Role of Liver Histopathology in Drug-Induced Liver Injury David E. Kleiner, M.D., Ph.D. Laboratory of Pathology/NCI FDA-PhRMA-AASLD Drug-Induced Liver Injury Workshop March, 2010
Transcript
Role of Liver Histopathology in Drug-Induced Liver Injury
David E. Kleiner, M.D., Ph.D.Laboratory of Pathology/NCI
I was asked to comment on the role of liver histopathology in drug-induced liver injury. To me, a liver biopsy is not like sending a glucose off to the lab and you get back a number. Liver biopsy can be a sophisticated medical consultation where you ask me, a trained medical professional in a different subspecialty, to comment on the findings in tissue for your patient. And particularly in the context of drug-induced liver injury, I think that the liver biopsy can provide a perspective that sometimes helps to sort out difficult cases. When I am called upon to do this in my regular clinical practice at the NIH, I review the history, I look at the laboratory values, I look at the reports and sometimes the actual scans. I sit down with the clinicians, and we talk about the case.
Liver Biopsies and DILI
• The utility of the liver biopsy, beyond its use as a sophisticated diagnostic test, has not been defined in DILI– What information can the liver biopsy
provide?– How do findings on liver biopsy correlate with
the biochemical presentation?– Can the liver biopsy help predict outcome?
Presenter
Presentation Notes
I interpret my findings in the context of all of this information and try to provide them with the best interpretation of the liver biopsy findings that I can. My role in DILIN is a little bit different because I am reviewing things blindly and don't really participate in the causality process. The local pathologists, be they at the university centers or at a community hospital, have already done this initial phase of consulting with the clinicians on their cases. So, it is a little bit of a different situation. Now, beyond what I just told you, beyond the sophisticated medical consultation, use the liver biopsy, the actual utility, how often it provides extra information that is useful in narrowing down the diagnosis, sort of what percentage of the time it is an effective test, that has really not been well-defined. So, it is useful to review what information the liver biopsy can provide and, then, as part of this prospective evaluation within DILIN, I have had a chance to look at how the liver biopsy findings correlate with biochemical presentation and whether or not there are findings on the biopsy that can help predict outcome.
An Illustrative Case• 42 y.o. male with
chronic hepatitis B• Enrolled in a 6 month
trial of a promising anti-HBV drug Fialuridine
• Previously participated in a 4 week trial (10 months prior to current trial) of the same drug, no symptoms of toxicity noted
Presenter
Presentation Notes
So, I am going to start with a case, too. This was one of my first exposures to drug-induced liver injury. It happened quite a long time ago when I was actually kind of just out of my residency. But this was a 42-year-old male with chronic hepatitis B. He was enrolled in a clinical trial that was being carried out at the NIH, the drug called Fialuridine. You probably are all aware of the story behind this drug. He had previously participated in a four-week trial the summer before, and now he was enrolled in this six-month trial. This is a picture of his pre-treatment liver biopsy. It shows typical findings of chronic hepatitis B, portal inflammation, interface hepatitis, et cetera.
• At week 11, the drug was stopped due to symptoms of neurotoxicity
• Three weeks later complained of lethargy, weakness, anorexia, myalgias and nausea
So, at week 11, though, he developed symptoms of neurotoxicity and the drug was stopped. You can see his clinical course in the diagram below. Three weeks after the drug had stopped, he developed other symptoms: lethargy, weakness, anorexia, myalgias, and nausea. His laboratory values at the time are shown. He was somewhat jaundice. He had a prolonged PT, low albumin, elevated ammonia level, and elevated lactate level.
Explanted Liver
Presenter
Presentation Notes
He underwent a transplantation. We got the tissue back at the NIH almost the same day for evaluation. This was what the explanted liver showed. From low magnification, it doesn't look too different from normal except that the liver cells look very pale. You can still see portal inflammation, even at this magnification, that would have gone along with his underlying chronic hepatitis B.
Presenter
Presentation Notes
You can see it here, a little lymphoid aggregate in the portal area, again, probably part of the underlying disease. But the hepatocytes are very pale. As you look very closely at them, what you see is microvesicular steatosis. They aren't dead, but they're metabolically-crippled.
Presenter
Presentation Notes
The overall pattern was microvesicular steatosis superimposed on a pre-existing liver injury pattern, that of chronic hepatitis B. It was not a pattern of fulminant hepatitis. There wasn't massive necrosis. There wasn't accelerated progression of the underlying liver disease. Microvesicular steatosis, though, is frequently the result of mitochondrial injury from drugs and toxins. And in particular, at the time, it had been reported in rare instances with ddI and AZT. It is also seen in some other liver disorders.
