The FDA has issued a warning that the popular Alli diet pill may cause a severe liver injury. The move comes three years after GlaxoSmithKline began marketing Alli as an over-the-counter version of Xenical, a prescription drug marketed by Roche (
background) which, you may recall, fared poorly due to side effects such as leakages and oily discharges. Both contain the active ingredient known as orlistat.
In explaining its decision, the FDA said it identified 13 cases of severe liver injury, with 12 coming from outside the US. The one case in the US involved Alli. Two patients died from liver failure and three required transplantation, although the agency emphasized that a cause-and-effect relationship has not been established (you can read more here).
21 Comments
Any second year medical student would tell you that a non-absorbale drug is not hepatotoxic. In rapid or prolonged weight weight loss the body switches from glucose to fat metabolism. This leads to mobilization of fatty acids and accumulation of droplet fat (triglycerides) in the liver. Over time this can lead to liver failure. This is straightforward pathophysiology. Any cause-effect relationship with orlistat is junk science.
Pharmavet,
Hate to tell you but it is absorbed. However due to likely absorption and subsequent biliary excretion of metabolites formed in the gut it's very difficult to get a good handle on how much is actually absorbed but it could be upwards of 20% of the dose.
The following is from the Xenical labeling:
Based on an oral 14C-orlistat mass balance study in obese patients, two metabolites, M1 (4-member lactone ring hydrolyzed) and M3 (M1 with N-formyl leucine moiety cleaved), accounted for approximately 42% of total radioactivity in plasma. M1 and M3 have an open β-lactone ring and extremely weak lipase inhibitory activity (1000-and 2500-fold less than orlistat, respectively). In view of this low inhibitory activity and the low plasma levels at the therapeutic dose (average of 26 ng/mL and 108 ng/mL for M1 and M3.
What strikes me is that both of these as well as other metabolites that are likely absorbed have an N-formyl group. I've said in a number of other posts that N-formyl metabolites appear to be involved in the hepatotoxicity associated with fatty livers seen with Zyprexa and Saphris, and may be the basis for the SAR for Hep C drugs under development by both Vertex and Merck.
Don't think that the concentrations quoted in the labeling mean that hepatic exposure are low either. These cleavage products are low molecular weight products so molar doses may be significantly higher than the mass concentrations would make them appear. Plus there's likely significant uptake and first pass effect since they're formed in the gut so the liver likely gets signficantly greater exposure than the rest of the body.
Salmon
Wow. Thanks Salmon.
Sounds like a very serious safety signal to me. I'll never prescribe it again. Not after this news. It certainly does not sound at all like the relatively innocuous fat blocker that I thought it was. Should probably be taken off the market.
One possible caveat: 12 of the 13 cases took place outside the U.S. Was there a "bad batch"? Hopefully, GSK and FDA are looking into this...checking lot numbers.
Disclaimer: one doc's opinion...not intended as medical advice.
Agree with the comment of Salmon above that this is probably drug-related, patients on Xenical just don't lose that much weight or lose it all that quickly.
The greater incidence of cases of hepatotoxicity in Europe probably follows from the fact that the drug is much more widely used there. On a prescription count basis, US antiobesity drug sales are dominated by generic phentermine (about 1.5M patients). Phentermine has been banned in the EU because of an enhanced risk of pulmonary hypertension, as has Meridia, so Xenical is the only game in town there.
There are about 500,000 patients on Xenical in the US and EU combined. Assuming that all five of the deaths and liver transplants occurred in a single year, and that all were drug-related, the risk is about 1 per 100,000 patient years. By way of comparison, 1 in 9000 Americans die each year in auto accidents.
I'm not sure Xenical is that great of a drug, but I wouldn't call it especially dangerous.
While it is a signal for a serious safety issue, i.e. the symptoms listed by FDA are consistent with end-stage liver failure that puts someone possibly in the game for a transplant. However, I don't have enough information to decide whether to prescribe it or not.
First we need to know the denominator, i.e. what is the absolute risk. We also need to consider what other risk factors did these patients have. As pharmavet notes starvation diets may place someone at increased risk. As might alcoholism, viral hepatitis, or concommitant use of certain drugs. For example valproate or as I alluded to Zyprexa, i.e. I probably would think twice about using it in a person who gained weight from Zyprexa especially as the mechanisms for hepatotoxicity might be similar and additive.
This also likely a dose and time related toxicity.
All drugs are dangerous and we need to have a healthy respect for them. I used to follow what other people said that a drug is too risky if it kills people whereas the disease doesn't. However I've come to realize that most drugs seriously injure or kill some people, so this is almost an impossible standard to meet. Some drugs the risks are so overwhelming that the decision is easy. Others it's not so clear and that's where I believe honest public hearings might be useful.
Although I can tell you where the company is almost always going to come down. It's extremely worrisome to me when I see a company backing off pushing a drug because of the risks. For example Lilly readily agreeing to make Zyprexa a second line agent in children because of the 12% incidence of hepatotoxicity within 2 weeks.
In my mind in less clear cases of risk:benefit it's best to inform prescribers about the risks, what factors may increase those risks, how to monitor and mitigate those risks, and to inform patients so that they can decide for themselves based on what they consider a tolerable or intolerable risk.
