Update, 02/15/2017: A section about "serious" versus "severe" liver injury was added. This offers further explanation as to the potential causes of the resulting CRL.
A few weeks ago, multiple news sources alerted the public about a death caused by a super-bacteria resistant to all known approved antibiotics. A few days before that, the FDA rejected the approval of a new antibiotic. What is it, and why was it rejected?
The new antibiotic is called solithromycin, developed by a small company called Cempra (NASDAQ: CEMP).
My thesis, in short: the FDA rejected solithromycin's NDA (new drug application) on the basis of bogus liver safety concerns, due to an institutional echo chamber effect.
If things continue on their current course, Europe and Japan will start treating their citizens with soli years before the American public gets one dose. EMA approval is slated for very early 2018, and Japan at some point in 2018. With this rejection, the FDA is sentencing thousands of patients to death, costing our healthcare system potentially billions of dollars.
Solithromycin took over 10 years and over 400 million dollars to develop. It's therefore no coincidence that large pharmaceutical companies would decide - and have decided - to avoid the risks of major antibiotic development. The risk is offloaded to small, vulnerable companies like Cempra. This recent FDA decision chills funding in antibiotic development and makes a mockery of Obama's 2014 Executive order and the spirit of the Cures Act.
Based on statements made by President Trump, investors and the public might ask if the FDA's decision could be, logically, reversed. The contenders for FDA Commissioner, including Joseph Gulfo, seem in agreement on current FDA overreach in not approving safe and effective drugs.
So, is solithromycin safe and effective, and will the new FDA Commissioner agree? Investors, and the public, stand to gain an enormous benefit if the answer is "yes".
= Section 1: Introduction =
The outgoing Director of the Office of New Drugs at the FDA's CDER claimed, in a recent interview, that companies often don't honestly portray private FDA communication, such as guidance for drug programs or NDA deficiencies given in complete response letters.
In early 2016, Robert Califf was appointed as head of the FDA; that year, the number of approved NDAs / BLAs was half of 2015's, despite a slightly reduced number of submissions from the year before. In fact, the number of 2016 approvals was at the lowest level since 2010, despite multiple drugs slated for early 2017 being approved in December.
In 2016, Sarepta's (NASDAQ: SRPT) accelerated approval request, though ultimately approved by Califf, was very nearly denied; this would've spelled an early death for many boys with a rare genetic muscle disease.
Perhaps it is the fault of greedy companies submitting faulty applications. But if this lowered rate is due to management, why did it go down so sharply from 2015? Was it just random luck?
Many drugs in 2016 were rejected on the basis of heightened safety fears: fears, not facts. This was the case with solithromycin, which, despite being a safe and effective drug, received a massive roadblock in the form of a requirement for an extra (potentially very time-consuming and expensive) 18,000 patient study, with a threat of a restrictive label, thus throwing the commercial viability of the drug into question.
People might think that the FDA exists to protect the public, but if isn't, why not, and can it be fixed? Could solithromycin actually be safe and effective, even if the FDA doesn't seem to think so? If so, what can we do about it, and what is the implication for Cempra? These are the questions and topics I will cover in my article.
= Section 2: What is Solithromycin and What Are Its Benefits? =
Antibiotic-resistant bacteria cause the death of 23,000+ people yearly in the US, and solithromycin, Cempra's lead antibiotic, can kill them. Out of the three most-urgent threats identified in the CDC's comprehensive 2013 report, soli has comprehensive activity against at least two: c. difficile and drug-resistant Neisseria gonorrhoeae.
Cempra ran, or is running, trials in moderate-to-severe CABP (community-acquired bacterial pneumonia), antibiotic-resistant gonorrhea, COPD, and NASH. This breadth of use could provide significant public health benefits over current treatment options. There are five verifiable benefits over current antibiotics:
(1) It kills antibiotic-resistant bacteria. It not only kills bacteria resistant to other antibiotics, but prevents the development of solithromycin-resistant bacteria.
(2) It's easy on the stomach and doesn't cause additional infection. It prevents c. difficile infections, and, in fact, cures them.
(3) It doesn't have many of the rare side effects of other similarly powerful antibiotics.
(4) Unlike many other antibiotics, it can be safely given in oral or IV form, saving money to the healthcare system.
(5) It's potentially more effective in older people.
1) Solithromycin kills antibiotic-resistant bacteria.
As reported by various news outlets, academic journals, and the CDC, resistance against many kinds of antibiotics & antifungals is rapidly rising; in the US, at least 23,000 people a year are dying as a result.
In Europe, for common macrolide antibiotics (e.g.: azithromycin / Zithromax), resistance is growing: there, about 25% of known isolates of the common s. pneumoniae bacterium are resistant to Zithro. In the US, it's already 50%, and in Asia, 80%. As resistant bacteria spread, commonly-prescribed antibiotics are becoming less effective:
When a bacterial infection cannot be treated due to drug resistance, another antibiotic drug class can be added; e.g.: ceftriaxone can be combined with azithromycin for greater efficacy. However, multi-drug treatment after single-drug failure: [a] compounds the safety risks of each drug, [b] increases the risk of mutations, leading to antibiotic resistance, [c] results in lost productivity, and [d] can come too late to save the patient, especially if they're older or immunocompromised.
Introduction to Plasmids
In the body, self-contained genetic mutations called plasmids hop from one bacterium to another. Thus, a mutation that makes a fairly benign bacteria antibiotic-resistant can hop to a more dangerous species during an infection: there's no need to develop resistance independently! This bacterial conjugation, combined with globalization, has accelerated the spread & mutation of difficult-to-treat bacteria.
In 2016, the CDC warned of a nightmare scenario: bacteria resistant to all known antibiotics may soon emerge if nothing is done. Simple infections would again become deadly, as they were before widespread use of antibiotics three-quarters of a century ago.
The CDC recently cited the first case of the mcr-1 gene in a bacterial infection in China. This gene, which sits in a plasmid, can make bacteria resistant to colistin, a last-resort drug for multidrug-resistant infections. Less than 6 months later, the first case in the US was reported. The CDC stated that if mcr-1 combines with carbapenem-resistant Enterobacteriaceae (featuring a mortality rate of possibly 50%), it could result in a nightmare bacteria that is impossible to treat with any known antibiotic.
That bacteria is getting closer to reality. In January 2016, the CDC announced the first documented instance of pan-resistant New Delhi metallo-beta-lactamase-producing Klebsiella pneumoniae, a new strain resistant to 26 commercially-available antibiotics. The identified strain didn't even need to combine with mcr-1 to gain its already-frightening characteristics. (See this for an excellent companion article.)
Solithromycin to the Rescue!
It's still not known how many Klebsiella isolates soli is effective against. However, in Phase 3 trials, patients treated with soli were able to clear away ~75% of Klebsiella infections, slightly more than moxifloxacin, at 66.7%. Soli even seemed to have activity against e. coli infections & other kinds of deadly bacteria. (see this, pages 154-158)
To kill azithromycin-resistant bacteria might require an addition of ceftriaxone, or the more potent moxifloxacin. But orally-dosed soli can knock out that bacteria just as well, without inducing these other drugs' side effects, & before a patient becomes critically ill.
As an example, let's examine the potency of solithromycin against the common, and deadly, s. pneumoniae. As seen in the following chart, soli inhibits growth of essentially 100% of bacteria isolates at a vastly lower drug concentration than azithromycin:
Soli achieves this potency in a similar way to other macrolides: it interacts with the bacterial ribosome, disrupting protein synthesis. However, unlike other macrolides, it has three binding sites (versus, e.g., one in azithromycin), owing to soli's aminophenyl plus 1,2,3 triazole ring on the side-chain (left side) and flourine atoms near its "base".
Action has been taken on a legislative level to protect the public against deadly bacteria. Congress & former President Obama recently enacted the 21st Century Cures Act, partly to encourage development of new antibiotics. Obama also signed an executive order designating the fight against antibiotic-resistant bacteria as a national security priority, a policy that will likely continue with the current administration.
Despite this broad legislative action, it is not enough. With the rapidly accelerating risk of antibiotic-resistant bacteria, we are left to wonder: where is the FDA? Why, with so many preventable deaths and complications, is solithromycin not approved?
2) Solithromycin is easy on the stomach.
Epidemiology, Risk Factors and Treatments for Antibiotic-Associated Diarrhea describes that the frequency of antibiotic-associated diarrhea for patients in community hospitals was 15% in the US and 22% for adults; c. diff. infections account for 15% to 20% of all cases. Usually, antibiotic-associated diarrhea is an "inconvenience". However, in the hospital, immunocompromised/-suppressed individuals (e.g.: transplant patients) are at a significant and increasing risk of complications and death from the spread of germs caused by diarrhea. The CDC noted in 2013 that c. diff. infections were responsible for 250,000 illnesses & 14,000 deaths in the US yearly:
In Phase 3 studies (oral and IV-to-oral), the percent of patients with diarrhea was consistently smaller than those on moxifloxacin: pooled results show 4.3% of subjects had mild diarrhea from soli, versus 6.2% from moxi. If a proportionate number of patients lived because of soli, over 4,000 people would be saved yearly.
