Drug and Toxin-Induced Hepatic Microvesicular Steatosis
From Captain Bligh to Bacillus cereus
by Steven Curry, M.D.
Department of Medical Toxicology
Banner – University Medical Center Phoenix
Drug and toxin-induced liver diseases comprise various histologic patterns. A few examples a toxicologist may encounter include perisinusoidal fibrosis with hypervitaminosis A, peliosis hepatis from sex steroids, veno-occlusive disease from pyrrolizidine alkaloids, autoimmune hepatitis from minocycline, and centrilobular necrosis from CCl4, acetaminophen, and other agents. But what medical toxicology fellows commonly don’t understand is the common pathophysiology behind microvesicular steatosis produced by a variety of drugs and toxins (and natural diseases). Understanding this pathophysiology assists in recognizing the syndrome and its potential etiologies, anticipating its complications, and instituting the most appropriate therapy. This topic must be simplified for a blog post, and we can’t go into great detail on each drug or toxin since each of those would be a book chapter on its own. But there is a lot of interesting stuff to learn. We’ll be emphasizing pathophysiology and begin by briefly taking a look at some major events in the history of microvesicular steatosis.
Meet William Bligh, born 1754, died 1817. This is the Captain Bligh of the HMS Bounty who survived the famous mutiny.
After being exonerated by a court-martial inquiry regarding his loss of the Bounty, he was given the command of the HMS Providence in the Royal Navy from 1791 through 1793. One of his tasks was to collect plant specimens from around the Caribbean and transport them to the Royal Society’s Royal Botanic Gardens in England. When Bligh arrived in England with specimens of the ackee tree from Jamaica, members of the Society honored him by naming the ackee, Blighia sapida.
The ackee originally had been transported to Jamaica from western Africa around 1778. The ripe fruit (attached to black seeds in photo) was and remains a popular food.
But by as early as 1875 it was recognized that eating the unripened fruit produced an illness characterized by vomiting, coma, and death which was named Jamaican vomiting sickness. As medicine advanced, it became apparent that vomiting sickness was accompanied by various degrees of hypoglycemia, hyperammonemia, mild transaminase elevation, and hepatic microvesicular steatosis. Hypoglycin A and B were isolated from ackee in 1955 and subsequently linked to vomiting sickness in 1976.1
Let’s jump to New York City in 1934.
Stander and colleagues in NYC provided the first detailed report of what is now known as acute fatty liver of pregnancy. After delivery, this woman continued to deteriorate with recurrent hypoglycemia, jaundice, and renal failure before dying. This disorder most commonly appears in the third trimester of pregnancy and is characterized by jaundice, mild transaminase elevation (usually < 600 IU/L), hypoglycemia, hyperammonemia, metabolic acidosis, with potential for coma and death. Microvesicular steatosis is found on liver histology.
Our next stop is September 1951. Mark Lepper, along with Hy Zimmerman, the father of drug-induced liver injury, published a paper describing 103 patients treated with IV chlortetracycline.
Fourteen of the patients were treated with concurrent IV and oral chlortetracycline, and things did not go well. Seven developed hepatotoxicity and five died. Microscopic examination of liver tissue showed microvesicular steatosis. We had better stop and discuss this thing called microvesicular steatosis. Whatever it is, it doesn’t sound good.
Hepatic steatosis refers to fat deposited in hepatocytes. What is meant by microvesicular steatosis? It is best to begin with macrovesicular steatosis.
I won’t say much about macrovesicular steatosis except to note it is extremely common and associated with alcohol use, diabetes mellitus, obesity, and many other disorders. What I want you to see is that large fat globules fill the cytoplasm and displace the nucleus to the edge of the hepatocyte, such as below:
In contrast, in microvesicular steatosis, countless tiny vesicles of fat fill the cytoplasm, giving a foamy appearance, and tending to centralize the nucleus in the hepatocyte (arrows):
Microvesicular steatosis can appear alone or, sometimes, mixed with macrovesicular findings. Older reports occasionally missed microvesicular fat since fat stains are sometimes required to recognize the entity.