What did we learn from the pathology?
• Overall pattern: Microvesicular steatosis superimposed on pre-existing chronic hepatitis B
• Not a pattern of fulminant hepatitis, progression of underlying liver disease or confluent necrosis
• Microvesicular steatosis is frequently the result of mitochondrial injury from drugs and toxins (notably ddI and AZT), also seen in rare disorders like acute fatty liver of pregnancy, alcoholic foamy degeneration
Pathology confirmed DILI, excluded competing injury patterns, accounted for underlying disease and provided a
clue to the mechanism of injury
Presenter
Presentation Notes
What did the pathology tell us? It confirmed that there was, in fact, drug-induced liver injury. It could be nothing else. It excluded competing injury patterns. It accounted for the underlying disease and provided clues to the mechanism of injury.
Role of Liver Biopsy in DILI• Characterize the morphologic changes
– Morphologic changes may confirm drug injury by matching known/reported patterns
– Morphologic changes may suggest mechanism of injury
• Assess the degree of injury• Rule out other causes of hepatic injury• May help to make diagnosis of DILI in complex
cases by careful clinical-pathological correlation• Sometimes biopsy can exclude DILI, permitting
continued use of a necessary drug
Presenter
Presentation Notes
So, these are all of the things that a liver biopsy can provide. You can characterize morphologic changes, which can help you confirm drug injury by matching to known patterns in the literature. It can suggest mechanisms of injury, as it did in the case of Fialuridine. You can assess the degree of injury. Is there a lot of necrosis? Is there a little necrosis? Is there fibrosis? Is there a lot of inflammation? How much cholestatis? Is there vanishing bile duct syndrome? You can rule out other causes of hepatic injury. It may help to make the diagnosis of DILI in complex cases by careful clinical pathologic correlation. This is where the consultant part of this comes in because I can help sort out those patterns. Sometimes the biopsy can exclude DILI. Sometimes it can tell you that the injury is not so bad and you can continue using it. For a long time, the liver biopsy was used like this in terms of methotrexate use.
Morphologic Change Toxic Mechanism(s)
Microvesicular SteatosisInhibition of mitochondrial β–oxidation; hepatocellular energy shortage via mitochondrial injury
Zonal necrosisMetabolite toxicity associated with P450 enzyme gradients, oxidative status
Hepatocellular cholestasis Inhibition of bile transport proteins
Duct destructionConcentration of toxic metabolites in bile; injury to peribiliary vascular plexus
Here are some morphologic changes and the toxic mechanisms that have been associated with them. It is not a complete list, and this isn't meant to be an absolute one-to-one correlation, but here are some examples. So, microvesicular steatosis, as I mentioned, is associated with mitochondrial injury. Zonal necrosis can be associated with metabolic toxicity that is based on enzyme gradients within the liver or the oxidative status of the individual cells. Cholestatis can sometimes indicate inhibition of bile transport proteins. Duct destruction can be linked to concentrating toxic metabolites in the bile duct lumen or injury to the peribiliary vascular plexus. Then, sinusoidal obstruction syndrome has been linked to endothelial cell damage.
Biopsy Evaluation in DILIN
• Blinded histologic review by a single hepatopathologist
• Classification of the overall disease pattern into one of eighteen categories
• Semi-quantitative evaluation of 50 individual features
Presenter
Presentation Notes
Now the biopsy evaluation in DILIN is a blinded histopathologic review by me. What I try to do is organize the visual information on the biopsy into something that can be put into a database. So, again, unlike a pathology report, which is a consultation, this is more looking at the biopsy, seeing what is there, and putting it in categories and boxes, and trying to make sense of it that way, so that this information can be put and incorporated into the database.
Hyperplasia• Unclassifiable mixed• Minimal changes• Absolutely normal
Presenter
Presentation Notes
So, I try to classify a pattern of injury, which is kind of a summation of everything that we see. Then, I look at a whole host of individual features. This is necessary because of the complexity of liver injury in DILI. These are the liver injury patterns that I look for. There are 18 of them, which is a lot, but that is because DILI can mimic almost every pattern that there is in liver disease. These are organized according to a scheme that Dr. Zimmerman came up with and published in his book, generally following that scheme. But they also match to patterns of liver injury produced by known other non-DILI diseases. So, the idea is that, if you have a chronic hepatitic pattern, it doesn't mean that you have chronic hepatitis in the sense of viral hepatitis, but it means that it looks like something that could be caused by, say, chronic hepatitis C or autoimmune chronic hepatitis, that sort of thing.
