A Severe Case of Idiosyncratic Hepatotoxicity with Unfractionated Heparin

Abstract

The authors describe a case of clinically apparent idiosyncratic hepatotoxicity in association with unfractionated heparin (UFH). A 52-year-old woman with increasingly symptomatic rheumatic mitral valvular disease and severe pulmonary hypertension underwent elective minimally-invasive bioprosthetic mitral valve replacement. The patient received 42 000 units of UFH intraoperatively 10 days after receiving 3100 units during a left heart catheterization. Standard prophylactic doses of unfractionated heparin were started on POD 2 for prevention of venous thromboembolism. On the evening of postoperative day (POD) 3, the patient was lethargic, encephalopathic, and hypoglycemic with an acute liver injury and hyperlactatemia. Similar events occurred on POD 7 after clinical improvement from the initial injury and an unintentional rechallenge with UFH. Heparins are usually not suspected of idiosyncratic hepatotoxicity due to their widespread utilization and reports of milder episodes of hepatotoxicity. This case highlights the need to consider UFH in the differential of drug-induced liver injury, including severe cases.

Intro

Drug-induced liver injury (DILI) occurs infrequently in the general population but is the most common cause of acute liver failure in Western countries.^1,2 Idiosyncratic types of DILI are unpredictable, poorly correlated with dose, and variable in their presentation. Drug-induced liver injury remains largely a diagnosis of exclusion and is reliant on a high index of suspicion since there are no definitive diagnostic tests or biomarkers. While 90% of patients are expected to fully recover, even when jaundiced, some may develop acute liver failure resulting in death or transplantation and chronicity may occur in up to 20% of patients. ³ The effective management of DILI involves rapid recognition, withdrawal of the suspected agent(s), and supportive care. However, the timely identification of potentially causative medications can be difficult because of the various etiologies of hepatotoxicity as well as the vast number of medications implicated in hepatotoxic episodes.

Unfractionated (UFH) and low molecular weight heparins are ubiquitous anticoagulants in hospitalized patients with indications ranging from general venous thromboembolism (VTE) prophylaxis to specific clotting prevention in cardiopulmonary bypass during surgery. We evaluated the U.S. Food and Drug Administration Adverse Event Reporting System (FAERS), National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) and National Library of Medicine’s LiverTox database, and medical literature in PubMed for cases of severe hepatotoxicity in patients receiving UFH before April 2020. Keywords for the various databases included heparin, heparins, unfractionated heparin, heparin sodium, hepatotoxicity, hepatic injury, livery injury, drug induced liver injury, abnormal hepatic function, and hepatic failure. These searches revealed UFH-associated DILI is unlikely and more commonly associated with transient aminotransferase elevations compared to low molecular weight heparin (LMWH) (up to 60% versus 13%), but both are usually asymptomatic, mild, and self-limited.^4,5,6 There has been no compelling evidence to date relating UFH or LMWH to clinically apparent idiosyncratic hepatotoxicity with jaundice.

Case Report

A 52-year-old Caucasian woman with increasingly symptomatic rheumatic mitral valvular disease and severe WHO Group II pulmonary hypertension underwent an elective minimally-invasive bioprosthetic mitral valve replacement for mitral stenosis and regurgitation. Her past medical history included polysubstance abuse, alcoholism, chronic heart failure, hepatomegaly, and hepatic steatosis. She experienced multiple transient serum aminotransferase elevations over several months before admission, with a serum aspartate aminotransferase (AST)/alanine aminotransferase (ALT) ratio matching the classical 2:1 excess seen in alcoholic hepatitis. ⁷ However, her smoldering hepatic dysfunction was not an acute preoperative concern with an AST and ALT the day before surgery of 25 and 18 units/L, respectively, and prothrombin time (PT)/international normalized ratio (INR) of 12.5 sec/0.9. The patient’s serum alkaline phosphatase (AP) was elevated at baseline (2 times the upper limit of normal [ULN]), but she had a normal post-cholecystectomy magnetic resonance cholangiopancreatography several months before surgery and her AP improved over the course of her hospitalization (1.3 times ULN at discharge). The patient received 3100 units of UFH during a left heart catheterization 10 days before surgery and an additional 42 000 units of UFH while on cardiopulmonary bypass (CPB). Goal levels of anticoagulation were achieved on CPB and aminocaproic acid was given intraoperatively to prevent excessive blood loss. Her immediate postoperative course was relatively unremarkable, including the absence of coagulopathy (PT 15.6 seconds, INR 1.2, activated partial thromboplastin time 27.2 seconds, fibrinogen 263 mg/dL) upon arrival to the intensive care unit (ICU). Oral amiodarone (200 mg twice daily) was initiated on postoperative day (POD) 1 for postoperative atrial fibrillation prophylaxis. Hemodynamic support with phenylephrine was quickly weaned and the patient was transferred to the surgical step-down unit on POD 2. Subcutaneous UFH 5000 units twice daily was given for VTE prophylaxis starting the evening of POD 2.

