Keratin 18 is a Diagnostic and Prognostic Factor for Acute Alcoholic Hepatitis

Abstract

Background & Aims

Acute alcoholic hepatitis (AAH) is a major cause of liver-related morbidity and mortality; there are no good blood biomarkers for diagnosis or determining magnitude of cell death. Keratin 18 (KRT18, also called K18), found in epithelial cells, is released from hepatocytes upon death. We investigated whether level of K18 is a better marker of hepatocyte death than standard biomarkers and might be used to identify patients with AAH at risk for death within 90 days.

Methods

We analyzed data from 173 participants in a large trial performed at 4 medical centers. Participants with AAH were classified as severe (n=57, model for end-stage liver disease [MELD] scores above 20) or moderate (n=27, MELD scores from 12 to 19); 38 participants had alcohol use disorder with mild (n=28) or no liver injury (n=10); 34 participants had non-alcoholic steatohepatitis; and 17 participants were healthy (controls). We quantified serum levels of K18 using ELISAs and APOPTOSENSE kits.

Results

Serum level of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and the ratio of AST:ALT did not correlate with MELD scores. Patients with alcohol use disorder had higher serum levels of ALT than patients with severe AAH. Levels of K18M65 and K18M30 had statistically significant increases as liver disease worsened, as did the degree of necrosis (ratio of K18 M65:M30). The ratio of K18M65:ALT was increased in serum from patients with AAH compared with controls. Serum levels of K18 identified patients who died within 90 days with greater accuracy than commonly used static biomarkers.

Conclusions

There is a stronger association between serum level of keratin 18 and amount of hepatocyte death and liver disease severity than for other biomarkers (AST, ALT, and the AST:ALT ratio). The ratio of K18M65:M30 might be used as marker of mechanism of hepatocyte death, and the ratio of K18M65:ALT might be used to distinguish patients with AAH from patients with non-alcoholic steatohepatitis. Serum levels of K18 might be used to identify patients with severe AAH at risk for death.

Introduction

Acute Alcoholic Hepatitis (AAH) consists of a constellation of clinical and biochemical findings.¹ Unfortunately, available biochemical biomarkers are non-specific for the diagnosis of AAH, and there are no blood biomarkers that clearly distinguish AAH from other forms of liver disease^{2, 3}. There are some biomarkers (Maddrey DF, MELD, ABIC, and Glasgow) that are used to predict prognosis and severity of alcoholic liver disease (ALD)^4–6, however, these do not reflect the magnitude of cell death nor the form of cell death (apoptosis/necrosis), which may be important in distinguishing various forms of liver injury, or may impact drug therapy^{7, 8}. In the U.S., liver biopsies are not the standard of practice to diagnose AAH⁹,. Thus, new biomarkers are needed for diagnosing AAH, assessing the degree of hepatocyte death, and predicting mortality.

Extracellular keratin 18 (K18) is a marker for epithelial cell death, and serum levels can be very elevated following hepatocyte death^{7, 8}. Following cell death, K18 and other proteins may be lost to circulation due to loss of cell membrane integrity. K18 is a substrate for caspase 3, and the cleaved form of K18 (K18M30) reflects apoptosis¹⁰. K18M65 (whole protein) has been shown to be predictive of prognosis for patients recovering from acute liver failure^{11, 12} and has been suggested as a potential biomarker for AAH^{7, 8, 13}.

The goal of this study was to compare AST, ALT, and the AST:ALT ratio with K18M65 and M30 levels in determining the diagnosis of AAH and severity of liver disease as assessed by MELD in chronic heavy alcohol users. A second objective was to compare serum K18 concentrations with commonly used biomarkers of AAH in predicting three-month mortality.

Materials and Methods

Patient Recruitment

The study included 173 individuals aged 23–68. Among these, 84 were AAH patients diagnosed using clinical and laboratory criteria described by the NIAAA consortium¹. Using MELD (Model of End-Stage Liver Disease), AAH patients were classified as having severe (MELD≥20, n=57) or moderate (12≤MELD≤19, n=27) liver disease severity. Thirty-eight heavy drinking AUD subjects with similar drinking profiles to AAH patients with minimal (n=28) or no liver injury (n=10), based on ALT concentrations were studied. Additionally, 17 healthy controls were enrolled. No AUD patient or healthy control had clinical signs or symptoms of liver disease, and admission criteria for the study included normal serum albumin, normal bilirubin, and normal nutritional assessment for both of these control groups.

All study participants had a complete history and physical examination and laboratory evaluation upon study enrollment. Last, de-identified blood specimens from our tissue biorepository were obtained from 34 patients with biopsy-documented NASH (liver disease controls; mean NAFLD Activity Score: 4.2).

Study design

This investigation was a single time-point assessment of AAH, AUD patients, NASH disease controls and healthy participants. This study was approved by the Institutional Review Boards at all 4 participating institutions. All study participants were consented prior to collection of data and bodily samples.