The level of c. diff. infection was significantly lower as well: in moxi, it was three, while in soli, it was zero. This is not a statistical anomaly: both in-vitro experiments and in-vivo trials previously identified that soli does not significantly impair bacteroides, usually beneficial & mutualistic bacteria that comprise a big part of the human gut flora.
Importantly, bacteroides keep c. difficile populations in check; this reduction of activity against them contributes to the zero c. difficile infections seen in solithromycin treatment. This difference was seen in both Phase 3 trials. In fact, solithromycin doesn't just not allow c. difficile infections from occurring: it kills most isolates.
As just described, soli, unlike many antibiotics on the market, doesn't significantly impact gastrointestinal flora [1, 2, 3]. With that in mind, there's a growing connection between antibiotic damage to gut flora & autoimmune disease. Recent research has revealed that autoimmune disorders, such as multiple sclerosis  and Crohn's disease [1, 2, 3], are correlated with a lack of certain gut flora. Many popular antibiotics on the market (like azithromycin) disrupt & kill gut flora, as shown in detail here.
3) Solithromycin does not have many of the side effects of other powerful antibiotics.
Currently available combination antibiotics for severe and drug-resistant infections cause rare side effects. Fluoroquinolones such as ciprofloxacin and levofloxacin cause tendon rupture. Powerful macrolides such as erythromycin and clarithromycin, and even the increasingly ineffective azithromycin cause Qt prolongation, inducing heart arrhythmia and sometimes death.
Qt prolongation is especially of concern for older people, who go to the hospital for CABP more frequently but are also much more susceptible to heart conditions such as arrhythmia. These side effects are imperceptible in soli.
Powerful antibiotics called fluoroquinolones (such as commonly-prescribed ciprofloxacin and moxifloxacin) have very high rates of serious arrhythmia: 15/100,000 in ciprofloxacin, 27/100,000 in levofloxacin, and 57/100,000 in moxifloxacin. As a comparison, the observed rate of serious drug-induced liver injury in telithromycin, pulled from the market by the drug maker, is 1 in over 123,000. (See this, page 160.)
In in-vitro models, soli has been shown to be "10x" safer than azithromycin and clarithromycin in terms of rare off-target liver, eye, and neurological effects, and "30x" safer than discontinued telithromycin.
4) Soli can be given in oral or IV (intravenous) form, saving money to the healthcare system and avoiding the dangerous side effects of powerful oral antibiotics.
When relatively weak drugs such as azithromycin fail, more powerful antibiotics are available. However, typically, the more powerful the antibiotic, the more severe the side effects. Fluoroquinolones, a powerful class of antibiotics, have a range of common and rare side effects (e.g.: diarrhea, or the Qt prolongation mentioned previously).
In oral-form fluoroquinolones, diarrhea is prevalent, especially in hospital mainstay ciprofloxacin. Thus, patients have a choice of either using IV, or go home and take oral antibiotics, with the associated side effects. More seriously ill patients thus tend to stay in the hospital (as you really don't want diarrhea if you're also taking a whole host of other drugs), racking up large costs to the healthcare system, if not to themselves.
Soli's effectiveness is at least comparable to that of fluoroquinolones (per Phase 3 trials); however, soli does not have the level or severity of diarrhea associated with oral fluoroquinolones. Thus, soli provides a safer and much more cost-effective treatment option.
5) Evidence of solithromycin's better efficacy in people over 75 years old.
In randomized, double-blind Phase 3 oral trials, patients with mild-to-moderate CABP infections took soli for five days, or moxifloxacin for seven days.
For patients 75+, the ECR (early clinical response) of soli was 83.9%; this was heavily favored against the comparator, moxifloxacin, at 69.8%. Although the SFU (short-term follow-up) data showed only a slight difference, that quick response to treatment, along with the 5 vs. 7-day treatment duration difference, provides a significant quality of life benefit and potentially reduces the likelihood of more serious illness between ECR and SFU. (See Note C1 for more details.)
6) "Bonus": Gonorrhea and NASH.
In Phase 2 trials, soli cleared gonorrhea after just one dose, in 100% of patients.
Ceftriaxone belongs to the cephalosporin class of antibiotics. Another cephalosporin, cefixime, which was previously used in 40% of cases 15 years ago, stopped being recommended as gonorrhea started to become resistant to it. Prevalent ceftriaxone dosage has doubled from 125mg in ~2007. It's thus only a matter of time before gonorrhea is resistant to both azithromycin and ceftriaxone in the US - worldwide, ceftriaxone-resistant gonorrhea is just emerging; resistance to azithromycin rose from .6% of isolates in 2013 to 2.5% in 2014.
Further, some people can't take ceftriaxone due to allergies (e.g.: those with penicillin allergies) so they must rely on older drugs, or azithromycin alone.
According to the CDC, the threat level is "urgent":
1) The convenience of one oral dose, leading to better compliance & fewer infections.
2) No potential side effects from azithromycin or ceftriaxone use.
3) Comparable efficacy: 90% of both soli & ceftriaxone isolates (MIC90) stopped growing at a dose of 0.12 μg/mL.
4) A different mechanism of action, providing a safety valve as gonorrhea becomes resistant to ceftriaxone.
5) In-vitro activity against many other STDs, such as Chlamydia.
The 21st Century Cures Act would have allowed off-label data to support approval for secondary indications like gonorrhea, if soli had been already approved. Fortunately, a proper Phase 3 trial is being run in this indication and is set to read out some time this year - because, unlike the FDA, Cempra values diligence.
NASH, or "non-alcoholic steatohepatitis", is liver disease which does not arise from alcohol abuse. Instead, it arises due to a combination of adverse genetics, poor diet, and obesity. NASH occurs in 3-12% of the US population. The NIH states that NASH "can lead to complications, such as cirrhosis and liver cancer. People with NASH have an increased chance of dying from liver-related causes. "
Solithromycin has been shown to reduce inflammation in the liver: it's currently being evaluated as a long-term therapy in a small study of 10 NASH patients, with NAFLD scores >= 5. Six patients have already been read out, with good results:
Compared with baseline values, after 90 days of soli treatment, all six NASH patients had a reduction in their nonalcoholic fatty liver disease (NAFLD) activity score (NAS) (mean reduction, 1.3) and their ALT (mean reduction, 17.8 U/L). Five of six patients had a reduction in their AST, and the 6th patient had an AST that was unchanged and normal (mean reduction, 10.1 U/L).
There are currently no FDA-approved treatments for NASH. Both the public health value and commercial value are high: Tobira Pharmaceuticals, recently acquired by Allergan for up to $1.7 billion dollars, failed a NASH trial in mid-2016 - with the same endpoints and same kind of patients that soli has already succeeded in.
= Section 3: Bilirubin, ALT, AST, ULN, and Mechanisms of Liver Injury =
This article extensively discusses liver function measures called bilirubin, ALT, and AST, in the context of "ULN". It discusses various means of liver injury as well. Note C3 briefly introduces these concepts.
= Section 4: Is the FDA an Effective Agency? =
The design and target of Phase 3 drug trials can often be critically evaluated to better understand the possibility of success. We can ask questions like: "Is the current trial set up the same way?", "How difficult is it to treat this disease?"; etc. Answering these doesn't obviate the need for trials, of course, but it helps spot potentially significant problems.
In contrast, predicting whether the FDA will decide to approve a drug is often, unfortunately, seemingly not a question of rational or scientific discourse. This must change if the public interest is to be served.
The FDA has acted, at times, irresponsibly in deciding on drug approval. There was a time, not too long ago, when the FDA acted irresponsibly in approving too soon or not providing restrictive enough labeling. In 2016, though, the pendulum swung the other way: the amount of applications very slightly decreased, but the number of approvals decreased by about half. One may further wonder why, in a fairly unusual string of events, some applications set to be approved in early 2017 were approved in Dec. 2016...
The FDA was set up and designed to safeguard the public from misleading or outright false drug marketing. It was not set up to act as an echo chamber, but that is what it has become. The FDA should not reject drugs based on potentially irrelevant safety concerns (as shall be discussed) if the current standards of care have worse safety profiles and efficacy. The risks of introducing a new drug to the marketplace must be balanced with the potential benefits. Increasingly, they are not.