• Special stains (5) - Iron (hepatocellular, REC), Copper, PAS• Size (No. of portal areas)
Presenter
Presentation Notes
This is the list of histologic features that I look at and record for the database. It is a very long list. It incorporates almost every aspect of liver pathology that I could think of that I thought I could reproducibly record. So, inflammation, fibrosis, steatosis, cholestatic and duct injury, hepatocellular injury which is non-inflammatory, vascular injury, and some miscellaneous findings, evaluation of special stains, and then kind of a size evaluation just for sort of quality control purposes.
• DDx: Acute Viral or Autoimmune Hepatitis, Early chronic hepatitis or PBC, Non-specific reactive changes
• Ex: Isoniazid, sulfamides, rifampin
Presenter
Presentation Notes
So, here are some of the patterns. This is an acute hepatitic injury pattern. These are lobular-predominant inflammatory injuries, usually lymphomas-plasmacytic infiltrate. There is often lobular disarray. And at least the way I categorize things, I don't include cases that have cholestatis in this category. This is a case of atomoxetine injury from DILIN and actually one that could be mimicked by autoimmune hepatitis. There's an abundance of plasma cells here, and that would be something I would come back to the clinician and say you had better make sure this patient doesn't have standard autoimmune hepatitis.
Chronic Hepatitic Injury(DILIN case – Likely Nitrofurantoin injury)
• Portal predominant, interface hepatitis, portal-based fibrosis, no cholestasis
• DDx: Chronic viral or autoimmune hepatitis, early PBC/PSC
• Isoniazid, minocycline, methyldopa
Presenter
Presentation Notes
This is a chronic hepatitic type of injury from nitrofurantoin. It was called likely nitrofurantoin injury in the causality assessment. You can see fibrosis and chronic inflammation, which is mostly in the portal areas with interface hepatitis. And again, this could mimic a severe case of viral hepatitis or autoimmune hepatitis. Early PBC and PSC can also look like chronic hepatitis.
Acute CholestaticDILIN Case – Probable Azithromycin injury
• Pure hepatocellular or canalicular cholestasis, mild injury and inflammation, mild portal changes
This is an acute cholestatic injury, from probably azithromycin injury. This is basically kind of a bland cholestatis. The upper panel shows kind of, well, a zonal cholestatic injury, and the lower panel shows that the cells are swollen and ballooned, and there's bile present. There are a few lymphocytes, but it is not really inflammatory.
Chronic CholestaticDILIN case – Likely Cefuroxime injury
• Duct injury/paucity with cholate stasis, copper accum, fibrosis, may have chronic hep changes
This is a chronic cholestatic injury from Cefuroxime. There's a lot of fibrosis in this case and a lot of copper accumulation, at least periportal fibrosis with ductular reaction, wasn't really bridging. In this particular case, the ducts were preserved, but there was a lot of evidence of chronic cholestatic change.
Mixed Hepatitic-cholestatic injury(DILIN Case – Likely Sevoflurane injury)
• Combination of hepatitis (usually acute) with canalicular/ hepatocellular cholestasis, duct injury
Mixed hepatitic cholestatic injuries are injuries, cholestatic hepatitis where you have both bile present and inflammatory injury. This was from a case of Sevoflurane injury. The inflammation is relatively mild and mostly in the portal areas. Then, there is cholestatis in the lower picture.
Zonal Necrosis(Acetaminophen Injury)
• Coagulative/confluent necrosis and/or hepatocyte drop-out in a zonal or pan-acinar pattern with little inflammation
• Hypoxic-ischemic injury, shock
• Acetaminophen
Presenter
Presentation Notes
Zonal necrosis is very common, it is the pattern for acetaminophen injury. This is a case of acetaminophen that I had. It is a coagulative or confluent necrosis with hepatocyte dropout around zone 3 mainly and steatosis around the edges. There's little inflammation here.
Relationship of Biopsy Findings to Biochemical Tests and
Presentation
Presenter
Presentation Notes
So, one of the things I looked at was the relationship of the biopsy findings to biochemical tests and the presentation, biochemical presentation. The issue here was whether or not the biochemically-defined R injury pattern was reflective of the histologic injury pattern. It is typical for publications that don't have liver biopsies to use pathology terms like hepatitis or necrosis. You really don't know if those are the things that are going on unless you have a liver biopsy. You can't assume from a transaminase elevation that you know that you've got inflammation or necrosis, or whatever.
Issues
• Is the biochemically defined injury “R” reflective of the histological injury pattern?– Typical for publications to equate elevated
transaminases to “hepatitis” or “necrosis” etc.• “R” is typically defined at the “onset” of
injury—is there another time point which would be a better comparator?