We collected clinical data, relevant medical history, measures of severity of ALD, and drinking history (Lifetime Drinking History [LTDH], denoted in years)¹⁴. Severe AAH patients were further grouped by K18M65 (IU/L) levels as: (1) <641, (2) 641–2000, (3) >2000, as reported recently¹³, corresponding to groupings used to indicate no AAH, possible AAH, or definite AAH on liver biopsy. We evaluated differences between moderate and severe AAH, examined the association of K18 and clinical measures of disease severity, and assessed 3-month mortality (considered a primary endpoint for current/future studies of drug therapy for severe AAH).

Specimen collection and sample analysis

Blood samples were collected at enrollment (within 48 hours) from each subject in the study, following completion of the consenting process. Serum was apportioned into 1ml aliquots and stored at −80°C until use.

K-18 assay

Serum K18 concentration was determined by two ELISA assays (Dipharma Group, West Chester, OH). The M65® ELISA assay measures total soluble K18 released from dead cells, representing total epithelial cell death from any cause. The concentration of the antigen is expressed as units/liter (500 U/L, upper limit of normal).

The M30 Apoptosense ELISA provides reproducible quantification of apoptosis of K18 positive cells. This assay is a solid-phase sandwich enzyme immunoassay. The concentration of antigen is expressed as units/liter (250 U/L, upper limit of normal).

Analysis

Data are presented as mean±standard deviation (M±SD) in tables and figures, unless otherwise noted. Demographics, drinking history markers, and clinical differences were analyzed using one-way ANOVA with Bonferroni’s correction. We compared K18 protein levels among different groups using Kruskal-Wallis test for overall significance followed by Mann-Whitney U tests for group comparisons. K18M65 and M30 levels were correlated with clinical markers within each severity group using linear regression models and smoothing splines. Logistic regression models were used to determine which variables were significant in predicting NASH versus AAH, and which were significant in predicting 90-day mortality. Receiver-operating characteristic (ROC) curves were further constructed to examine the sensitivity and specificity of the new and standard biomarkers as diagnostic (i.e., predicting NASH versus AAH) and prognostic tests (i.e., predicting mortality), and to find an optimal cutoff for each test by maximizing the Youden index. Kaplan-Meier (K-M) survival curves were constructed corresponding to the cutoff points obtained from the ROC curves and compared using the log-rank tests. Significance was set at p<0.05.

Results

Patient demographics, drinking profile, liver injury, and mortality

There were significantly lower percentages of females in the severe and moderate AAH groups compared to the healthy controls (HC) (Table 1). Age and weight were similar in all the study arms involving alcohol abuse, and we did not find any other demographic differences. Lifetime drinking history was similar between the severe AAH, moderate AAH, and AUD group patients (drinking controls). As anticipated, MELD and Maddrey DF were significantly higher in the severe AAH group compared to the moderate AAH patients (Table 1). Among the 57 severe AAH patients, six patients were excluded from the survival analysis since they were lost to follow-up. The 3-month mortality for the remaining severe AAH patients was 27% (14/51), with seven patients dying at one month (all due to liver disease). Only one patient in the moderate group died, as he developed liver failure during his hospitalization.

Table 1:

Demographics, drinking history, and clinical markers in patients with severe acute alcoholic hepatitis (AAH), moderate AAH, alcohol user disorder (AUD), non-alcoholic steatohepatitis (NASH), and healthy controls. Data are presented as Mean±SD (standard deviation).

	Severe AAH (n=57)	Moderate AAH (n=27)	AUD (n=38)	NASH (n=34)	Healthy Control (n=17)	p-value
Sex (Female, %)	19 (33.3%)$&@	11 (40.7%)$&@	8 (21.1 %)$&	22 (64.7%)	13 (76.5%)	<0.001
Age	47.2±9.5	47.8±9.8	44.9±10.6	45.7±13.3	43.9±12.5	0.657
Weight (lb.)	192.6±46.9	202.6±76.7	173.1 ±39.7	NA	NA	0.219
LTDH (years)	20.2±13.4	20.0±12.5	17.3±9.3	NA	NA	0.949
MELD	26.04±4.94#*	16.96±2.16#	6.84±1.42	NA	NA	<0.001
Maddrey	60.13±28.82*	24.06±15.33	NA	NA	NA	<0.001
Total Bilirubin (mg/dl)	17.89±8.31n#*	5.75±3.69n#	0.61±0.39	0.6±0.4	NA	<0.001
Albumin (g/dL)	2.64±0.59n#	2.71±0.66n#	3.99±0.59n	4.4±0.36	NA	<0.001
AST (U/L)	127.2±63.2$&#	109.7±56.1$&#	76±50.8$	61.6±31.3$	21.7±6.4	<0.001
ALT (U/L)	46.7±21.7$&#	53.9±33.8$&	66.2±35.7$&	89.0±44.3$	18.4±10.0	<0.001
AST:ALT^	2.95±1.29$&#*	2.48± 1.39$&#	1.21 ±0.53&	0.72±0.24$	1.41±0.54	<0.001
Creatinine (mg/dL)	1.17±1.11#*	0.71±0.28	0.79±0.17	NA	NA	0.009
ALP (U/L)	199.8±146.1#	162.0±78.6#	93.03±38.69	NA	NA	<0.001
INR	2.02±0.55#*	1.48±0.36#	1.02±0.12	NA	NA	<0.001
CPT	10.89±1.35#*	8.70±1.54#	5.27±0.69	NA	NA	<0.001