Although many companies sometimes do a terrible job in presenting the benefit of their new drug to the FDA, that is only part of the problem and does not really change yearly. What other problems started to emerge in 2016?
1) Wages: An increasing problem is the lack of qualified and experienced personnel at the agency, due to lower salaries than the private sector. This is exacerbated when the biotech industry experiences high levels of growth, and thus an increase in wages - as had been happening for the past few years.
2) Lack of staff: Another related problem is simply the lack of staff to handle increasing amounts of NDA submissions from small companies. Despite a drug application fee intended to ameliorate such shortfalls, this fee is not collected from the first-time submissions of the small companies submitting NDAs in increasing proportions in recent memory.
3) Political pandering: An undue emphasis on safety can be accentuated not just when less-qualified people are employed at the FDA, but when less-qualified people are invited to advisory committees. This year, in a "interesting" event, multiple experts were un-invited from advisory committees due to their faint ties to industry, after Senator Ron Wyden (D-Oregon) complained to the FDA - in, unsurprisingly, an election year. This has happened before, and might happen again.
4) Complexity: Although this has been a continuing process, the increasing complexity behind the science of drug development itself requires more expertise, not less. Thus, there is a need to pay better wages to more competent and less overworked staff, and a necessity to invite experts in their fields into advisory committees, regardless of whether they have been employed in more lucrative private industry.
The structure of an advisory committee meeting exacerbates the issues above: what's usually one or even half a business day of meetings provides very little ability for the committee members to understand complex scientific and medical issues:
* FDA and company briefing documents frequently go to 150 pages or more.
* Committee members aren't typically experts in the field they are supposed to advise on. Can statisticians & pediatric doctors become liver toxicology experts in hours? Unlikely.
* There is no possibility for either the company or public to respond with thoughtful and complete answers to committee questions. Instead, answers must be quickly given verbally. Only expert public speakers can do so - but how many of them are scientists?
* The FDA's presentation may have multiple presentational issues, but the company always presents first. Thus, the FDA always gets the "last impression".
The cost of FDA inaction and organizational incompetence is borne by the public. Major safety scandals in the past have made the FDA more conservative. Individual FDA reviewers no longer bear any cost of rejecting a drug based on safety. Due to the heavy layers of bureaucracy, lack of talent, and detrimental politics from the left, a relatively minor safety concern becomes something that is almost fantastical as it is passed through the echo chambers of the FDA.
Such is the case with solithromycin.
= Section 5: The Story of Telithromycin =
Part 1: The Beginning of Telithromycin US Sales
To understand the history of solithromycin, we must know a little bit of the history of telithromycin (trade name: Ketek), a structurally similar drug - in the same way, carbon dioxide is almost identical to carbon monoxide, but has vastly different properties.
Telithromycin, a novel and potent antibiotic produced by Aventis, was developed as a response to growing antibiotic resistance to current therapies - telithromycin can kill resistant bacteria that many other antibiotics cannot.
In Phase 3 trials, there were two liver-related severe adverse events (SAEs) attributable potentially to the drug. This was discussed in an FDA advisory committee meeting in 2001 (page 26). There were some concerns about liver safety and cardiac safety. There were also concerns about drug exposure elevations in patients with renal and hepatic impairment and due to potential concomitant medications.
As a condition of approval, Aventis was compelled to run a large comparative safety trial of 24,000 patients, paying trial docs $400 a patient. Although this took ~5 months, due to various extra trials and delays, the drug was approved only in April 2004. Aventis's large resources meant the delay didn't significantly damage its investors.
After submission of more Phase 3 trials, European post-marketing data, and data from the 24,000 patient safety trial, the FDA approved telithromycin for the treatment of acute bacterial sinusitis, acute exacerbation of chronic bronchitis, and CABP. Notably, the trial was the biggest antibiotic safety trial of its kind (see Note C4).
Rare visual, muscle, & neurological side effects were seen in clinical trials, and the drug's label reflected this.
By April 2005, telithromycin had already been prescribed millions of times in the US, and 17+ million times worldwide, with only 4 reports of acute liver failure (see page 75 here). During that month, Carolinas Medical Center in Charlotte reported three cases of acute liver failure possibly related to the drug. One person quickly recovered, one required a liver transplant due to a severely atrophied liver, and one died.
In January 2006, those three cases were published in the March issue of Annals of Internal Medicine, with the article highlighted prominently in the editorial section. After publication, the number of liver failure reports to the FDA significantly increased:
In that article, a possible correlation between higher ALT elevations versus other drugs and what the authors saw as a significant amount of idiosyncratic drug toxicity was considered; however, there was no correlation, and in fact, it was later found that elevations of ALT > 3x ULN favored telithromycin (see Note 2).
There is a large number of confounding factors in liver problems; thus, doctors only report perhaps 1 in 16 events that may be tied to a particular drug. Given the tiny number of liver failure reports through September 2006 (13) versus 5.65 million prescriptions, any increase in doctors' awareness after the Jan. 2006 article could've had an immense impact on the report number. The FDA agreed with the concept of possible reporting bias, per this transcript (page 49), & accompanying presentation (page 28).
Part 2: 42 Cases of Toxicity
In late 2006, the FDA adjudicated 42 possible telithromycin toxicity cases that occurred between July 2004 and April 2006, during which 5.2 million US prescriptions were written. The results were published three years later by Brinker et al:
This agent received approval by the FDA in 2004 for treatment of upper and lower respiratory infections. Following market introduction, spontaneous reports of telithromycin-associated hepatotoxicity, including frank liver failure, were received.
Separate from whether telithromycin liver injury rates in general are actually any different than that of other antibiotics, the unusual finding of ascites in many of the more severe cases presented a pattern of liver injury unique to telithromycin. Would it have been possible to catalogue and identify the root cause of this idiosyncratic liver injury before telithromycin was put on the market? Let us consider some numbers...
Fourteen out of the 42 patients had a DILIN severity of 4+ on a 5-point scale. Thus, in Brinker et al's report, the rate of reported severe liver injury was defined as 1 in ~371,400 (14/5.2m). However, reports are not incident rates, as the FDA and others have stated. Incident rates are what we really want, but there is no way to get exact data in a post-marketing commercial setting. We can only approximate.
Using the results of a 2002 French-based population study of liver adverse events, per the FDA's suggestion, we obtain a rate of 16 actual incidents to 1 report. Given that (1) there was probably a higher rate of reporting after the Jan. 2006 Annals article and that (2) the FDA noted that European doctors report adverse events less often than US doctors, we can comfortably use the 16:1 ratio as a max. The estimate of serious DILI (drug-induced liver injury) thus becomes, at most, 1 in 371,400/16 or 1:23,212.
Let's consider a few other measures, as well. Using all 42 reports, the ratio of any drug-induced liver injury is 1:7,738. Using 13 ALF (acute liver failure) cases per 5.65 million prescriptions through September 2006, we get 1:27,163.
Without a comparator arm (to only detect DILI at the rates above) the trial size would need to have been between 23K patients (to detect minor DILI - 7,738*3) and 70K patients (to detect severe DILI). With this trial size, we would find 1 event with 95% certainty. But when we do, we'd need even more to establish a liver injury pattern.
In the 24,000 patient study, there were 12,159 patients in the telithromycin arm and 11,978 in the AMC (amoxicillin/clavulanate) arm. Just 5 patients - 3 in the telithromycin arm and 2 in the AMC arm - were judged to have DILI that was possibly-related to the drug. (see page 61-67) In other words, this trial did not definitively identify telithromycin's safety level compared to the drugs it was intended to replace.
Given what we know now about possible telithromycin DILI rates, Sanofi-Aventis would have had to have run a 58K-sized safety only trial to identify a single severe DILI event, and a separate (24K?) comparative trial to address the non-DILI side effects. Could Sanofi-Aventis have feasibly run two trials of 82K patients?
The larger the trial, the longer it takes and/or the more expensive per patient it is. Accounting for the overhead of monitoring and analyzing the trial, we could maybe estimate a $1,000 per-patient price tag. The two trials of 82K patients might have cost $82m and taken 2-3 years to design and run, at an opportunity cost of $200-$300m (if $100m yearly income). Further, there would have been no guarantee of success.
Perhaps an 82K patient trial would have been able to catch an imbalance in liver injury if one existed, or one event of idiosyncratic liver injury. It would have been completely insufficient to identify a pattern, however. This would have either required advanced predictive tools, or mass commercialization.
Is telithromycin really more dangerous than other antibiotics? The answer is: we don't really know for certain, but we can make educated guesses:
1) In the December 2006 FDA advisory committee for telithromycin, moxifloxacin was identified as 3.5 times safer based on acute liver failure reporting rates.
2) In one study, case reports show that telithromycin use increases liver toxicity risk versus other non-macrolide antibiotics by 82%, but the study also showed that macrolide antibiotics in general increase the risk by 85%.