• How do individual histologic findings relate to serum enzyme levels?
Presenter
Presentation Notes
In typical use, R is defined at the onset of injury. There might be a better time, though, that might match with the liver biopsy.
Hypothetical DILI Enzyme Profile
0
5
10
15
20
25
0 5 10 15 20 25 30
ALT AP R
Time (days)
X U
LN
ONSET
ALT MAX
AP MAX
BIOPSY
Presenter
Presentation Notes
One of the other questions was, how do individual histologic findings relate to the serum enzyme levels? So, this is a hypothetical profile. You might have the biopsy located anywhere along this profile of ALT changes and alk phos changes. In this particular example, R goes through hepatitic or hepatocellular at onset down and eventually becoming cholestatic towards the end.
Data Analysis
• All DILIN cases with biopsies adjudicated as Probable, Very Likely or Definite DILI
• Laboratory data (AST, ALT, AP, tBili, R) at 5 points defined by profile or protocol – Onset, ALT Max, AP Max, At Biopsy, Baseline– “At biopsy” labs were within 7 days
Presenter
Presentation Notes
So, the data analysis included DILIN cases that had biopsies that were adjudicated as probable or more likely than that. We looked at this biochemical data, transaminase elevations, bilirubin, and R at five points, although I am only going to show you the data from onset, basically, and then at biopsy. The at-biopsy labs were within seven days.
A Hep26%
C Hep14%
A Chol8%
C Chol12%
Mixed27%
Zonal Nec5%
Other8%
Relative Frequency ofPathology Patterns inDILIN Patients with atLeast Probable CausalityDetermination
There were 106 biopsies that fell into this category where we knew what the adjudication was and had biochemical data. These were the patterns. The most common patterns that I see are the acute hepatitic pattern and the mixed cholestatic hepatitic pattern, followed by the other cholestatic patterns and the chronic hepatitis pattern. Then, everything else kind of fits into the zonal necrosis and the other injuries into a minority of cases.
-200 -150 -100 -50 0 50 100 150 200
At Biopsy
Baseline
MaxALT
MaxAP
Onset
DAYS
Time Between Lab Test and Biopsy at Various Time Points
Presenter
Presentation Notes
This is just a bar diagram or a box plot to show you where the biopsy labs fell or the biopsy fell with respect to the labs. You can see the at-biopsy labs very tightly clustered near the biopsy. Then, most of the time, the comparator labs preceded the biopsy, but baseline evaluation, though, often follows the biopsy.
Biochemical Injury at Onset (R)
Pattern Total Hepatocellular Mixed CholestaticAcute Hepatitic 27 18 5 4
Distribution of Pathology by Onset Biochemical Injury
Presenter
Presentation Notes
This is the distribution of pathology patterns by the onset biochemical injury. The thing I want to point out here is, if you look at, focus on those top five lines where I have most of the cases, acute hepatitic, chronic hepatitic, acute and chronic cholestatic, and mixed, you see that all of them have the whole range of biochemical injuries at onset.
Biochemical Injury at Biopsy (R)
Pattern Total Hepatocellular Mixed CholestaticAcute Hepatitic 25 17 5 3
Distribution of Pathology by Biochemical Injury at Biopsy
Presenter
Presentation Notes
Now some things are more present, more likely to be present. So, for instance, acute hepatitic injury pattern is more likely to present with a hepatocellular R, but four cases were cholestatic. So, it is not an absolute correlation. If you look at the distribution of pathology by biochemical injury at the time of the biopsy, it is a little bit more what you would expect. There's maybe a little bit shift in the data from biochemical patterns that you wouldn't necessarily expect, the ones that you would, but it is still spread out over the whole range. So, you cannot make assumptions based on the biochemical profile about what you are going to see on the biopsy because you are going to be wrong about a third of the time, if not more.
-20
0
20
40
60
80
100
120
1 2 3 4 5 10
Variation of “R” at Onset with Pathologic Pattern of Injury
AcuteHepatitis
ChronicHepatitis
AcuteChol.
ChronicChol
MixedH/C
ZonalNecrosis
R
Presenter
Presentation Notes
So, this is the variation of R. So, it is, again, a box-and-whiskers plot showing the variation at onset with the pathology pattern, just looking at these six most common patterns. You can see that the variation is kind of what you would expect with the acute hepatitic and chronic hepatitic and zonal necrosis patterns having the highest R's, and acute cholestatis, chronic cholestatis low; mixed patterns, kind of spread out over the whole range.
