reply

    What happened to Jatenzo?

    • Healthonym June 20, 2018 at 10:45 am

      Jatenzo’s developer is a private company and so this may impact what we hear, if, and when. Their NDA was resubmitted in June of last year; years after the original review cycle ended. This means we should’ve heard about the FDA’s decision by now. Given the silence, and a poor advisory committee result in early 2018, we have to assume they received a CRL again but have failed to disclose as much.

Leave A Comment