3) Yet, in another study, Kaye et al. look at emergency department and hospital discharge claims diagnoses indicating severe liver injury, and find that, as a measure of incident rates per person years, telithromycin was the safest antibiotic out of all those tested. This included amoxicillin, moxi, etc. Telithromycin was roughly 5 times "safer" than moxi!
Spontaneous report likelihoods better identify drugs with unusual toxicity profiles, as they stand out more to doctors. However, this author would be more comfortable with emergency room visits as the best proxy for safety.
Part 3: Telithromycin US Sales Stopped
In mid-2006, Sanofi-Aventis modified the label to reflect the newly uncovered liver injury risk. Then, in late 2006, the FDA held an advisory committee meeting. In it, multiple items were considered, including the consideration by both the FDA and outside experts that reports are not incident rates.
Because the risk seemed to potentially outweigh antibiotic resistance benefit, the advisory committee voted that the labeling for telithromycin be reduced to only CABP.
In early 2007, the FDA implemented this new marketing label. The threat of antibiotic resistance was still significant enough to warrant a CABP label, but in other indications, telithromycin's efficacy was re-evaluated and found not to be sufficient to warrant the extra risks - including the liver risk and the cardiac, muscle, and neurological risk.
Despite this compromise, Sanofi-Aventis pulled the drug from the US, perhaps because of reputational damage. The rest of the world had a less restrictive response to Ketek.
= Section 6: The Background of Antibiotic Development & Resistance Between 2006 and 2012 =
The Shadow of Ketek (and Its Predecessors) Looms Large
The initial publication of rare Ketek side effects in January of 2006, an article in the NY Times (see Note 3), and finally "strong" FDA label action in early 2007 chilled antibiotic development in the United States (and even worldwide). As Spellberg shows below, the number of new approved antibiotics steadily declined since the 1983-1987 time period, and the decline continued post-Ketek.
Between 2006 & 2010, 10 new antibiotics were sent to the FDA; 7 were not approved.
Antibiotic Resistance Trends
Over the past several decades, antibiotic resistance has steadily increased worldwide. Massive surges in airplane travel rates & population density booms in India & China are major contributory factors, as well as simply the natural progression of bacterial adaptation. Per a Cempra poster presentation:
As noted previously, bacteria recently briefly emerged that was resistant to all 26 commercially available antibiotics in the US. Even in the hospital setting, this development is a frightening prospect. The urgent real need of new antibiotics thus provides incentive to develop them, even in the face of inexplicable regulatory reticence.
The GAIN Act
In 2012, the GAIN Act was passed, creating a "QIDP" incentive. This allowed Qualified Infectious Disease Products to gain an additional 5 years of market exclusivity and an expedited review of such products.
The GAIN Act incentivized antibiotic development, generating the investor confidence needed for money to flood into the market. On September 23, 2014, Janet Woodcock, the FDA's CDER director, proudly waxed about all the GAIN Act had accomplished, with the FDA approving three new antibiotic drugs in a short period of time in 2014.
In 2013, Cempra filed for its IPO, generating much-needed funds for solithromycin development, and solithromycin quickly gained QIDP status. 2013 is not 2006, and, as we shall find out, 2013 is not 2016...
= Section 7: The Early Story of Solithromycin =
The Birth of Solithromycin
It was in the 2006 regulatory context that a small company from North Carolina called Cempra began developing solithromycin for the treatment of CABP, for a 5-7 day treatment period. Solithromycin's initial development came as a result of several years of pre-clinical development and algorithmic testing of hundreds of molecular combinations, visually similar to telithromycin.
Solithromycin, although visually similar to telithromycin (although one could make this argument for other macrolides as well), differs in two significant respects.
1) The side chain (left side) is aniline-triazole, as opposed to pyridine-imidazole. Aniline-triazole is much more metabolically stable, and safer, than pyridine-imidazole.
* As early as 1987, pyridine has been found to cause problems in the liver, eye, muscle, and brain.
Telithromycin's side chain, left, versus solithromycin's, right
2) Soli contains a fluoride molecule. With its addition, soli has three sites of disabling/killing interaction with bacteria, versus two or one in older macrolide antibiotics. This gives the drug a unique ability that no other macrolide can match - not even the potent telithromycin. A mutation would need to be made simultaneously in three different physical sites for a previously vulnerable bacterium to gain soli resistance. Thus, the probability of a mutation creating soli-resistant bacteria is extremely low.
A Clinical Hold, 2008-2010
From 2008 through 2010, Cempra was unable to continue human trials, as the FDA had put a partial, and then full clinical hold. The company then performed fundamental mechanistic research to try to remove the hold.
From what I have found, the FDA had two concerns: one concern focused on a potential repeat of telithromycin toxicity, which the FDA thought might be related to an immune reaction to the drug - hypersensitivity. The other focused on the potential new toxicity of the aniline-triazole side-chain.
A quick detour then, to the question of aniline-triazole toxicity. In The solithromycin journey-It is all in the chemistry (2016), Fernandes et al. write that after extensive experimentation, soli's aniline-triazole is not metabolized in liver cells, because it is stable: "unlike imidazoles that are metabolically unstable in vivo, the 1,2,3-triazole ring has the advantage of being metabolically stable." (see Note C5)
Cempra's research then focused on the pyridine-imidazole side-chain of telithromycin, and what it might be doing. As early as 1987 (pages 25-29), researchers discovered similar issues in animals to what had been seen idiosyncratically in telithromycin. For example (see Note 4 for more details):
Pyridine exposure has been associated with hepatic effects in rats. In a 90-day study, female rats that received pyridine by gavage at dosage levels of 10 mg/kg/day and higher had significantly increased liver weights (Anderson 1987). Inflammatory hepatic lesions were found in 70% of male rats that received 50 mg/kg/day. Lesions included bile ductule proliferation, mixed peribiliary infiltrate, and enlarged vacuolated hepatocytes.
The Role of nAChRs in Telithromycin Toxicity
Cempra commissioned Dr. Daniel Bertrand to perform an in-vitro study to examine the role that pyridine-imidazole plays in the rare eye (syncope), neurological/muscle (myasthenia gravis), and liver adverse events seen in telithromycin. Dr. Bertrand saw a common theme: each of these systems had nAChRs, or nicotinic acetylcholine receptors.
Bertrand's 2010 study evaluated various macrolide inhibition levels on nAChR receptors in the eye, the neuromuscular junction in muscles, and the vagus nerve supplying the liver. The soli concentration needed for 50% inhibition (IC50) was shown to be 3x more than azithromycin/clarithromycin, & 30x more than telithromycin:
Thus, telithromycin, via its pyridine side-chain, heavily inhibited nACh receptors.
Recall that nAChRs are contained in the vagus nerve supplying the liver. Years of research (Wang et al, Gallowitsch-Puerta et al, Tracey 2009, Pavlov et al 2012) has shown that the vagus nerve, liver cells called hepatic stellate cells, and white blood cells called macrophages all contain nACh receptors. It further showed that the level of inhibition controls the release of cytokines - small proteins responsible for cell signaling. The more that nACh receptors are inhibited, the more cytokines are released from them.
Cytokines' signaling, in turn, directs certain white blood cells into the liver. Both cytokines themselves and white blood cells may damage liver cell walls, causing them to release ALT. Thus, inhibition of nAChRs causes an immune system overreaction, leading to inflammation.
As an example, this paper shows, via biopsy of human liver cells, that inhibition of nAChRs with nicotine causes liver fibrosis:
Given that we do not presently advise smokers with liver disease to refrain from smoking, our results imply that ... we should in fact be advising patients with liver disease to abandon smoking ... for fear that it may exacerbate liver fibrogenesis.
The Role of Concomitant Alcohol Use In Telithromycin Toxicity
But wait, there's more! It is not just about the nAChRs, but about pyridine and all the effects it can cause. For example, pyridine and alcohol do mix: pyridine has been implicated in increased alcohol metabolism in a 1998 animal study. In fact, pyridine was noted as a mechanism of de-naturing alcohol in 1912.
Dore et al. studied macrolide liver toxicity case reports done between January 2005 and June 2005. They found that, among macrolide antibiotics, men between 45 and 64 seemed more likely to be affected by possible telithromycin liver toxicity:
The study showed (see Note C6) that patients on telithromycin who experienced liver injury had comparably higher rates of both alcohol and Tylenol use - both drugs can instigate liver injury and an immune response.
In 2/3 Carolinas Medical Center cases, alcohol was a potential factor in inducing serious telithromycin toxicity. The Carolinas patient who died also had a nasal tumor, further augmenting his immune reaction.