-20
0
20
40
60
80
100
120
1 2 3 4 5 10
R
0
20
40
60
80
100
120
140
160
1 2 3 4 5 10
ALT/ULN
0
2.5
5
7.5
10
12.5
15
17.5
20
1 2 3 4 5 10
AP/ULN
0
5
10
15
20
25
30
35
40
45
1 2 3 4 5 10
tBili
Onset Lab Data
Presenter
Presentation Notes
If you look at the other laboratory data, and the graphs are laid out exactly the same way, so at onset the ALT is going to parallel R because ALT kind of drives that ratio. You can see that in the bottom lefthand graph that the alkaline phosphatase is highest in patients with chronic cholestatic patterns, mixed, and acute cholestatics. And the bilirubin is kind of all over the place, actually. Even in patients where I did not see visual bile on the biopsy, they could be quite jaundiced at presentation.
R ALT/ULN
AP/ULN tBili
-20
0
20
40
60
80
100
120
1 2 3 4 5 10 0
20
40
60
80
100
120
140
160
1 2 3 4 5 10
0
2
4
6
8
10
12
14
16
18
20
1 2 3 4 5 100
5
10
15
20
25
30
35
40
45
1 2 3 4 5 10
Lab Data At Biopsy
Presenter
Presentation Notes
If you look at the lab data at the time of biopsy, there is a little bit more variation, particularly in the bilirubin. Now what you see is very high bilirubins and the patient showing acute cholestatic patterns and mixed injury patterns on biopsy and lower bilirubins elsewhere. But everything else kind of matches what was there before.
Range of Injury by Pattern(Comparison to Zimmerman)
DILIN (onset labs) Zimmerman
ALT/ULN AP/ULN ALT/ULN AP/ULN
Acute Hepatitis 4-38x 0.7-4x 10-100x 1-3x
Chronic Hepatitis 3-15x 0.6-4x 3-50x 1-3x
Acute Cholestasis 4-15x 1-10x 1-5x 1-3x
Chronic Cholestasis 3-14x 1-6x 1-5x 3-20x
Mixed H-C 2-60x 1-5x 1-10x >3x
Zonal Necrosis 6-140x 0.7-1.4 100-1000x 1-3x
Presenter
Presentation Notes
If you look at the range of injury by pattern, this is a table that was in Dr. Zimmerman's book where I think, since he doesn't reference where this data comes from, I assume this is like seat-of-the-pants, rule-of-thumb sort of data on the righthand side of this table. So, he was giving what he thought were the ranges. Then, on the lefthand side of the table are the ranges that I see in DILIN, mostly from sort of 10 percent to 90 percent. So, I have cut off the ends of the range. And things pretty much match up, although there are some surprises, particularly with respect to the acute and chronic cholestatic injury patterns. They can have a much higher ALT than what was presented in Dr. Zimmerman's book.
Serum Biochemistries at Time of BiopsyFeature R ≥ 5 ALT ≥ 5xULN AP ≥ 2xULN tBili ≥ 3
Interface Hepatitis 0.025 0.0016 >0.2 >0.2
Lobular Inflammation <0.0001 <0.0001 >0.2 >0.2
Portal Inflammation >0.2 0.18 >0.2 >0.2
Granulomas 0.057 >0.2 0.0005 >0.2
Plasma Cells 0.043 0.0009 >0.2 >0.2
Eosinophils >0.2 0.18 0.19 >0.2
Bridging Necrosis 0.0003 <0.0001 >0.2 >0.2
Canalicular Cholestasis >0.2 >0.2 >0.2 <0.0001
Chronic Cholestasis 0.089 >0.2 0.11 >0.2
Ductular Reaction 0.11 0.13 >0.2 >0.2
Ductular Injury (0.047) >0.2 0.037 0.0007
Ballooning >0.2 >0.2 (0.016) >0.2
Apoptosis 0.0001 >0.0001 >0.2 0.16
Coag/Confluent Necrosis 0.15 0.005 >0.2 >0.2
Pathologic Correlates of Biochemical Parameters
Presenter
Presentation Notes
The pathological correlates of these biochemical patterns are shown here. So, this is where I took those individual features that I looked at and compared them to the serum biochemistry. So, for instance, and this was more of a validation thing for me because you expect certain things to result in certain biochemical changes.