No nutrition was ordered until POD 3, at which point the patient refused to eat. The patient went into atrial fibrillation that morning, 2 amiodarone 150-mg intravenous (IV) boluses were given, and her oral amiodarone was increased to 400 mg twice daily. On the evening of POD 3 a rapid response was called due to increasing lethargy and encephalopathy. The patient’s blood glucose (BG) was 26 mg/dL and, despite her hemodynamic stability, lactic acid was 14.1 mmol/L. She was transferred back to the ICU, made nil per os, and ultimately given 50 g of dextrose 50% and 3 slow IV pushes of sodium bicarbonate 50 mEq without any additional IV fluids. Her preoperative hemoglobin A1c was 4.8% and the only insulin administered during hospitalization was regular insulin 2 units/hr during CPB.

The patient experienced an acute liver injury during these events on POD 3 with an elevated AST, ALT, and bilirubin (12.6, 4.5, and 2 times the ULN, respectively) (Figure 1). A computerized tomography (CT) scan of the chest, abdomen, and pelvis confirmed diffuse hepatic steatosis without definite focal hepatic lesion or biliary dilatation. On the morning of POD 4, her lactate declined to 2.5 mmol/L and she developed a respiratory alkalosis subsequent to severe epigastric and back pain. Hepatic arterial thrombosis was ruled out by ultrasound after her AST and ALT continued to worsen. The patient’s INR climbed to 1.8 and subcutaneous UFH was discontinued. Four doses of prophylactic UFH had been given since the evening of POD 2. Piperacillin/tazobactam and vancomycin were administered on POD 4 to empirically cover a nosocomial infection due to an elevated procalcitonin (PCT) level (3.91 ng/mL) with no signs of acute kidney injury. She became febrile (peak temperature 39.4°C) with altered mental status on POD 5, yet an ensuing infectious workup was ultimately negative, including hepatitis serologies and active Epstein-Barr virus (EBV) and cytomegalovirus (CMV) infections. Other diagnostic testing revealed hypogammaglobulinemia and the absence of serum antinuclear antibodies (ANA).

Figure 1.

Liver test trends during hospitalization.

Note. 7a = early morning; 7b = mid-morning; 7c = mid-afternoon; Amio = amiodarone; AST = aspartate aminotransferase; ALT = alanine aminotransferase; AP = alkaline phosphatase; CPB = cardiopulmonary bypass; D/C = discontinued; IV = intravenous; PO = oral; SQ = subcutaneous; TBili = total bilirubin; UFH = unfractionated heparin; ULN = upper limit of normal; VTEp = venous thromboembolism prophylaxis.