Thus, nAChR-sensitive patients would experience heavy inflammation after telithromycin use, in the presence of pre-existing or ongoing liver damage.
One might ask: what is the relationship between these findings and identifiable liver injury events? Given the probable alcohol/Tylenol liver injury effect, plus the fact that many patients with severe side effects did not use either alcohol or Tylenol, nACh receptors in the vagus nerve cannot be the whole story. Are there other mechanisms of drug-induced liver injury at play with telithromycin?
Individual reports of telithromycin toxicity often manifest in several unusual and specific profile rarely found in any other antibiotic side effect. One of them is ascites, or "the accumulation of fluid in the peritoneal cavity that exceeds 25 mL" -- any fluid between the abdominal cavity and the abdominal wall. People with ascites look really, really bloated.
But, there are more. Fernandes et al. noted that "the liver toxicity observed with telithromycin was unusual in that it involved eosinophilic infiltration and inflammatory response that rapidly led to necrotic cell death."
Eosinophilic infiltration is (typically) an extremely rare condition where white blood cells called eosinophils accumulate where they should not - in the gastrointestinal tissue (GI). This process is described in here.
Simply put, inflammatory cytokines (small proteins) called IL-3 and IL-5, combined with another cytokine process in macrophages (white blood cells that eat bacteria, viruses, etc.) called GM-CSF "recruit and activate" these eosinophils to travel into the GI.
What is the outward sign of eosinophilic infiltration? Ascites. Were there any instances of ascites in the 42 cases of telithromycin injury? Yes; many. The FDA stated that:
An unexpected but striking finding among the more severe cases in this series was the presence of ascites occurring early during the course of injury in a number of cases even in instances of only moderately severe hepatic injury. This clinical signature is rare in cases of drug-induced liver injury... the presence of diffuse abdominal pain in some of the patients who developed ascites suggests the possibility of drug-induced serositis and serosal membrane transudation at the time of telithromycin-induced liver injury.
Why would telithromycin cause this? Recall those pesky nAChRs. As previously introduced, macrophages have nACh receptors, and they release cytokines when there is an injury in the body and receptors are inhibited (see Note C7), thus initiating the process of eosinophilic infiltration. Fernandes et al. further describe multiple animal experiments which confirm the mechanism of action (see Note C8).
Bertrand et al. 2010 and related research provided a plausible mechanistic rationale for the observed rare telithromycin effects, as supported by detailed experiments and numerous real-world cases.
= Section 8: FDA Briefing Document and Advisory Committee Meeting =
Fast-forward to 2016. Cempra is running small trials in COPD and NASH patients and a Phase 3 trial in gonorrhea, had just completed two Phase 3 trials in adult CABP patients, and had just begun its trial of pediatric CABP patients.
Further, Cempra had been consulting with the FDA on its trial design and helping develop antibiotic development guidance.
Phase 3 trials showed that there were some increased temporary site infusion side effects (redness, etc.) (in IV) and a slight increase in the percent of patients with rapidly-resolving asymptomatic ALT elevation versus the comparator.
Cempra found that in Phase 3 oral trials, the percent of subjects experiencing ALT elevations above 3 times ULN was 5.4%, compared to 3.6% for moxifloxacin. In Phase 3 IV-to-oral trials, it was 9.1%, again compared to 3.6% for moxi. (Ketek's label shows 1.6% versus 1.7%.)
Cempra then applied for a CABP indication for solithromycin in mid-2016.
After receiving Cempra's NDA, the FDA anti-infectives division (division head Sumathi Nambiar, 301-796-1400) became concerned that the rare side effects of telithromycin would again be seen with soli. The FDA was concerned about soli ALT numbers that were higher than the comparator, one case of jaundice in a COPD trial, and the visual similarity of soli to Ketek.
The division called for a review by FDA liver injury doctor Mark Avigan and scheduled an advisory committee meeting.
Are 9.1% and 5.4% numbers of concern?
1) We have evidence that macrolides increase "liver injury risk" by 85%.
2) The 9.1/5.4 numbers represent the highest attained ALTs at any recorded time point in the trial, including baseline numbers.
3) Serious infections, as well as any number of other risk factors, raise baseline ALTs. Telithromycin trials included mostly less vulnerable patients (those in the PORT I and PORT II risk class), while soli, indicated for mild-to-moderate CABP, included PORT II, III, and IV patients.
So, let us consider what happens when only those with normal baseline ALT readings are included. The numbers drop from 5.4% to 3.2% (oral), and 9.1% to 5.5% (IV-to-oral). In moxi patients, the number drops from 3.6% to 1.6%. In the soli IV-to-oral administration, the number could have been lower by removing the high oral "loading dose" after IV switch, without compromising efficacy, per the FDA's concurrence (see Note 5).
We should further take a keen look as to what is being compared. Telithromycin data includes baseline, short-term follow-up (about Days 12-17) and late follow-up (about Day 28-38). Soli trial data, unlike most other historical ALT data, also includes Day 4 (where elevation was highest -- see page 32) and end-of-treatment (Days 7-9 - last dose of study drug).
A further point of differentiation is that those elevations very quickly resolved. So, to make a more accurate comparison against other macrolides, one should only look at the short-term follow-up and late follow-up time points, and not at all at Day 4 data.
But, in any case, is there a correlation between ALT elevations and idiosyncratic liver injury, and what is the natural variability in these numbers?
ALT correlation to liver injury?
Causality is important. Experts working with the FDA as well as the FDA itself (see this & this) explain that simply ALT elevation in the blood is NOT a sign of liver function decrease. According to the FDA (and as discussed in Section 4), reduction of liver function only occurs if ALT elevation is combined with bilirubin elevation.
And, to make matters "worse", we cannot state that ALT elevations are responsible for all hepatocellular damage, because:
a) The leakage of ALT from liver cells only indicates how permeable the cell walls are, not the degree to which a cell's main functionality is impaired.
b) The blood plasma ALT level is not necessarily the same as the liver ALT level.
Thus, the FDA stated that "It is important to note the degree of transaminase elevation is not predictive of the degree of hepatocellular damage" (2006 telithromycin adcom briefing documents, page 68).
In the FDA's Serious Drug Induced Liver Injury presentation (page 10), Pauls & Senior show the ALT > 3x ULN percents of three different drugs, and their acute liver failure rates. There is no discernible relationship:
In other words, as we examined in the previous section, and as the FDA itself has shown: immediate, transient, and asymptomatic ALT elevation seen in many antibiotics, and especially in macrolides, has no causal relationship with idiosyncratic serious liver injury.
Well, what caused the observed ALT elevations in soli trials, then?
Causes of ALT Elevations
A mechanistic look into the cause of soli's ALT elevation was done by DILIsym, a partnership between the FDA and a majority of the largest 20 pharmaceutical companies. DILIsym software has sophisticated modeling to evaluate multiple modes of liver injury.
At Cempra's request, DILIsym modeled soli's Phase 3 trials. DILIsym Director Watkins' December 2016 presentation (page 31) shows that the simulation closely approximates the observed elevations:
(The first number (e.g. 5.4%) is the % of study patients that achieved an ALT > 3x ULN. The second number (3.2% and 5.5%) counts only those with ALT > 3x ULN and normal ALT at baseline.)
As Dr. Paul Watkins explained in the Cempra advisory committee meeting:
... As I will show you, these elevations have been mechanistically characterized at an unprecedented level for an NDA submission...
The DILIsym initiative has shown that three properties account for dose-dependent elevations in serum ALT in greater than 90 percent of the drugs that have been modeled to date. These properties are oxidative stress and mitochondrial dysfunction that can each be measured in cultured cells, and bile acid transporter inhibition measured in express transport proteins....
If requested, I would be happy to describe this modeling process further and also share with the committee data confirming the high success rate of this modeling approach with all the drugs that have been modeled in this way to date, much of which has already appeared in peer-reviewed journals.
DILIsym's modeling software predicted that soli causes a reversible ALT increase called "mitochondrial respiration inhibition": mitochondria produce ATP molecules, used as energy for cellular function. Soli's inhibition of enzymes involved in the electron transport chain causes mitochondria to temporarily produce fewer ATP molecules via ATP synthase. Thus, liver cell walls are less frequently repaired due to a smaller supply of ATP, and ALT leaks out of these weakened cells.
The ALT elevation predicted for telithromycin by DILIsym software, however, is called "bile acid transporter inhibition", which reduces bile flow. According to the DILIsym model, telithromycin ALT elevation was predicted to be near-zero, although DILIsym did not model immune reactions caused by nACh receptor inhibition. Though not included in the labeling, this is confirmed by the 24,000 patient comparative study: post-dosing ALT elevations over 3x ULN for those patients with normal baseline readings occurred in only 0.4% of subjects (page 16), compared to 0.2% in the comparator of amoxicillin-clavulanic acid (see Note 6 for more details).