Distribution of Inflammation Severity with respect to ALT at Biopsy
05
101520253035404550
0 1 2 3 4
ALT<5ULN ALT>5ULN
0102030405060708090
1 2 3 4
ALT<5ULN ALT>5ULN
Interface Hepatitis Lobular Inflammation
%
p=0.0016 p<0.0001
Presenter
Presentation Notes
So, hepatitis is correlated with high ALT, as you would expect. The presence of plasma cells was also correlated with high ALT and necrosis. Bridging necrosis was also very highly correlated. Some things were, well, I don't know if they were unexpected, but kind of surprising. Granulomatous inflammation, when it was present, was associated with high alk phos. Canalicular cholestasis, not surprisingly, associated with total bilirubin and some other things, as I have shown. This is just looking at some of those individual things. The interface hepatitis on the left and the lobular inflammation on the right, both highly correlated with the amount of or the height of the ALT. Patients with more inflammation tend to have higher ALTs, no surprise.
Distribution of Necrosis Severity with respect to ALT at Biopsy
0102030405060708090
100
None BN MLN
ALT<5ULN ALT>5ULN
0
10
20
30
40
50
60
0 <1/hpf 1-3/hpf >3/hpf
ALT<5ULN ALT>5ULN
Bridging/MultiLobNecrosis
Acidophil bodies
%
p=0.0016 p<0.0001
Presenter
Presentation Notes
The same way with necrosis. So, the presence of bridging and multi-lobular necrosis tended to be associated with high ALT as well as the number of acidophil bodies that were present on the biopsy.
Canalicular Cholestasis and Duct Injury in Relation to total Bilirubin at Biopsy
0
10
20
30
40
50
60
70
80
None One Duct Mult Ducts
tBili<3 tBili>3
0102030405060708090
100
Absent Present
tBili<3 tBili>3
Duct Injury Canalicular Cholestasis
%
p=0.0077 p<0.0001
Presenter
Presentation Notes
Canalicular cholestasis and duct injury were related to the total bilirubin. So, the more ducts that were injured, the more likely the patient was to have a high bilirubin. And the same way with the presence of canalicular cholestasis. All of these things make sense and just provide validation for what I do as the pathologist, sorting out these individual features.
Relationship of Biochemistry to Histology
• Biochemical Classification using “R” only loosely correlates with the actual histologic injury pattern– Patients with “cholestatic” R values can have an acute hepatitis
pattern on biopsy– Patients with “hepatocellular” R values can have acute or chronic
cholestasis– Biochemical classification not useful for varient patterns: e.g.
granulomatous and steatohepatitic• Individual biochemical tests (ALT, AP, Bilirubin) do
correlate with some histologic features in a rational way• Some histological features, such as the cell types
involved in the injury, are not reflected in laboratory tests
Presenter
Presentation Notes
Just to summarize, the biochemical classification using the ratio of ALT to alk phos only loosely correlates with the actual histologic pattern. So, patients with cholestatic R values can have acute hepatitis. Patients with hepatocellular R values can have acute or chronic cholestasis. They are really not useful at all for some of the minor patterns, granulomatous injuries, steatohepatitis, that sort of thing. So, if that is what is in your patient, you are not going to know that. You may not even think about that. Individual biochemical tests do correlate with the histologic features in a rational way. Then, some histologic features, like eosinophilia, was not represented at all by the lab tests that I looked at, but might correlate to things like peripheral eosinophilia, which I did not look at.
Liver Histology and Outcome in DILI
Presenter
Presentation Notes
Now just a few words on outcome.
• Meta-analysis of 570 case reports of DILI with liver biopsy• Study evolved from a study of disulfiram DILI in which these factors
were associated with outcome• “Eosinophilia” was recorded if patient had any degree of eosinophilic
infiltrate on biopsy• “Hepatic necrosis” was recorded if biopsy was described as having
any of the following: “centrilobular drop-out, centrilobular necrosis, confluent necrosis, panlobular necrosis, submassive necrosis and massive necrosis”
• These factors were related to fatal outcome
Not-Fatal Fatal p
Eosinophilia 48% 18.8% <0.0001
Hepatic Necrosis 24% 84% <0.0001
Presenter
Presentation Notes
This was a paper that was published by Einar Bjoernsson a couple of years ago now. It was a meta-analysis of 570 case reports. So, he culled the literature for case reports that had liver biopsies done, looked at eosinophilia both on peripheral blood measurements and whether or not any degree of eosinophilia was reported in a biopsy, and then captured also hepatitic necrosis very broadly, looking for all of these different terms, and then related those to fatal outcome, and showed that eosinophilia, either on biopsy or in peripheral blood, was more likely to be seen in patients who survived than those who didn't; whereas, the opposite was true with hepatitic necrosis. This all, of course, makes sense, but it hadn't really been demonstrated in this large-scale way before.