On POD 6, hepatic function, INR (1.6), lactic acid (2.1 mmol/L) and neurological status were improved despite continued abdominal pain. Oral medications and subcutaneous UFH 5000 units twice daily were restarted. On POD 7, the patient’s AST and ALT continued to decline but the total bilirubin and INR rose to 1.8 mg/dL and 2.6, respectively. Signs of slight jaundice with mild scleral icterus were first noted on this day. Similar to the events on the evening of POD 3, the patient became tachypneic and agitated. Levothyroxine was the only oral medication the patient received between the order to restart oral medications and her decompensation. Nine hours after the second dose of UFH a random BG was 24 mg/dL. The patient had multiple hypoglycemic excursions over the next 6 hours and was thrice given 25 g of dextrose 50% and started on a continuous infusion of dextrose 5% in Lactated Ringers (D5LR). Her lactic acid was 15.4 mmol/L and D5LR was transitioned to dextrose 5% with sodium bicarbonate 150 mEq/L. UFH was once again discontinued. The patient became increasing less arousable throughout the day and was reintubated later that afternoon. Phenylephrine was required for hemodynamic support.

Enteral nutrition was initiated on POD 8. No further episodes of hypoglycemia occurred, and the patient only received insulin intraoperatively during her hospitalization. Serum PCT peaked at 11.88 ng/mL on POD 9 and trended down to 0.52 ng/mL on POD 24. A single 150-mg dose of intravenous amiodarone was given on POD 9 for recurrent ectopy. The ectopy resolved and no further doses were needed. Her lactic acid normalized by POD 11. The patient’s liver biochemistries continued to improve during the UFH washout. Peak ALT and bilirubin levels declined from 871 to 42 units/L and 5.2 to 0.7 mg/dL, respectively, but did not return to baseline levels by discharge. All other causes of hepatotoxicity were ruled out. Nutritional, ventilatory, and hemodynamic support were discontinued on POD 14 after the patient stabilized.

The patient’s hospital course was complicated by an extensive left middle cerebral artery embolic stroke of the proximal (M1) segment with subsequent aphasia and right-sided hemiplegia which was identified on POD 10. A cardioembolic source was suspected, but no intracardiac thrombus was identified on transesophageal echocardiogram. Anticoagulation was delayed and only given at prophylactic levels due to the high risk for hemorrhagic conversion, relative thrombocytopenia, and ligation of the left atrial appendage during surgery. UFH and low molecular weight heparin were listed as drug allergies for the patient. Fondaparinux and was administered for VTE prophylaxis beginning on POD 20 after her INR normalized. Aspirin was also administered as antithrombotic therapy for her valve prosthesis and secondary stroke prevention. She experienced partial recovery of motor function and speech prior to discharge. The patient was transitioned to inpatient rehabilitation on POD 26 and ultimately discharged home in stable condition on POD 54.

Discussion

Most cases of hepatotoxicity with UFH are mild and asymptomatic. In the reportedly rare instances of clinically apparent liver injury due to UFH, other possible causes of hepatotoxicity were present. Although more severe cases (serum aminotransferase levels above 5 times the ULN) occur in approximately 2% of patients receiving high doses of heparin, this case represents a highly probable severe case of idiosyncratic DILI in a patient receiving standard prophylactic doses of UFH. ⁶

The diagnostic approach to idiosyncratic DILI is typically guided by determining an R-value which identifies the pattern of liver injury at presentation. ³ The R-value is equal to the serum ALT/ULN divided by serum AP/ULN. Hepatocellular DILI have an R-value ≥5, while cholestatic DILI have R-value <2, and R-values for mixed injuries lie between the two. No guidance exists for calculating an R-value in patients with pre-existing liver disease. Since our patient’s preoperative liver biochemistries were normal, we calculated the R-value for this patient using the local ULN values. This patient experienced 2 separate postoperative episodes of acute liver injury with initially different phenotypic patterns. Following the events in the late evening of POD 3, the R-value was 4.6, which correlated with a mixed cholestatic and hepatocellular injury. During the second hepatotoxic episode on POD 7, however, the R-value (8) was consistent with a hepatocellular injury. It is possible that the laboratory measurements and phenotypes of liver injury can present differently among patients, but currently there is a lack of information about a medication resulting in categorically different R-values within an individual patient. This is likely due to the avoidance of subsequent exposures to suspected causes of DILI. It is also possible that a singular injury produces an evolving pattern as the aminotransferase levels change, though usually an initial hepatocellular pattern transitions to cholestatic later in the clinical course. ⁸ However, an unintentional rechallenge of UFH occurred in this patient and likely resulted in more than one injury. The mechanism of heparin-induced hepatotoxicity is not known but is likely due to a direct hepatocellular injury. ⁶ The differences between R-values in this patient could be due to latency, pre-existing liver disease, or immunologic priming.