In summary: neither mitochondrial respiration inhibition nor bile acid transporter inhibition is related to the "idiosyncratic" DILI seen in telithromycin.
To CABP or Not to CABP
The telithromycin rate of 1.6% of patients over ALT > 3x ULN in the Ketek label is a pooled value for all clinical trials: not just CABP trials, nor normalized for baseline-only readings. Was the rate in CAPB-only higher?
The FDA stated in 2004 that:
Analysis of liver function tests from the comparative Phase 3 CAP studies in patients who were normal at baseline showed a greater proportion of telithromycin-treated patients with low level elevations of AST and ALT (< 5x Upper Limit of Normal) relative to comparator... This pattern was not seen in non-CAP patients.
1) Soli's ALT increase is significantly dependent on dose, whereas telithromycin's is not.
2) Soli elevates ALTs via a different mechanism of action compared to telithromycin.
3) In telithromycin's 24,000 patient trial, patients who were normal at baseline had an essentially zero > 3x ULN ALT incident rate post-therapy.
4) As shown in comparative soli and telithromycin Phase 3 trials, CABP is linked to higher ALT elevations versus other types of infections.
5) Infection severity can increase baseline ALT elevations due to the immune response: soli enrolled mostly PORT II, III, & IV patients while telithromycin enrolled mostly PORT I & II patients.
6) Telithromycin's substantial nACh receptor inhibition is the likely cause of its rare and dangerous toxicity; even so, per Kaye et al. 2014, telithromycin liver safety might be overall better than most other antibiotics.
7) The measurement time points were different. Telithromycin ALTs were measured in post-therapy and late-follow-up; soli ALTs were also measured at Day 4, where the value was typically the highest, by a significant margin.
Thus, [a] there is no viable way to compare telithromycin and soli ALTs, and [b] there is no correlation between instances of high ALTs and serious liver injury levels.
Despite this, FDA medical officer Ramya Hopinath, M.D., painted a correlation between asymptomatic soli ALT and "dangerous" telithromycin toxicity, by stating in presentation slides that:
Aminotransferase signal for hepatotoxicity seen with solithromycin in the Phase 3 trials is greater than was seen with telithromycin in Phase 3 trials; telithromycin was associated with severe hepatic injury post-marketing
...and FDA liver doc Mark Avigan stated the following in FDA briefing document:
In the pooled Phase 3 clinical trials of solithromycin for CABP there was a substantially higher percentage of study subjects receiving the ketolide who developed ALT elevations greater than 3X ULN compared with subjects randomized to receive moxifloxacin [7.2% vs 3.6%].
Except, this is wrong. That number includes high baseline readings. You don't "develop" baseline readings. It should have been stated as 4.35% vs. 1.6%.
And then there was the COPD patient...
A "Spectrum of both hepatocellular and cholestatic signatures of hepatotoxicity"
The FDA briefing document stated that there was "a spectrum of both hepatocellular and cholestatic signatures of hepatotoxicity" which brings us to the other major issue: the slightly jaundiced COPD patient, and the supporting troupe of 12 other patients with varying levels of temporarily increased ALTs. (moxi had some as well, but of course it wasn't mentioned!) As we'll later see, these patients share a common thread.
As noted previously, in 2015, Cempra initiated a very small placebo-controlled COPD trial, in London. The trial involved a course of 400mg oral soli for 28 days. The rationale for running this was likely two-fold: (1) a longer azithromycin trial described in 2011 (dosing 250mg) was successful in reducing symptoms, and (2) rat and in-vitro studies ( ) showed that soli may have superior performance to that of azithromycin.
In 3/4 patients in the trial, ALT steadily increased to over 3x ULN during therapy. In the patient we are currently focused on, up to 11.9x ALT > ULN was observed. What was happening with these patients, and with our jaundiced patient in particular?
Before we continue with a description of what happened with the jaundiced patient, some more information about soli activity is in order...
Soli, as well as many other antibiotics and drugs in general, both is a substrate of and inhibits CYP3A4 enzymes and P-gps (permeability-glycoproteins). In other words, soli is metabolized, or broken down, by CYP3A4; at the same time, it is an enzyme inhibitor of CYP3A4, which prevents metabolism of both soli and other molecules. Similarly, soli both inhibits and is a substrate of P-glycoprotein, a glycoprotein that transports substances out of a cell membrane.
Part of the exclusion criteria of the COPD trial (#11) was the use of any other drugs principally metabolized by CYP3A4: competition between different drugs for CYP3A4 enzymes may limit the metabolism for one or both drugs, and, over time, cause their accumulation in the body.
The jaundiced patient initiated finasteride simultaneously with soli as part of his medication regimen... and, of course, finasteride is principally metabolized in the liver, by CYP3A4 enzymes - oops! Predictably, finasteride levels were highly elevated. The FDA stated that, "finasteride concentrations were elevated 3- to 4-fold above the reported values for a 5 mg dose." Soli concentrations in the blood, however, didn't increase at all.
The FDA noted that: "By Day 23, the patient had become mildly icteric and developed pruritus..." Or: the patient's bilirubin count increased, causing him to have slightly yellow eyes and skin and develop a slight itch.
There are at least five possible non-exclusionary explanations for this event.
Scenario 1: As Cempra and the FDA's liver expert says, this patient's symptom profile is a typical case of cholestatic hepatitis, a "'a well-recognized' adverse event tied to macrolides". It's indeed well-recognized, but the mechanism of action is, technically, completely unknown, and the incident rate is historically extremely low - estimated at less than 4 cases per 100,000 prescriptions. Thus, to have one "extremely rare" event after long-term dosing of 400mg of just 4 patients sounds like a case of really bad luck...
Scenario 2: Soli and finasteride competed for metabolism via CYP3A4. We know that finasteride levels in the blood increased 3-to-4-fold, while soli levels did not. Soli, lacking matching CYP3A4 enzymes due to finasteride competition, was excreted out of the liver... or binded to any available P-glycoprotein in liver cells.
Thus, soli may have reduced the number of available escape routes for molecules (including itself), and increased its concentration in liver cells to an amount much bigger than in the blood. There are hints of this in animal studies: soli, clarithromycin (a drug recognized to be fairly safe), and telithromycin, all with the same interactions against CYP3A4 and P-gp, showed similarly high amounts of concentrations in liver tissue at very high doses after 28 days. (see Note 8).
With a high enough concentration of solithromycin in liver cells, there is thus more of an opportunity to inhibit mitochondrial respiration and cause ALT elevation, as observed.
There was a slight decrease of bilirubin flow with ALT increase: DILIsym does predict a "minor" bile acid transporter inhibition with soli, so perhaps extreme concentrations would have started to show this effect. Further, blockage of P-gp modulation of foreign substances out of liver cells likely affects myriad other systems.
Scenario 3: A rare hypersensitivity reaction to soli may have played a role, as suggested by the FDA. Immune cells (B-cells and T-cells) could identify soli as an antigen via an IgE-mediated hypersensitivity reaction.
Since there is an extremely diverse number of antibodies that may bind to any number of antigens, extreme concentrations of soli in the liver could cause an allergic response based on a mechanistic cascade:
a) Soli concentration increases in the liver due to Scenario 2 above.
b) A single antibody may, by chance, bind weakly with soli, causing generation of memory B-cells.
c) Memory B-cells clone themselves, generating more antibodies that bind with soli.
d) Cloned b-cells undergo somatic hypermutation, with the potential of generating a more potent antibody to soli.
e) Macrophages start eating free-floating soli; T-cells may start attacking liver cells containing soli.
f) The process continues until soli is removed from the body.
Scenario 4: Could severe nACh receptor inhibition have occurred, leading to liver inflammation? NAChR inhibition driving a cytokine flood, liver inflammation, eosinophilic infiltration, and then ascites is not limited to telithromycin: it can be achieved by any macrolide, given a high-enough drug concentration.
Regarding the case of 42 telithromycin treated patients, the FDA briefing stated that:
Of note, distinct clinical features of these cases included some with a very short latency from initiation of treatment to onset of liver injury (median, 10 days; range 2 - 43 days; 4 cases had known previous telithromycin exposure) and rapid onset of fever (29%), abdominal pain (45%) and jaundice, with in some instances, reported eosinophilia (19%) and/or ascites (17%).
The COPD patient suffered a mild case of eosinophilia - an increase of eosinophils in the blood. However, in the case of ascites in telithromycin patients, it was eosinophilic infiltration into the GI tissue, not the eosinophilia itself, that was a defining characteristic of the most severe patients. It is known that eosinophils accompany allergic reactions ( ), which points to Scenario 2/3.