Mechanisms and Pathophysiology
When we see drug- and toxin-induced microvesicular steatosis, a variety of possible etiologies arise. Regardless of the cause, the disorder usually results from one of two mitochondrial dysfunctions:
- primary impairment of fatty acid β-oxidation; or
- primary decreased oxidative phosphorylation from impaired electron transport which, in turn, secondarily impairs fatty acid (FA) oxidation.
The purpose of this column is to get us to the point where the figure, above, makes sense. So, let’s begin with impaired FA oxidation as the primary event. We are going to gradually get to a complicated, busy figure, which will be very understandable if we create it step-by-step.
Primary Inhibition of Fatty Acid β-Oxidation
In the figure below, we have the inside of an hepatocyte with the inner mitochondrial membrane delineated by the white square. Inside the inner membrane, of course, is the matrix.
Briefly, glucose goes to pyruvate via glycolysis, which enters the mitochondrial matrix where it is decarboxylated and converted to acetyl-CoA. Acetyl-CoA, in turn, enters the tricarboxylic acid cycle, producing NADH and FADH2. Electrons from NADH and FADH2 move down the electron transport chain onto oxygen, and we generate ATP. We will simplify this figure, as shown below, keeping in mind all that it represents.
Now, we must move some fatty acids (FAs) into the mitochondrion for purposes of generating acetyl-CoA, NADH, and FADH2 for energy and some other purposes. In the cytoplasm, FAs can be converted to triglycerides for storage or can undergo transport into mitochondria. Long FAs (composing the majoriity) can’t cross the inner membrane on their own. They first are “activated” in the cytoplasm by being esterified to CoA to form acyl-CoA (“acyl” simply representing variable lengths of FAs). Then an enzyme that goes by several different names, but what we will call CPT I (carnitine palmitoyltransferase I), exchanges out the CoA for carnitine, producing an acyl-carnitine ester. Carnitine-acylcarnitine translocase in the inner membrane then moves acyl-carnitine esters into the matrix while transporting free carnitine (and other carnitine esters) out. Finally, CPT II in the mitochondrial matrix converts the acyl-carnitine ester back to an acyl-CoA ester and we are ready to begin β-oxidation.
Let’s quickly remind ourselves how this beta-oxidation occurs. We all learned this in biochemistry at one time or another. Take a look at the figure below just to get the big picture as we remove two-carbon fragments to produce acetyl-CoA, FADH2, and NADH. NADH and FADH2 can feed into the electron transport chain to power oxidative phosphorylation or into other redox reactions, and acetyl-CoA can fuel the TCA cycle and do many other things, as well.
Just recognize there are four sequential steps, ending in acetyl-CoA. The first step involves one of three isozymes (with some overlap), depending on the length of the fatty acid (very long [VL], medium [M] and short [S] chains). If the FA is ≥ C12 in length, then the next three steps occur in a large complex comprising three enzymes and named the mitochondrial trifunctional protein that resides on the inner mitochondrial membrane. If the FA is < C12 in length, then those last three steps occur via separate enzymes in the matrix. The last major point to make is that congenital deficiencies of these enzymes, including those in the trifunctional protein, are not rare. We’ll come back to that in a bit.
Acetyl-CoA sits at a very major intersection in mitochondrial biochemistry and influences several different pathways. We’ve already seen that acetyl-CoA fuels the TCA cycle and, thus, oxidative phosphorylation. But now take a look at gluconeogenesis, highlighted in blue, that I’ve added to our figure.
The enzyme, pyruvate carboxylase (P.C.) adds a CO2 to pyruvate to make oxaloacetate, fueling gluconeogenesis. Pyruvate carboxylase exists as subunits with relatively low activity. However, acetyl-CoA causes a conformational change that stabilizes four subunits as a highly active tetramer that carboxylates pyruvate.2 Thus, acetyl-CoA is required for and activates gluconeogenesis, preventing hypoglycemia.
Now we’ll add a couple more pathways, ketones in yellow (to the right) and the urea cycle in green (lower left).
Acetyl-CoA provides the acetyl groups for generation of acetoacetate, β-OH-butyrate (not a ketone) and acetone (another ketone).