When hepatocellular or mixed DILI is suspected, acute viral and autoimmune hepatitis serologies and imaging studies are recommended first-line tests. ³ Testing for hepatitis A, B, and C viruses was non-reactive in this patient. Hepatitis C viral (HCV) RNA was not measured due to the non-reactive test for antibodies toward HCV. The patient was not suspected of any exposures to HCV within the previous 6 months, but HCV RNA testing could have been considered given the subsequently diagnosed hypogammaglobulinemia and possibility of initial false negative anti-HCV antibody testing. Autoimmune hepatitis was not suspected given the absence of ANA and the presence of hypogammaglobulinemia. As previously mentioned, imaging studies did not reveal any findings to suggest other causes of hepatotoxicity.

Second-line tests for hepatocellular and mixed DILI are considered on a case-by-case basis. Hepatitis E virus (HEV) antibodies and ceruloplasmin levels were not tested due to the low clinical suspicion for HEV infection or Wilson’s disease. Testing for EBV and CMV was non-reactive. Serologies for acute herpes simplex viral infection were not completed; however, the patient did not have atypical lymphocytosis, rash, or lymphadenopathy. Budd-Chiari syndrome was ruled out by hepatic ultrasound, though it was not suspected. A liver biopsy is not mandatory in the evaluation of DILI and was not clinically indicated per guideline recommendations. ³

The most likely causes of fever in this patient included sepsis, colitis, and atelectasis. Sepsis could cause ischemic liver damage or direct and indirect damage to hepatocytes, but no definite source of infection was identified. Microbiologic cultures remained negative but were drawn 4 hours after the first doses of empiric antibiotics following the patient’s initial decompensation. Instead of infection, this patient’s PCT level could have been elevated due to the inflammation associated with hepatocyte necrosis in the setting of acute liver failure. ⁹ Focal colitis and bilateral patchy atelectasis were identified on CT scan and were the most likely alternative sources of fever. This patient completed 10 days of broad-spectrum treatment and only required mild to moderate doses of phenylephrine for hemodynamic support.

The two most likely causes of hepatocellular injury in this patient were ischemic hepatitis and idiosyncratic DILI; however, the former is doubtful. This patient required minimal postoperative hemodynamic support, maintained steady arterial pressures, and was transitioned off a phenylephrine infusion by the morning of POD 2. Symptoms of the first episode of liver injury occurred on POD 3 despite continued mean arterial pressures in the 70 to 80 mm Hg range. Similarly, the second episode of liver injury occurred after patient decompensated on POD 7 without any preceding hemodynamic compromise. An intraoperative ischemic insult could have manifested by POD 3, but a partial recovery with subsequent worsening on POD 7 is not the predicted pattern for that type of injury.

Drug-induced liver injury remains a diagnosis of exclusion. Since few medications have never been linked to hepatotoxicity, isolating a singular probable cause of DILI can be difficult. Furthermore, a degree of subjectivity lies within the assessment of DILI with the available scoring tools. Heparin and amiodarone were the most likely causes of DILI in our patient. Other medications were ruled out based on the timing of administration relative to the 2 incidents of acute liver injury in our patient. Medication exposures preceding the 2 episodes of acute liver injury on postoperative days 3 and 7 are outlined in Table 1. Amiodarone is a well-established cause of clinically apparent liver injury, while heparin is not an established likely cause of severe acute liver injury as no definitive cases are reported in the literature.^6,10,11 However, the pattern of our patient’s injury was more likely explained by UFH. We used multiple assessment algorithms to objectively assess the possibility of one of these medications causing the hepatic injury. The Roussel Uclaf Causality Assessment Method (RUCAM) system is the most widely used method in clinical practice.^3,8 Although a hepatocellular injury was clinically determined as more likely, we calculated the RUCAM score for both hepatocellular and mixed injuries since the 2 separate postoperative episodes presented with different R-values. Heparin scored as a highly probable cause of hepatocellular injury and probable cause of mixed injury (Table 2). Heparin was not considered a causative agent after the initial hepatic injury on POD 3. Consequently, an unintentional rechallenge occurred when UFH was restarted on POD 6. The positive UFH rechallenge escalates the probability by one category for each type of injury.