Each one of us if we check will typically find that there are some people who won't accept risks that we would and other people who think we're worry warts for the risks we're concerned about.
Salmon
John,
Having worked on several marketed drugs that caused hepatotoxicity. I would venture that your risk estimates are likely too low.
First there were 13 cases of end-stage liver failure in the report. Of these two patients died (2/13 or 15%) and three required transplantation. With these small numbers the 15% is consistent with the 10% incidence of death associated cases consistent with Hy's Rule. We also generally state that only 10% of cases are reported in the US. However having seen what the FDA reports out and other databases I can tell you that a 10% reporting rate is high. Depending on the country in Europe and their reporting system and culture the reporting rates may be higher or lower. Even so if we use a 10% reporting rate we still get 130 cases. On top of this I can guarantee that both companies and FDA will only count pure cases, i.e. if there is any underlying risk factor the case will not be considered drug related. Thus only 10% - 20% of cases reported to companies or the FDA will be adjudicated as drug related with the risk attributed to something else, even though as I indicated these other risk factors likely put patients at greater risk. So we're now up in the 100's or even as high as over a thousand likely cases out there.
On top of this if you follow this area you can see that the FDA's hepatotoxicity expert John Senior always bends over backwards to discount drug induced hepatotoxicity. In fact a few years ago at a meeting he stated that a drug that kills 1 in 10,000 people can't be considered toxic. (Of course this is the number that hasn't been adjusted for underreporting as I've laid out. So his real number is more like 1 in 1000.
Considering all this and how things are typically downplayed and dragged out for years before we hear anything. You can read between the lines and guess that the FDA putting out a PR release is really a sign of a significant signal.
Salmon
Salmon, I had the brief opportunity to work with the late Dr. Hyman Zimmerman in the 1980's. As you know, Hy was probably the world's foremost authority on drug-related liver injury at that time. I would posit that if Hy were to review the data that you presented, he would probably suggest that orlistat-related hepatotoxicity is more likely to be idiosyncratic rather than dose and time-related, as you suggest. You had indicated in your commentary that there is variable and difficult-to-measure absorption, which I think would make it difficult to make the case for a dose and time-dependent relationship.
Idiosyncratic hepatoxicity can be even more worrisome than dose-dependent toxicity since we know nothing about the pharmacogenomics of this drug.
Sorry, I typed in pharmalot instead of pharmavet. I'm working off a computer in the library and had to manually enter my screen name.
Pharmavet
Pharmavet,
While I highly respect Dr. Zimmerman's work we know a lot more about hepatotoxicity now than we did 30 years ago.
A great many causes of hepatotoxicity that we would have called idiosyncratic years ago are now known to have specific mechanistic causes and different types of structures are associated with different or multiple types of hepatotoxicity.
I base my suspicion of dose and time related toxicity on dose and time related hepatotoxicity with drugs with similar structures. Some of those compounds also have highly variable absorption. However if you go from say 5 mg to 10 mg to 20 mg with these compounds you clearly see dose and time related toxicity in the population at large. That does not mean that there isn't variability within the population at each dose level.
In addition interindividual variability due to differences in pharmacogenomics does not negate dose related toxicity. They may simply both be contributory. For example both valproic acid and lamictal both have dose related toxicity yet they also appear to be effected by pharmacogenomic differences in CYP3A5 and epoxide hydrolase. Yet these were not known until the 1980's and 1990's after Hy Zimmerman's time. Yet in the case of valproic acid I know that in the early 80's the hepatotoxicity was considered idiosyncratic.
Salmon
Hi Salmon,
I actually think I was pretty conservative, at least in drawing a comparison to a common and familiar risk. Here's how I think of it.
We have 2 reported deaths from Xenical in 500,000 patient years vs 35,000 deaths in 300 million driving years. One in 250,000 vs. 1 in 9,000.
(I used 5 for Xenical in my previous calculation, counting the transplant cases, even though the patients did not die. Since I counted only deaths from autombobile accidents, the appropriate number to use for Xenical was 2.)
If we are going to count all 13 cases of hepatoxicity, for an apples to apples comparison we would have to count all the people with brain injuries, hospitalizations, etc, from traffic accidents. About 1 American in 1000 is admitted to the hospital each year for automobile accident related injuries.
I assumed that all of these Xenical cases occurred in a single year, which is unlikely.
I think your number for under-reporting is probably a little high. Reporting of serious unexpected (not on product label) adverse events by companies is non-discretionary and not up to their judgment regarding causation.
If we nonetheless apply your assumption that actual cases are 15 fold greater than reported,we get an annual risk of death from Xenical of about 1 per 17,000 patient years. About half as risky as driving a car.
This is not to trivialize the suffering of those who got hurt. But its the "anal leakage" that keeps me from using this particular drug, not my concerns about my liver.
Understand how you arrive at your comparison. If you want to use all injuries from accidents then for drugs like Cymbalta and Zyprexa that also causes significant hepatotoxicity you need to include suicide, cardiac death, etc..
However I think my number for underreporting is actually conservative.