Finally, acetyl-CoA combines with glutamate to form N-acetylglutamic acid (NAG), a major positive allosteric effector (activator) of the first enzyme of the urea cycle, carbamyl phosphate synthetase. Thus, acetyl-CoA turns on the urea cycle.
We have now arrived. Let’s take a person who might suffer from a primary impairment of FA oxidation. The prototypical patient would be a child with congenital medium-chain acyl-CoA dehydrogenase (MCAD) deficiency. Most all pediatricians know this disorder, but the first presentation can be in adults. Many states, including Arizona, screen all newborns for genetic disorders of FA oxidation. A child with MCAD deficiency may do just fine as long as they are eating regularly and maintaining acetyl-CoA levels within an acceptable range with oral carbohydrate intake. But now, for some reason, they may begin fasting. It might be viral gastroenteritis or even elective surgery. Glycogen stores are depleted in a few hours, and the liver relies on burning FAs to maintain acetyl-CoA levels. But with MCAD (or some other) deficiency, acetyl-CoA levels fall. And the potential consequences of low acetyl-CoA levels are, all or in part, and in no particular order:
- metabolic acidosis from decreased ATP formation via the TCA cycle and oxidative phosphorylation.
- transaminase elevation, usually no more than 2-3 times upper limit of normal, and almost always < 600 IU/L.
- hypoglycemia from impaired gluconeogenesis.
- hyperammonemia from an impaired urea cycle (which also needs ATP).
- decreased or absent serum and urine ketones, despite catabolism.
- secondary carnitine deficiency from urinary excretion of incompletely oxidized FAs as acyl-carnitine esters. This compounds impairment of FA oxidation.
And because FA oxidation is impaired, cytoplasmic FAs are converted to triglycerides which accumulate to produce microvesicular steatosis.
Now it all makes sense. And those of you who were practicing in the 1970s and 1980s will recall Reye syndrome in children, which looked just like this, most commonly reported in association with varicella, influenza or other viral illnesses. In Reye’s original description3, he noted that the illness and hepatic histology of the children he described were similar to reports of patients who died from Jamaican vomiting sickness. The reason we rarely hear of Reye syndrome today is that we discovered most, if not all, of these patients, actually had disorders of FA oxidation that are now recognized with appropriate diagnostic studies. This was nicely pointed out by Orlowski in a 1999 paper.4
We also now recognize that many, but not all, women with fatty liver of pregnancy carry a child with a recognized impairment of FA oxidation and/or have such an impairment, themselves, with LCHAD (long chain hydroxyacyl-CoA dehydrogenase) deficiency being among the most common. We don’t have time today to go into what is understood of the pathophysiology of how a gravid woman is affected by her in utero child.
But with this background, we need to take a look at ackee fruit, tetracyclines, and some other agents to understand how fatty acid oxidation is becoming impaired and learn what other drugs and toxins can produce the syndrome of microvesicular steatosis.
Hypoglycin A in unripened ackee undergoes conversion to MCPA-CoA which ties up CoA, inhibits the transporter responsible for moving acyl-carnitines into the mitochondria, and also inhibits enzymes of mitochondrial FA oxidation, among other possible actions. Many of you may recall that lychee (litchi) fruit has recently been described as producing hypoglycemia and an identical illness in India and other areas.5
Hypoglycin A and methylene cyclopropyl glycine have been isolated from the aril of lychees, and metabolites of these compounds have been found in the urine of affected hypoglycemic children.6,7
The mechanism by which tetracyclines inhibit β-oxidation of FAs is not completely understood, but microvesicular steatosis is most common after IV therapy.8 That inhibition can be observed with in vitro assays demonstrates that such inhibition does not rely on tetracycline’s possible effect on mitochondrial protein synthesis. Some NSAIDs have also been associated with microvesicular steatosis and impair FA oxidation in vitro.