Histopathology and Outcome in the DILIN cohort
• Reviewed pathology data and DILIN severity score on patients in the database with both a biopsy reviewed and causality completed with DILI causality of probable or greater
• 111 patients met this criterion• Age 49.8 (7.8 to 87.2 years)• 41% Male• Analyzed individual histologic observations
including necrosis, inflammation, fibrosis, cholestasis and others comparing fatal vs non-fatal and severe/fatal vs less than severe
Presenter
Presentation Notes
So, I looked at the DILIN cohort pathology data with respect to these same things. This represents data through September last year. There were 111 patients who had both a biopsy and causality completed with a DILIN causality of probable or greater. They were, as shown before by Bob and Naga, sort of close to 50 years old. About 40 percent were male. And I looked at the individual histologic observations and the pattern of injury comparing them, sort of dividing the cohort of patients with those who died and those who didn't, and those who had severe injury, severe or fatal injury, and those who had less degrees of injury.
DILIN Severity ScoreMild Elevated ALT, AST or AP, tBili <2.5, no
coagulopathy (INR<1.5)
Moderate Elevated ALT, AST or AP, and tBili >2.5 or coagulopathy (INR>1.5)
Moderate-Hospitalized
As for Moderate but patient is either hospitalized for DILI or pre-existing stay is prolonged
SevereElevated ALT, AST or AP, with tBili >2.5 and either hepatic decompensation or other DILI-related organ failure
Fatal DILI related death or transplantation
Presenter
Presentation Notes
This is the DILIN severity score, which basically classes patients as to how bad the injury is: mild, moderate, moderate and hospitalized, severe, or fatal. So, I either grouped severe and fatal together or fatal by itself.
Mild Moderate Mod-Hosp Severe Fatal Total
Acute Hepatitis 4 5 14 2 2 27
Chronic Hepatitis 10 1 1 3 2 17
Acute Cholestasis 1 3 4 1 0 9
Chronic Cholestasis 2 3 4 0 0 9
Mixed Hep-Chol Inj 0 7 14 6 1 28
Granulomatous Hep 0 0 2 0 0 2
Steat, Microves 0 1 0 0 0 1
Steatohepatitis 1 0 1 1 0 3
Coag. Necrosis, zonal 1 1 2 2 2 8
Vascular 1 0 0 0 0 1
Hep. Alteration 2 0 0 0 0 2
Unclassifiable 0 0 2 1 1 4
Total 22 21 44 16 8 111
Presenter
Presentation Notes
This is just looking at all of those patterns of injury again with respect to outcome. There isn't any real clear outcome related to pattern that is evident. Basically, any pattern, especially the ones that are represented the most frequently, resulted in both fatal and non-fatal outcomes.
Relationship of Most Common Pathology Pattern with Outcome
Mild Moderate Mod-Hosp Severe Fatal Total
Acute Hepatitis 4 5 14 2 2 27
Chronic Hepatitis 10 1 1 3 2 17
Acute Cholestasis 1 3 4 1 0 9
Chronic Cholestasis 2 3 4 0 0 9
Mixed Hep-Chol Inj 0 7 14 6 1 28
Coag. Necrosis, zonal 1 1 2 2 2 8
Total 22 21 44 16 8 111
Presenter
Presentation Notes
This just kind of focuses it on the six most common patterns. I guess the one standout was that the chronic hepatitis pattern did seem to present mostly with mild injury, with rare or fatal or severe cases. As far as the individual histologic features and outcome, again, looked for non-random associations with either severe/fatal or fatal. There were very few features, actually, that were predictive. Most had no association at all, except for the things that I have listed here.
Histologic Features and Outcome
• Individual features assessed on biopsy were evaluated for non-random association with either severe/fatal or fatal outcome
• Analysis somewhat limited by the small number of fatal cases (8/111)
• Very few features showed any suggestion of correlation with outcome with a few exceptions:– Granulomas associated with better outcome– Eosinophils showed a non-significant trend to better outcome– The presence of multiacinar or bridging necrosis but not the
degree of confluent necrosis was associated with poor outcome– Ductular reaction was associated with poor outcome
Presenter
Presentation Notes
Granulomas were associated, interestingly, with better outcome. Eosinophils showed a non-significant trend. It came close, but didn't quite make it to statistical-significance. And the presence of multiacinar or bridging necrosis, but not the degree of confluent necrosis, was associated with poor outcome. This, again, I think is partly due to the low numbers of fatal cases that we had. There were only eight in this group. Then, ductular reaction, interestingly, was associated with poor outcome. I will go into that a bit.