However, nAChR inhibition still could have played a role. As Scenario 2 describes, concentrations in liver cells, including in Kupffer cells, may have been much higher than in the blood. An immune reaction driven by nAChR inhibition could encourage the arrival of more B-cells than usual in the liver, increasing the likelihood of hypersensitivity.
Recall Nicotine induces fibrogenic changes in human liver via nicotinic acetylcholine receptors expressed on hepatic stellate cells. It shows that nicotine inhibits nAChRs and causes liver fibrosis. Any former smoker who wishes to stop smoking (or relapses) may try to use nicotine patches, which could just be yet another cumulative factor in nAChR inhibition in the body and in macrophages.
Due to pre-existing lung damage, perhaps, after nAChR inhibition and the resulting cytokine response, the resulting immune system response would be stronger than in a normal patient, resulting in more liver damage.
Scenario 5: Could it have been the finasteride after all, or some combination of finasteride and solithromycin?
Despite the FDA liver expert's assertion that: "the sponsor is correct in pointing out that in the face of its extensive post-marketing use finasteride has NOT been identified as an agent that causes idiosyncratic liver injury", I have found at least three cases outside the United States of idiosyncratic liver injury due to finasteride.
Two mild cases were reported in Spain, and one extreme case occurred in Japan. The Japanese patient had fatigue and poor appetite, and ALT 5416 IU/L was observed: 135 times the upper-limit of normal of 40 IU/L. A delayed (2 month) allergic reaction to finasteride was hypothesized to be the cause.
COPD: Other Patients
Out of four patients total, the mildly jaundiced patient and two other COPD patients had minor elevations of ALTs above 3x ULN.
1) One was taking atorvastatin, among other drugs - per the label, the regular 10-to-20-mg dose should be reduced when taking another CYP3A4 substrate, but it wasn't.
2) One was taking an asthma-attack medication called "Seretide". For the fluticasone propionate portion of the drug, the labeling instructions recommended that "concomitant treatment with ... potent CYP3A4 inhibitors should therefore be avoided unless the benefits outweigh the potentially increased risk of systemic side effects of salmeterol treatment". Day 9 ALT elevation for this patient showed normal readings. At Day 15 (the next reading), his ALT elevation was 4.1x ULN, at 165, along with a minor bilirubin and eosinophil increase in blood - did he have an asthma attack?
The COPD patient's case is very similar to multiple other cases of cholestatic hepatitis arising from macrolide use described in the literature, such as this one.
If we're to consider whether soli is more "dangerous" than telithromycin, and we look to this one mildly jaundiced COPD patient as a potential warning case of idiosyncratic liver injury that is supposed to occur at a rate of 1 in thousands, that would mean that soli is orders of magnitude more dangerous, especially since that 3 out of 4 COPD patients developed ALT elevations. But, if this were true, we would have seen huge amounts of jaundice in CABP trials.
Grapefruit juice inhibits CYP3A4 and, just like soli, makes a drug linger in the body longer than it is supposed to, increasing its potency and side effects [1, 2, 3]. In fact, the FDA advises not to drink grapefruit juice while on statin drugs. So then, why not throw in some grapefruit juice to the COPD patient's medication list, along with alcohol, nicotine, and Tylenol, while we're at it?
This study examines the effect of ALT elevation in the daily dosing of Tylenol. When taken daily for 14 days, a peak ALT of > 3x ULN was seen in 38% of subjects, and roughly 19% at 5-7 days. By comparison, the peak ALT of over 3x ULN for oral soli on a 5-day course was seen in 5.4% of subjects, or 3.2% for those that were normal at baseline. Based on this, a 5-7 day course of oral soli is safer than Tylenol!
Three of four COPD patients had elevated ALTs and took other medications that were metabolized by CYP3A4. These patients' lung injury, combined with a reduced metabolization of soli due to concomitant medication, seems to be a logical cause of their ALT elevations.
In FDA briefing documents, in almost every individual description of high ALT elevations, a concomitant drug is taken that is metabolized by CYP3A4. Thus, the minor ALT elevations seen in soli-treated patients can be well-explained by corresponding dosage elevations and concomitant use of drugs metabolized by CYP3A4.
Simply put, this patient should not have been taking this combination of drugs, especially not for such a long period of time.
The FDA noted (incorrectly) that finasteride has never been implicated in significant injury. Regardless, it is not finasteride itself that's the problem, but its competition with soli for CYP3A4 enzymes. There was no discussion about the possible effects of concomitant use of finasteride (or other CYP3A4 substrates) on ALT elevation.
With little discussion, the FDA presented ALT elevation data from this one patient, for a different indication than what was proposed, and for 23 days instead of 5-7. This severely clouded the judgment of some advisory committee members.
FDA liver expert Mark Avigan, who had previously been involved in telithromycin adcoms, analyzed the situation. He suggested that, due to the "observed" safety signals in Phase 3 trials, there could be two paths forward, depending on whether the advisory committee believed there was substantial efficacy advantage against current treatments:
(1) The first path was to increase the safety database from 924 to 12,000 patients, to rule out "clinically serious hepatocellular DILI" to less than 1/4,000. This would imply a new non-comparative trial of 11,000 patients. This may not be such a big issue if mild-to-moderate patients are enrolled - but recall that Cempra is targeting moderate-to-severe CABP patients...
If we were to take the Aventis trial as a comparison, this might take as little as, perhaps, 5-6 months to enroll, and cost $13,000,000 ($1,000 per patient, and another $2m for trial analysis). But a resubmission to the FDA, after the data is already generated, could take another 6 months, thus delaying much-needed soli for at least a year and resulting in many otherwise avoidable deaths.
(2) The second path was approval, with heavy monitoring and labeling to restrict use to 5-7 days for CABP only, per Cempra's plan all along.
During the advisory committee meeting (all links, transcript), the FDA's heavy tilt against soli due to these safety "concerns" was not matched by the same passion on Cempra's side. The telithromycin mechanism of action explained in previous sections was barely mentioned ("if you read the excellent work of so and so", was Dr. Fernandes's [CEO] answer to the question of idiosyncratic hepatotoxicity.
Of course, the role of nAChRs and macrophages in telithromycin toxicity were somewhat explained in Cempra's 173-page briefing document. But, does anyone believe that most committee members pored over every word in this document?
FDA Presentations, Then And Now
As the advisory committee meeting dragged on, with Cempra's presentation and then the FDA's presentation, curious omissions started appearing.
Of course, the FDA did not describe what it believed to be the benefits of soli compared to current treatment. They only focused on questions of efficacy and safety, but neither exists in a void - they must be compared to the standard of care.
The FDA compared telithromycin's purported liver safety against soli's liver safety. And yet, there was no word about telithromycin's nausea rates (double that of the comparator), its Qt prolongation, or high rate of eye/muscle adverse events - none of which were present with soli.
There was no discussion about soli potential to save thousands of lives as a destroyer of drug-resistant bacteria, nor any word about its prevention of the side effects of multi-drug courses. Zero FDA banter regarding soli's potent effects against c. diff. was had.
This lack of relativity is just one more example of the echo chamber within the FDA.
The 2006 telithromycin briefing documents (pages 55-57, 68, etc.) contained a table of patients who had normal readings at baseline - but there was no such table in soli's FDA briefing or presentation. And in fact, the FDA's liver expert presented the total elevations as if all patients had normal readings.
Last time, the FDA noted that "It is important to note the degree of transaminase elevation is not predictive of the degree of hepatocellular damage." This time, the FDA suggested a strong link between mild ALT elevation and level of liver damage.
And then scatter plots. Cempra & FDA presented the same data, but visually, the FDA's scatter plot looked much more ominous: the FDA included baseline elevations, but didn't state that for many patients, bilirubin levels declined on therapy. Great job, FDA!
And what was the result? Jonathan Honegger, advisory committee member, stated: "But with 7 percent risk of an ALT rise that's significant and the history of Ketek..." It wasn't 7%, however, and ALT has no relationship to Ketek's liver injury signature.
Multiple advisory committee members discussed the possible fallout from the FDA approving a drug and then have doctors inappropriately prescribe it, despite the label. However, as potential Commissioner Joseph Gulfo points out, this kind of second-guessing ("off-label uses") is not what the FDA should be doing.
Nor is it appropriate, Gulfo states, to believe that the physician and patient do not do a risk-benefit assessment: If you're currently chronically taking a CYP3A4 substrate, are young, and enjoy alcohol maybe a little too much, perhaps solithromycin isn't for you!
And so, after a few hours of Cempra and FDA presentations, questions, audience presentations, and more questions, the discombobulated advisory committee voted.