When valproic acid is administered to young children, an illness uncommonly develops that would have been called Reye syndrome in days past. This can be more rarely seen in adults, too. This syndrome is apart from the isolated hyperammonemia resulting from valproate therapy or acute overdose in patients of all ages (perhaps a later topic). More than 40% of valproate undergoes β-oxidation since valproate is a branched-chain FA. A small fraction is oxidized by CYP450. There are numerous actions of valproate on FA oxidation and hepatic function, and I’ll just offer some examples. Through β-oxidation, valproate competes for activation and esterification to Co-A to form valproyl-CoA ester, sequestering CoA.9 Valproate metabolites are excreted as carnitine esters, leading to secondary carnitine deficiency, which further impairs FA oxidation. There is some evidence that valproyl-CoA and various ester metabolites may directly inhibit the first enzymes of FA oxidation (e.g., SCAD, MCAD) as well as the trifunctional protein. In vitro data also support valproate’s inhibition of CPT I.
Every toxicology fellow knows that salicylate uncouples oxidative phosphorylation. Electron transport and oxygen consumption accelerate while ATP production falls. However, about 80% of children who die from salicylate poisoning display microvesicular steatosis.10 Fraser performed fat stains on liver specimens obtained at autopsy from 17 subjects without a history of liver disease to estimate the prevalence of microvesicular steatosis. Four of 17 patients had extensive microvesicular findings, and all four of these patients had been using salicylates.11 Don’t forget the historical concern that use of aspirin might worsen or even cause Reye syndrome (though epidemiology did not support this very well). It turns out that salicylate, in part, is conjugated to CoA, which may contribute to impaired FA activation. Long chain FAs appear to be particularly affected in animal models.12
Primary Inhibition of Oxidative Phosphorylation
Apart from primary inhibition of mitochondrial FA β-oxidation, various agents that inhibit the electron transport chain and, thus, oxidative phosphorylation, will secondarily inhibit β-oxidation. Agents that act quickly to shut down electron transport and oxidative phosphorylation (e.g., cyanide, sulfide) can produce death very quickly, long before evidence of microvesicular steatosis could appear. But other agents that produce inhibition of electron transport over hours to months can cause microvesicular steatosis.
Let’s look at oxidative phosphorylation briefly, like we did in the last Fellow Friday post on cyanomythology. Here we will focus on the NADH/NAD+ ratio rather than cytochrome oxidase.
If we take pure NADH and place it in some water, it will quickly release electrons to form NAD+ until it reaches its desired equilibrium or mass action ratio. (We will leave the H+ out of the equilibrium for simplicity.)
NADH ←→ NAD+
This is when NADH is happiest and when the equilibrium is at its lowest energy state. In the TCA cycle in the matrix, NAD+ is converted to NADH, which raises the NADH/NAD+ ratio, moving it away from its desired state of equilibrium. That is how much of the energy released from the oxidation of carbohydrates is stored (we also raise FADH2 / FAD+ ratios). NADH really wants to give up a couple electrons to form NAD+ and reestablish its desired equilibrium mass action ratio. It does so by unloading electrons onto complex I of the electron transport chain. As those electrons move down the chain and eventually onto oxygen, the energy released is used to create a proton gradient, with a lower pH (higher proton concentration) and more positive charge (voltage gradient) in the intermembrane space than in the matrix. A typical voltage potential across the inner membrane is about -180 mV (matrix negative). In fact, at steady state, the amount of energy found in the increased NADH/NAD+ ratio is about equal to the energy stored across the inner mitochondrial membrane as voltage and proton concentration (pH) gradients. Those protons re-enter the matrix by traveling through a pore in ATP synthase, and the released energy is used to make ATP.