Table 1.

Medication Exposures Preceding the 2 Episodes of Acute Liver Injury on Postoperative Days 3 and 7.

Medication	Pre-op	Day of surgery	POD 1	POD 2	POD 3	POD 4	POD 5	POD 6	POD 7
UFH	3100 units IV POD –10	42 000 units IV		5000 units SQ x1	5000 units SQ x2	5000 units SQ x1		5000 units SQ x2
Amiodarone			200 mg PO x2	200 mg PO x2	150 mg IV x2; 400 mg PO x2
Isoflurane		0.5%-1% with oxygen
Propofol		X
Cefazolin		X	X
Vancomycin		X				X
Famotidine		X		X	X	X	X	X
Ondansetron		X		X
NAC nebules		X	X	X	X
Ipratropium/albuterol		X	X	X	X	X	X	X
Morphine		X
Ketorolac		X	X
Oxycodone		X	X
Fentanyl		X	X	X		X	X	X
Diazepam		X	X	X
Phenylephrine		20-40 mcg/min	Weaned off	40 mcg/min at midnight; off at 0930					X
Hydrocodone			X	X
Duloxetine			X	X	X
Gabapentin			X	X	X
Levothyroxine			X	X	X			X
Omeprazole			X	X	X
Clonazepam				X	X
Furosemide				X	X
Lamotrigine				X	X
Metoprolol				X	X	X	X	X
Aspirin					X	X	X
Acetaminophen					X		X
Pantoprazole						X	X	X
Lorazepam						X	X	X
Piperacillin/tazobactam						X	X	X
Hydromorphone								X
IV fluids		NS	NS	NS			D5NS		D5LR; SW/HCO3
Nutrition		NPO	NPO	NPO	PO ordered but refused	Minimal to none PO	Minimal to none PO	Minimal to none PO	Minimal to none PO

Note. D5LR = dextrose 5% in Ringer’s lactate; D5NS = dextrose 5% in normal saline; IV = intravenous; NAC = N-acetylcysteine; NPO = nil per os; NS = normal saline; PO = by mouth; POD = postoperative day; SQ = subcutaneous; SW/HCO₃ = sterile water with sodium bicarbonate 150 mEq/L; UFH = unfractionated heparin; X = dose(s) receive.

Table 2.

Roussel Uclaf Causality Assessment Method (RUCAM) Scoring.

	Hepatocellular		Cholestatic or mixed
	UFH	Amiodarone	UFH	Amiodarone
1. Time to onset	13 d from LHC, 3 d from OR, +2	2.5 d, +1	13 d from LHC, 3 d from OR, +2	2.5 d, +1
2. Course (after stopping the drug)	≥50% decrease in ALT between peak value and ULN in 3 d, +3		≥50% decrease in TBili between peak value and ULN in 4 d, +2
3. Risk factors	Alcohol, +1
4. Concomitant drug(s)	Known hepatotoxin (amiodarone) with a suggestive time to onset, –2	Concomitant drug (UFH) with a clear evidence for its role (eg, rechallenge), –3	Known hepatotoxin (amiodarone) with a suggestive time to onset, –2	Concomitant drug (UFH) with a clear evidence for its role (eg, rechallenge), –3
5. Exclusion of other causes of liver injury	Ruled out 4 of 6 causes from Group I (alcoholism and recent history of hypotension, shock or ischemia are possible), 0
6. Previous information on hepatotoxicity of the drug	Reaction labeled in the product characteristics, +2
7. Response to readministration	Positive, +3	Negative, –2	Positive, +3	Negative, –2
Total	+9 highly probable	+2 unlikely	+8 probable	+1 unlikely

Note. ALT = alanine aminotransferase; LHC = left heart catheterization; TBili = total bilirubin; UFH = unfractionated heparin; ULN = upper limit of normal.