Granulomas and Eosinophils
0102030405060708090
100
None/Mild Increased
Not Severe Severe/Fatal
0102030405060708090
100
NoGranulomas
Microgran Epith Gran
Not Severe Severe/Fatal
Granulomas Eosinophils
Per
cent
of C
ases
p=0.008 p=0.06
(46)(58) (6) (56) (54)
Presenter
Presentation Notes
These are the granulomas and the eosinophils. You can see that, as the degree of granulomatous inflammation increases, the less likely you are to find fatal cases among the group. There was that trend in the eosinophils, as to whether they were definitely increased or just none at all or only a few seen. It may have to do with the threshold. I didn't draw it at none because my feeling was I could find you an eosinophil in almost any liver biopsy, if I looked hard enough.
Necrosis
0102030405060708090
100
<33% >33%
Not Fatal Fatal
0102030405060708090
100
Not Present Present
Not Fatal Fatal
Per
cent
of C
ases
p=0.004p=0.18
(77) (33) (99) (11)
Bridging Necrosis or Multiacinar Necrosis Confluent Necrosis Fraction
Presenter
Presentation Notes
Necrosis, here is the bridging necrosis, multiacinar necrosis, data on the left. It showed that almost all the fatal cases had some degree of bridging necrosis or multiacinar necrosis. Confluent necrosis showed the same trend again, but, again, not a significant P value.
Ductular Reaction
0102030405060708090
100
None/Mild Prominent
Not Fatal Fatal
0102030405060708090
100
None/Mild Prominent
Not Severe Severe/Fatal
Per
cent
of C
ases
p=0.004p=0.005
(67) (42) (67) (42)
Presenter
Presentation Notes
This is the ductular reaction data. So, when ductular reaction is prominent, I am going to score that as a positive. Again, a majority of the severe/fatal cases and fatal cases were in this group of patients with prominent ductular reaction. That is not usually a feature that you might think of as being associated with poor outcome. It is a lesion that is characterized by small ductules, often without a visible lumen. There is often fibrosis around it and inflammatory cells. In fact, neutrophils are quite frequently seen associated with ductular reaction.
Ductular Reaction• Ductular reaction is a lesion characterized by small
ductules, often without a visible lumen, fibrosis and inflammatory cells
• In impaired hepatic regeneration, it has been shown to be a secondary pathway of hepatocyte progenitor replication– Roskams T, et al., Am J Pathol 2003;163:1301–1311
• In some chronic liver diseases, it has been assocatiated with replicative failure in primary hepatocytes – Richardson MM et al., Gastroenterology 2007;133: 80–90;– Roskams T. J Hepatol 2003; 39: 431–434.
Presenter
Presentation Notes
However, it has been shown that in an impaired hepatic regeneration, they become activated as a secondary pathway of hepatocyte progenitor replication. In some chronic liver diseases, it has been associated with replicative failure in primary hepatocytes. So, what we may be seeing in some of these cases is a failure of the remaining liver cells to replicate appropriately, and these other systems are being turned on. Hence, I am seeing ductular reaction.
Conclusions• Biochemical classification of injury along with the
results of specific laboratory tests at most allow an educated guess at the histologic pattern of injury
• Specific features on biopsy may be useful in prognosis
• Liver biopsy remains a valuable tool that may give insight into mechanism of injury, potential for chronicity and help with difficult clinical differential diagnoses
Presenter
Presentation Notes
So, just conclusions from my part of the talk: biochemical classification of injury, along with the results of specific laboratory tests that, at most, allow an educated guess at the histologic pattern, and it is really not appropriate to use histologic terms unless you have actually done a biopsy. Specific features on the biopsy can be useful in prognosis, although it may not be soon enough to do anything about it. Liver biopsy, I feel, remains a valuable tool that can give insight into mechanism of injury, particularly if you are dealing with a new agent. Potential for chronicity in terms of showing you fibrosis and other things like vanishing bile duct syndrome, which might make for a prolonged course, and I think it can help with difficult clinical differential diagnoses. That has been my experience, anyway.
AcknowledgmentsDILIN Investigators
– Herbert Bonkovsky (Petr Protiva)-UConn– Naga Chalasani-IU– Timothy Davern-UCSF– Robert Fontana-UMich– Paul Watkins-UNC– William Lee-UTSW– Andrew Stolz-USC– Jayant Talwalkar-Mayo– Victor Navarro-Jefferson/UPenn– James Rochon-DCRI– Jose Serrano-NIDDK Project Officer
Presenter
Presentation Notes
That's it. So, these are my acknowledgments. I am very grateful to the Network and to the other DILIN investigators for providing me with this opportunity to look at all of these amazing cases. The people who are not italicized were the investigators in the first part of the study, mostly carried forward now, but those are the ones who contributed the cases that I talked about. Thank you. (Applause.)