If electron transport is impaired, NADH can’t unload electrons onto complex I since complex I remains reduced and can’t send its electrons down the line. As long as the cell is still alive, and the TCA cycle can function reasonably well, the TCA cycle will sustain an elevated NADH/NAD+ ratio that would be used to power oxidative phosphorylation. There are a couple consequences of this sustained NADH/NAD+ ratio in the face of impaired electron transport. Some electrons will be unloaded onto oxygen to create reactive oxygen species (ROS), explaining increased oxidant stress resulting from impaired electron transport. These electrons most commonly originate from complex I, ubiquinone (Q), or complex III, depending on where the block in electron transport occurs. But what is important to us today is that the elevated NADH/NAD+ ratio will impair FA β-oxidation as shown below:
What are some examples of agents that impair electron transport primarily and then secondarily inhibit FA oxidation to produce microvesicular steatosis? The first that come to mind are nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs), which are used to treat HIV and hepatitis B infections. Each one of our hepatocytes (and other cells) has roughly 500 to 2000 mitochondria, and each mitochondrion contains an average of 5 circular mDNA molecules. Thus, we have about 2,500 to 10,000 circular mDNA strands per cell. This DNA codes for 13 peptides that contribute to complexes I, III and IV along with the tRNAs and rRNAs needed to produce them. Every time one of our hepatocytes divides, about 2,500 to 10,000 circular strands of mDNA must be replicated, and that is accomplished by DNA polymerase-γ. Reverse transcriptases found in HIV and hepatitis B viruses are themselves, DNA polymerases. We shouldn’t be surprised, then, that NRTIs might inhibit some of our DNA polymerase-γ. Patients taking NRTIs can experience impaired mDNA synthesis which leads to decreased mitochondrial DNA content and decreased complexes I, III and IV.13 This explains most side effects associated with these drugs, including lactic acidosis, peripheral neuropathy, pancreatitis, and microvesicular steatosis. In 2001, Coghlan and colleagues reported on 12 patients with symptomatic lactic acidosis while receiving NRTIs. AST values ranged from 41 to 1455 IU/L; 11 of 12 patients had ASTs < 600 IU/L.14 Microvesicular with some macrovesicular steatosis was found on histopathologic examination of livers. The degree to which DNA polymerase-γ is inhibited varies with the particular NRTI. An early inhibitor, ddI, was quite famous for microvesicular steatosis and metabolic acidosis, and I saw deaths as a result. Agents used most commonly today are less likely to produce such a severe syndrome, but it is still possible. In situations in which the primary event is impaired electron transport, metabolic acidosis is commonly more of a clinical feature than hypoglycemia, probably because of the more severe impairment of oxidative phosphorylation.
Next to mind is the antibiotic, linezolid, an inhibitor of bacterial translation, and thus, protein synthesis.
If we look at animals receiving linezolid doses equivalent to those used in human beings, a decline in quantities of complexes I, III, and IV are found on immunostaining of the inner mitochondrial membrane. Such changes have also been found in human liver (and other organs) following prolonged linezolid therapy. This results from linezolid’s inhibition of mitochondrial protein synthesis.15–17 Proteins of complex II are coded entirely by nuclear DNA and, thus, are unaffected. Impaired oxidative phosphorylation is thought to explain most adverse effects from linezolid, as well, including lactic acidosis, visual impairment, pancreatitis, renal failure, and microvesicular steatosis. Hypoglycemia from linezolid has been reported repeatedly, commonly with lactic acidosis.18,19 We should not be surprised.
B. cereus exotoxins produce a common form of food poisoning, mainly characterized by gastroenteritis. On occasion, though, tragedy strikes. Consider this case reported in 1997.20 A father and son ate spaghetti and pesto that was 4-days-old. Gastroenteritis developed within 30 minutes. The 17-year-old boy worsened over two days with lethargy and jaundice. AST was 2,140, ALT 5,270, and pH 7.27. He died on day 3, and liver displayed microvesicular steatosis and necrosis. What was this all about?
B. cereus produces a couple different exotoxins. Enterotoxin primarily produces diarrhea. The second toxin, cereulide, is responsible for much of the emesis, possibly through interactions at 5-HT3 receptors.21 And when cereulide levels are high enough, hepatotoxicity (and other organ dysfunction) and death result from mitochondrial actions.
Cereulide is a cyclic peptide similar in structure and action to the mitochondrial toxin, valinomycin. Cereulide inserts itself into the inner membrane and behaves as though it were an ion channel that allows K+ to move down its electrochemical gradient into the matrix. There is an H+/K+ exchanger in the inner membrane to move K+ back out, but K+ influx via cereulide can exceed its capacity. Cereulide is an uncoupler of oxidative phosphorylation. By allowing positively charged K+ to move into the matrix, the voltage potential across the inner membrane decreases (moves towards zero), decreasing an important component of the driving force required to move protons through ATP synthase, thus decreasing ATP production. Revving up the H+/K+ exchanger also allows proton influx apart from ATP synthase. And the drop in membrane potential makes it a lot easier for complexes I, III and IV to pump protons out of the matrix (decreased voltage and proton concentration gradient to fight against), accelerating electron transport and oxygen consumption. As you would anticipate, this would decrease the NADH/NAD+ ratio. Not what we would expect from a toxin that produces microvesicular steatosis.