Amiodarone is a well-established cause of clinically apparent liver injury with several possible patterns and varying latency.^10,11 Most amiodarone-related hepatotoxicity is chronic and associated with high doses of prolonged oral therapy. Acute hepatotoxicity may occur but is generally associated with high intravenous doses given to elderly or frail patients. Our patient was middle-aged and received standard doses of oral and intravenous amiodarone on POD 1-3 as described above. The initial pattern of our patient’s liver injury resembles the typical pattern of marked AST and ALT elevations (10- to 100-fold) and minimal increases in ALP within a day of intravenous amiodarone infusion followed by quick recovery (within days) after drug removal. However, our patient did not experience a typical reappearance of acute injury following a single reexposure of amiodarone on POD 9. Consequently, amiodarone was identified as an unlikely cause by the RUCAM system (Table 2). If we conservatively score UFH as a known hepatotoxin with suggestive time to onset instead of a concomitant drug with clear evidence for its role (eg, rechallenge), then amiodarone becomes a possible cause of hepatocellular injury and remains an unlikely cause of mixed injury. Either way, UFH remains the more probable cause of DILI using the RUCAM system.

Scoring by the Maria and Victorino (M&V) System of Causality Assessment also identifies UFH as the most likely cause of DILI in our patient (Table 3). ¹² Latency and response to rechallenge are the main reasons UFH is differentiated as the more likely cause with the RUCAM and M&V algorithms. Although the Naranjo algorithm was not specifically designed for identifying DILI, it also classifies UFH as the most likely cause of an adverse drug reaction in this patient (UFH, +6 probable vs. amiodarone, +3 possible) (Table 4). ¹³

Table 3.

Maria and Victorino (M and V) Clinical Scale Scoring.

	UFH	Amiodarone
I. Temporal relationship between drug intake and onset of clinical picture
Time to first clinical or lab manifestation	+3	+1
UFH: initial 10 d from LHC, subsequent 3 d
Amiodarone: about 2.5 d oral therapy for POAF
Time from withdrawal to onset of manifestations	N/A	N/A
Time from withdrawal until normalization (<2x ULN) of labs [<6 mo in cholestatic and mixed forms of injury, <2 mo in hepatocellular forms of injury]	+3	+3
UFH: 19 d; discontinued POD 7, normalized POD 26 (not checked between POD 12 and 26)
Amiodarone: 23 d from initial; 17 d from re-bolus
II. Exclusion of alternative causes
Viral hepatitis (HAV, HBV, HCV, CMV, EBV) excluded	−1	−1
Alcoholic liver disease possible
Biliary tree obstruction excluded
Pre-existing liver disease possible
Other possible (intraoperative acute hypotension)
III. Extrahepatic manifestations
Rash (no), fever (yes), arthralgias (no, epigastric/back/abdominal pain only), eosinophilia (no), cytopenia (no)	+1	+1
IV. Intentional or accidental reexposure
Positive (rise of at least 2x ULN for ALT or AP)	+3	0
V. Previous report in literature of cases of DILI associated with drug
Yes	+2	+2
Total	+11 possible	+6 unlikely

Note. ALT = alanine aminotransferase; AP = serum alkaline phosphatase; CMV = cytomegalovirus; EBV = Epstein-Barr virus; HAV = hepatitis A virus; HBV = hepatitis B virus; HCV = hepatitis C virus; LHC = left heart catheterization; N/A = not applicable; POAF = postoperative atrial fibrillation; UFH = unfractionated heparin; ULN = upper limit of normal.

Table 4.

Scoring Using the Adverse Drug Reaction Probability Scale (Naranjo) in Drug Induced Liver Injury.