Companies only have to report SAEs that they know about. (Which they don't always do).
The 10% reporting rate is the number that is used by both the FDA and Companies and is supposed to represent the cases that physicians and other health care providers simply don't report to anyone. Years ago I did a literature search on deaths due to a drug that was pulled from the market. I found 10 times as many cse reports of deaths in the literature as the FDA and the company were reporting. (The FDA published the cases so I know which ones in the literature were included by the FDA and the company). Even so we know that most people won't even take the time to write up a case report and publish. There are also many cases that are never even recognized as drug related. Thus I believe my underreporting rate is likely conservative.
Salmon
Salmon
Underreporting of SAE's is a big concern with ALL marketed drugs. The LABAs are a case in point. I am concerned that patients on LABAs who die "cardiac-related" deaths are not being reported. That is why ALL CAUSE mortality and morbidity needs to be meticulously databased by an independent entity for ALL marketed drugs. So too does SURVIVAL.
It's just that reporting is so futile. I would never bother to do it again. It wasn't until someone here explained to me that what is covered in the labeling gets treated very differently- did I realize that the system is so faulty. It doesn't seem to matter if the problem didn't exist for you before some major change in a product. Just as long as you have been warned that it could.
I don't know how someone is supposed to get around that. You are taking medications for a reason and the burden of proof falls on sick and often helpless people who don't begin to know what to do. And from personal experience, I can tell you that many doctors really just want to treat patients and not get into whatever process is involved in reporting problems themselves. Maybe they don't want to have to appear in court. In my case it became obvious that, for people with a NTI, the reformulation of the drug was a disastrous situation- but despite all of the people having new problems, how do you prove that sub-therapeutic levels or toxicities wouldn't have occurred anyway? Tough beans. Same problem with psych drugs that might cause someone to commit suicide. If they are unstable enough to need medication, whats to say the drug caused the action? The only thing surprising to me is that drug makers don't want every possible reaction under the sun on product warning labels.
Hi Salmon,
I understand your point about clinicians under-reporting SAEs, and also that some idiosyncratic drug related deaths will go undetected by spontaneous reporting. I guess in the final analysis you would have to compare deaths from all causes in treated and untreated patients, but outside of cancer indications that is going to be pretty difficult data to collect. Overall though, my impression is that it is difficult to come up with reasonable inputs to this calculation that give you a deeply troubling result.
To approach this problem form another direction, I looked up the Xenical clinical trials to see if they were sufficiently powered to provide some light on this issue. On casual inspection, it looks like they accumulated about 5000 patient years experience in the combined treatment arms. Not really enough to be useful for looking at rare events.
In the 4 year XENDOS trial they saw a reduction in the risk of progression to diabetes in obese patients (9% vs 6.2% over 4 years), which is something I had not known.
Salmon, one more minor point, then I'll drop this subject before I get too wrapped up in quibbling (if its not already too late).
From the FDA press release that Ed provided a link to, it appears to me that these 13 cases are the total reported thus far, and not a subset of "pure" cases judged to be drug-related. So I don't think John Senior's role is relevant to this discussion or that it is appropriate to multiply by a factor to correct for this judgment factor.
Well the problem is we don't have the raw data.
I've been involved with several drugs that cause hepatotoxicity and know what tends to occur. Sometimes the warning letters or advisory committee hearings dismiss the non-pure cases. In other cases they're simply not mentioned and only pure cases are reported. In every instance the companies I've been with will do everything they can to dismiss cases or provide inadequate information and followup so FDA can't make any determination at all for most cases. In any event if FDA is stating 13 cases without any qualifiers it's most likely these are very strong cases as FDA doesn't want to raise any concerns when it isn't warrented.
Salmon
Thanks, Salmon. I stand corrected. Just as an historical point, and perhaps my metabolic bias. My dad is a surgeon who performed some of the earliest jejonoilial bypass operations for morbid obesity before the more modern techniques. In addition to being a technically demanding procedure it produced a number of cases of weight loss-associated fatty liver leading to liver failure. Since this was in the pre-liver transplant era there was a high mortality rate, and the procedure was eventually abandone3d.
Hi Vet - Dieing of liver failure is a terrible way to die. Living life with a transplant liver and lifelong immunosuppresion is not exactly a picnic, either. There have to be safer ways to loss weight than risking cardiotoxicity with stimulants or risking hepatotoxicity and liver failure with lipase inhibitors. Just one doc's opinion...not intended as medical advice.
http://www.cmaj.ca/cgi/content/full/166/10/1307
Interesting. Here's another thought. Bile acids are not only reabsorbed in this region along with fats but bile acid metabolites (e.g. sulfates, glucs) are also cleaved, reabsorbed and reutilized by the liver. It's likely that a drug like orlistat not only sequesters fats but also bile acids and increases their fecal elimination. What if the fatty livers you describe are a consequence of decreased bile acid enterohepatic recirculation and the consequential metabolic effects on the liver. If something like this were occuring then orlistat wouldn't need to be absorbed in order to cause hepatotoxicity.
Salmon
http://www.tmz.com/2013/01/07/al-roker-shart-white-house-poop-fart-dateline-gastric-bypass-surgery/