But cereulide has another action best characterized in experiments with valinomycin that may explain the microvesicular steatosis that occurs. At low to moderate concentrations, the influx of K+ alkalinizes the matrix.22 Matrix pH rises because K+ influx via cereulide is occurring while protons are being pumped out via the electron transport chain. And alkalization of the matrix, it turns out, decreases electron transport and raises the NADH/NAD+ ratio. The ability of alkalization to decrease electron transport is believed, at least in part, to be secondary to pH-mediated structural changes in complex I. Others have reported stabilization of ubiquinone (Q) when pH rises, preventing it from shuttling electrons, as well.23 So, as cereulide first enters mitochondria, electron transport is inhibited, and the elevated NADH/NAD+ ratio will impair FA oxidation and produce microvesicular steatosis. As cereulide levels continue to rise over time, uncoupling wins out over matrix alkalization and becomes the dominant effect. Mitochondria can swell and rupture, causing hepatocellular necrosis, producing transaminase levels higher than those typically seen from other causes of microvesicular steatosis. Both cereulide and salicylate uncouple, and through such action, will lower NADH/NAD+ ratios. But additional actions they possess explain their production of microvesicular steatosis.
Let’s finish up by examining this figure again.
When the primary event is impaired oxidative phosphorylation, we should not be surprised that the drop in ATP production may be greater, more commonly producing a significant lactic acidosis and even occasional significant hepatocellular necrosis. But the overlap between the two major mechanisms of microvesicular steatosis is significant, and mild transaminase elevation, hypoglycemia, acidosis, low ketones, and hyperammonemia can be seen in either situation. We have mainly discussed mitochondrial events in the liver. But in congenital disorders of FA oxidation and drug- and toxin-induced microvesicular steatosis, other organs can be affected, as well (e.g., cardiomyopathy, rhabdomyolysis). No time for that today.
I’ll offer a couple of points on therapy, not the major topic of this post. The general principles of treatment mainly comprise stopping the offending agent, giving full caloric support in the form of glucose to maintain normal acetyl-CoA concentrations without the need for FA oxidation, providing supplemental levocarnitine to correct secondary carnitine deficiency, and treating for hyperammonemia (e.g. carbohydrates, hemodialysis, scavenging agents, etc.) and cerebral edema, as required.
Please provide some feedback on this post so as to provide some direction for the future.
This topic was meant both to educate and to get you thinking. For example, here are some thoughts you might discuss with your colleagues and faculty members.
- Should we be giving large doses of glucose and levocarnitine to patients with serious salicylate poisoning?
- We know salicylate produces hypoglycemia by markedly increasing glucose consumption while inhibiting ALT to impair gluconeogenesis. But is part of salicylate’s ability to produce hypoglycemia secondary to low acetyl-CoA levels from impaired FA oxidation?
- Aren’t salicylate-poisoned patients very ketotic, unlike patients with low acetyl-CoA levels, or would ketones be even higher if not for impaired FA oxidation?
- There are certainly reports of chronic salicylate therapy producing hyperammonemia (and hypoglycemia). Should we be checking ammonia levels in encephalopathic patients with salicylate poisoning?
No results from randomized controlled trials provide answers to these questions. But levocarnitine therapy is amazingly benign when given for other causes of impaired FA oxidation, and administration of glucose, alone, increases survival in rodent models of salicylate toxicity, whatever that’s worth.
Finally, if you happen to visit London, stop by and pause at William Bligh’s tomb in the Lambeth district. He was buried at St. Mary’s, which is now the Garden Museum.
Admiral Bligh would feel proud that you knew of his adventures and accomplishments, but would probably be surprised that your main interest centers on those ackee plants he delivered to England more than 200 years ago.