Advisory Committee Questions
There were three questions given to the advisory committee after the company, FDA, and outside experts presented their cases, and after the advisory committee discussed amongst themselves. (This meeting was publicly recorded.) Roughly, they were:
1) Is the drug efficacious? (13-0, yes)
2) Has the risk of hepatotoxicity been adequately characterized? (12-1, no)
3) And finally: Does the efficacy of solithromycin for the treatment of CABP outweigh the risks including hepatotoxicity? (7-6 yes)
Importantly, one of the "no" voters on Question 3 (Demetre Daskalakis) stated that a "no" vote was not necessarily a vote against approval; another (Michael Green) stated that if the company had a specific plan to deal with an event of serious DILI or a death in the commercial setting, he would have voted yes - but this is something that cannot be negotiated with the FDA before the advisory committee meeting!
The chairman, Lindsey Baden, Director of Clinical Research in the Division of Infectious Diseases at the Brigham and Women's Hospital in Boston, voted yes, citing increasing antibiotic resistance.
= Section 9: Analysis of Meeting Minutes and FDA CRL =
After this meeting came the meeting minutes on Dec. 20, which was inaccurate/false in its description of the meeting in several respects. The most significant problem was the statement that the advisory committee felt that solithromycin should not be used as first-line therapy for CABP, but only when other treatment alternatives were not appropriate. In fact, only Marc Scheetz expressed this opinion in the voting discussion.
Note on manufacturing: It was already expected that Cempra would receive a fairly resolvable manufacturing CRL due to issues with the supplier (see Note 9).
A week after the meeting minutes were published, Cempra received a heavy-handed CRL from the FDA, mashing together the negative and potentially expensive aspects of both pathways FDA expert Avigan described, ignoring the adcom vote, and mis-quoting the adcom entirely using the inaccurate meeting minutes as a guide:
1) The FDA stated it would require a strict post-approval pharmacovigilance requirement.
2) The FDA suggested a potentially impractical 9,000 solithromycin-exposed patient study, along with a comparator arm requirement of what is likely another 9,000 patients.
3) Regardless of the results, the FDA stated that it would restrict solithromycin use to second-line therapy.
It seems clear that whoever wrote this CRL did not attempt to decipher the science, and did not even listen to the meeting. They only read the inaccurate meeting minutes and perhaps glanced at the FDA briefing documents - maybe everyone else was on vacation?
1) Why didn't the FDA fully specify the parameters of the required safety study? For example, should the trial consist of mild-to-moderate CABP patients, despite the proposed labeling? It would make sense, given that efficacy has already been indisputably determined. The trial would enroll much faster. But then again, nothing makes sense here except for the concept of a bureaucratic echo chamber.
2) Cempra showed that nACh receptor inhibition contributes to the reported rare telithromycin side effects. But, if the FDA "isn't sure", and the purpose of the study is to exclude the possibility of telithromycin-like toxicity, 9,000 soli-treated patients would not show this: we'd need 70,000!
3) If a 9,000-soli-exposed patient study is needed to "enable exclusion of serious drug induced liver injury", given that the incidence is already well-characterized for approved antibiotics, why does the FDA desire a comparative study of 18,000 patients?
A) If the purpose of the study is to determine that the incident rate of serious DILI is below a specified value, what is the utility of a comparison arm?
B) If the purpose of the study is to assess the safety of soli compared to another drug - given that ALT elevations are asymptomatic - how is safety going to be compared?
4) Hy's Law, per FDA, outlines an assumption that 10 Hy's Law cases reflects one serious DILI event. Thus, to understand whether solithromycin-treated patients have serious DILI levels comparable to that of clarithromycin treatment (3.8 in 100,000), a safety-only trial of 7,000 patients in mild-to-moderate CABP can be run instead of an 18,000 patient trial (see Note C10).
5) Why would any company run a 9,000 (or 18,000)-patient study, if the drug is limited as a second-line treatment, making it commercially dead?
6) Given that this drug can kill bacteria other drugs can't, why would the FDA not allow second-line treatment as soon as it's available? Why wait to let people die?
7) Is there striking incompetence in this CRL? Can the FDA even do simple math? If the FDA states that a trial should be run of 9,000 solithromycin-exposed patients to reduce the risk of serious liver injury to 1/3,000, what happened to the 924 or so that are already in the safety database? Shouldn't it be 8,076?
8) The FDA stated that it would like to reduce the risk of "serious liver injury" to less than 1:3,000, but what is "serious"?
In the FDA's training document, serious liver injury is defined (page 16) as a DILI severity of 3 on a 5-point scale. A Hy's Law case is defined as 2, acute liver failure a 4, and death/transplant is 5. But in the FDA briefing document, "serious liver injury" is equated to a Hy's Law case:
First, with approximately only 1,000 study subjects treated by the sponsor with the antibiotic in a therapeutic framework so far, a lower boundary for risk for clinically serious hepatocellular liver injury or Hy's Law cases causally linked to the antibiotic that can be excluded by the rule-of-three is only approximately 1/330 (7).
Reference #7 is to (Guidance for Industry Drug-Induced Liver Injury: Premarketing Clinical Evaluation), which states:
For example, if the true incidence of severe injury is 1/10,000, and the rate of Hy's Law cases is 1/1,000, about 3,000 exposed subjects (Rule of 3) would be needed to have a 95 percent probability of observing at least one Hy's Law case in the treated population (Rosner 1995).
That reference, in turn, references Hy's law: predicting serious hepatotoxicity - a paper by Robert Temple, the man who coined "Hy's Law". In it, "severe hepatotoxicity" and "serious hepatotoxicity" are used interchangeably. Thus it is not a surprise that FDA reviewers may also view "serious" as interchangeable with "severe", when Dr. Avigan actually had two different meanings for these terms.
If FDA reviewers believed that an 8,076-patient trial could only reduce the risk of severe injury to less than 1:3,000 (clarithromycin is estimated at < 1:26,300), it would make sense to restrict soli to second-line therapy. But, per Hy's Law, the current safety database of 924 patients is already enough to reduce the risk to less than 1:3,080 - not the 1/330 (or 1/308) number that Avigan writes!
= Section 10: Conclusion =
In The role of solithromycin in the management of bacterial community-acquired pneumonia, it was written that:
Pneumonia remains one of the most severe diseases among community-acquired infections. It is associated with a high risk of morbidity and mortality, and represented the second cause of death worldwide in 2013, after ischemic heart disease and stroke.
The main question that the FDA should ask is: what is the possible risk of soli liver damage compared to the benefit gained? As I have shown, the FDA performed a risk-benefit analysis, oblivious to many facts, impervious to reason. Whether it was due to an echo chamber, politics, basic incompetence, or a mix is an open question.
I've described the current history of soli, its many benefits, and multiple reasons why soli is safe and effective for the treatment of CABP. For Cempra investors, it helps answer the question of whether a Trump review of solithromycin may yield more a fruitful result. For the public, it informs about the current FDA process.
In this final section, I'd like to take a moment to again touch on the FDA's "QIDP" program and Sarepta's Exondys 51.
This article describes the QIDP designation that Cempra requested & obtained: "in the four years since the law took effect, there has been a 'mild uptick in antibacterial drug development,' the FDA said. However, it added, 'the pipeline remains very fragile.'"
The pipeline was fragile in mid-December and became more fragile with Cempra's CRL. Was QIDP the product of a more reasonable FDA in the past, and can it be more reasonable in the very near-future?
Sarepta's drug was approved when it showed some evidence of efficacy. With the "encouragement" of many lawmakers, the Commissioner overrode the opinion of the head reviewer, who suggested the bar was "significant evidence of efficacy" - in favor of the experience of the Director of CDER, Janet Woodcock. Exondys is a success story that will benefit the public good.
But Exondys had photogenic little kids and the support of moms who knew the treatment was working for their kids, as well as moms who hoped it would work for theirs. What does Cempra have?
The history of soli is not as well known, so it has not yet had this type of chance. Bacterial pneumonia patients are not little kids, but they deserve to live, too. Sometimes they are popular actors/actresses. They can be your grandparents, or aunts/uncles, parents, and brothers/sisters. Sometimes they are your children. Sometimes they are you.
Just like with Sarepta, lawmakers & the citizens they serve can effect change. Were soli to obtain a reasonable path forward in CABP, it would align FDA's public-facing message of strengthening the "antibacterial drug pipeline" versus the current negative one it actually sends to drug makers.
The expeditious approval of solithromycin under a new FDA administration would reduce costs on the healthcare system, halt the spread of drug-resistant bacterial strains, and prevent the entirely unnecessary American deaths of otherwise treatable bacterial disease.
Disclosure: I am/we are long CEMP.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.
Additional disclosure: I'd like to thank Summer Schrader for editorial support, and TheUgly over at Stocktwits for picking up on a typo and identifying some nice references.