	UFH	Amiodarone
Are there previous conclusive reports of this reaction?	Yes, +1	Yes, +1
Did the adverse event appear after the drug was given?	Yes, +2	Yes, +2
Did the adverse reaction improve when the drug was discontinued or a specific antagonist was given?	Yes, +1	Yes, +1
Did the adverse reaction reappear upon readministering the drug?	Yes, +2	No, –1
Were there other possible causes for the reaction?	Yes, –1	Yes, –1
Did the adverse reaction reappear upon administration of placebo?	N/A, +0	N/A, +0
Was the drug detected in the blood or other fluids in toxic concentrations?	Not measured, but unlikely, +0	Not measured, but unlikely, +0
Was the reaction worsened upon increasing the dose? Or, was the reaction lessened upon decreasing the dose?	Large dose on CPB, then standard prophylactic doses, +0	200 mg PO x4, then 150 IV x2, 400 PO x2; later 150 IV x1, +0
Did the patient have a similar reaction to the drug or a related agent in the past?	UFH w/ LHC 10 d pre-op but no labs, +0	No, +0
Was the adverse event confirmed by any other objective evidence?	Yes, +1	Yes +1
Total	+6 probable	+3 possible

Note. CPB = cardiopulmonary bypass; IV = intravenous; LHC = left heart catheterization; N/A = not applicable; PO = oral; UFH = unfractionated heparin.

The details of our patient’s alcohol consumption are unknown, but she was sober on admission. Currently there is no evidence suggesting chronic alcohol abuse is a risk factor for DILI. ¹ Chronic liver disease possibly increased the patient’s susceptibility to DILI, but there are limited data to declare this with certainty. ³ It is possible the patient experienced an icteric flare of her alcoholic hepatitis, which has been mistaken for DILI, but the Naranjo algorithm suggests an ADR is possible and the RUCAM scoring tool indicates a DILI due to UFH is highly probable.

Idiosyncratic drug reactions are rare and unpredictable. Our patient had multiple UFH exposures over a 2-week period ranging from her preoperative heart catheterization to postoperative VTE prophylaxis, but the most apparent DILI occurred later in the course of UFH exposure. The events on POD 3 may not have been the initial UFH-related DILI in our patient. Laboratory testing was not done over the 9 days between heart catheterization and the eve of surgery. It is possible our patient suffered an anamnestic response where she initially had an unrecognized and milder DILI after the heart catheterization but experienced a more severe and rapid return of injury upon subsequent UFH exposure. ³ It is also possible that a combination of potentially hepatotoxic medications, including heparin and amiodarone, led to the initially observed liver injury.

Our patient concurrently experienced multiple hypoglycemic events and episodes of acute liver injury on POD 3 and 7. These glycemic excursions were likely a consequence of liver injury instead of vice versa or a direct adverse effect of UFH. Impaired gluconeogenesis and decreased uptake of insulin by hepatocytes may lead to hypoglycemia during acute liver injury. Patients with diminished glycogen stores due to chronic liver failure may be at higher risk for hypoglycemia. Although our patient’s liver enzymes were normal preoperatively and her postoperative fasting blood glucoses were normal prior to these hepatotoxic events, she may have been at higher risk for hypoglycemia due to reduced glycogen stores and dysregulated glucose metabolism from her underlying liver disease.

According to Hy’s law, patients without cholestasis who experience a hepatocellular DILI with jaundice have a 10% or higher risk of death. ¹⁴ However, these criteria lack specificity for clinical predictions of unfavorable outcomes. A recent modification of Hy’s law improves its ability to predict which cases of idiosyncratic DILI are most likely to progress to acute liver failure or liver transplantation by identifying hepatocellular patterns of injury with a new ratio and utilizing a composite algorithm. ¹⁵ An increased risk of severe outcome is predicted by Hy’s law in our patient; yet, the modified Hy’s law suggests our patient was at lower risk of developing acute liver failure or requiring transplantation. Recognition of the temporal relationship of UFH exposure and hepatotoxicity, discontinuation of the potentially offending agent, and provision of supportive care resulted in a fortunate outcome with nearly complete hepatic recovery.

Conclusion

This case report describes a highly probable, but not definitive, occurrence of clinically apparent idiosyncratic hepatotoxicity in association with UFH. Although cases of DILI with UFH and LMWH are generally mild and asymptomatic, clinicians should include these agents in the differential etiology when treating patients with acute hepatotoxicity.