^$significantly different from healthy control

^&significantly different from NASH

^#significantly different from AUD

^*significantly different from moderate AAH

^{^}indicates non-parametric test statistics. MELD: Model for End-Stage Liver Disease; AST: aspartate aminotransferase; ALT: alanine aminotransferase; ALP: Alkaline Phosphatase; INR: International Normalized Ratio; CTP: Child-Turcotte-Pugh. NA: Not applicable.

AST, ALT, and AST:ALT ratio

Compared to AUD subjects, serum AST was modestly but significantly higher in the severe AAH and moderate AAH groups, (Fig1a), while ALT was lower in the severe AAH patients. (Fig1b). The AST:ALT ratio showed the best separation between groups, but there was still major overlap (Fig1c). Thus, liver enzymes normally used to determine liver injury did not consistently distinguish severe AAH patients from subjects with AUD without clinically evident liver disease (Fig1).

Figure 1:

Liver injury markers in alcohol use disorder (AUD) patients, moderate acute alcoholic hepatitis (AAH), and severe AAH. Fig.1a: Aspartate aminotransferase (AST) levels. Fig.1b: Alanine aminotransferase (ALT) levels. Fig.1c: AST:ALT ratio. Data are presented as mean ± standard deviation. *p<0.05, **p<0.01, and ***p<0.001.

K18 levels

K18M65 and K18M30 were elevated above the normal range in most patients with moderate AAH.K18M65 was elevated in all patients with severe AAH. Further, severe AAH patients had significantly higher K18M65 and K18M30 levels compared to patients with moderate AAH (Fig2a and Fig2b). Both severe and moderate AAH patients had significantly higher K18 levels compared to AUD drinking controls (Fig2).

Figure 2:

K18 levels in alcohol use disorder (AUD) patients, moderate acute alcoholic hepatitis (AAH), and severe AAH. Fig. 2a: K18M65 levels. Solid line represents the upper limit of normal (−500u/l). The dotted line (2000 u/l) represents the level above which patients are classified as having AH based on criteria of Bissonnette and coworkers¹³. Fig. 2b: K18M30 levels. Solid line represents the upper limit of normal (250u/l). Observed data, mean and standard deviation are presented. *p<0.05, **p<0.01, and ***p<0.001.

In order to further understand the association of K18M65 and severity of AAH, we analyzed the distribution of AAH patients by K18M65 grading, described by Bissonnette et al [13]. Specifically, severe AAH patients were grouped by K18M65 (IU/L) levels as those with: (1) <641, (2) 641–2000, (3) >2000 IU/L. Using a Χ² test, we found significant association between severity of AAH and K18M65 grading: 51.9% of moderate AAH and 38.9% of severe AAH patients had K18M65 levels between 641 and 2000, and 25.9% of moderate AAH and 61.1% of severe AAH patients had K18M65 levels >2000 IU/L. All severe AAH patients had levels >641 IU/L.

We separated AUD patients into two subgroups—those with minimal liver injury and those without liver injury, as defined by elevated serum ALT. We compared ALT, AST, K18M65 and K18M30 levels to healthy controls. ALT (Suppl. Fig1a) and AST (Suppl. Fig1b) were elevated in minimal liver injury AUD patients compared to healthy controls and AUD patients without liver injury. There were modest statistical differences in K18M65 levels between healthy controls, AUD patients without liver injury and AUD patients with minimal liver injury (Suppl. Fig1c). Similarly, K18M30 was modestly but significantly elevated in AUD patients with liver injury compared to AUD patients without liver injury and healthy controls (Suppl. Fig1d).

Association of AST, ALT, AST:ALT ratio and K18 with MELD and Maddrey Scores

We next assessed traditional markers of liver injury (AST, ALT and AST:ALT ratio) and scales of liver disease severity (MELD and Maddrey DF). None of these biochemical markers of liver injury correlated with MELD or DF scores (Suppl. Fig2). However, MELD and, to a lesser extent, DF showed some association with K18 (Suppl. Fig3).

Association of K18 with AST, ALT and AST:ALT Ratio

We evaluated the potential associations of K18 with AST, ALT and AST:ALT ratio^{3, 15}. However, neither K18M65 nor K18M30 correlated significantly with any of these traditional biomarkers (Suppl. Fig4).

K18M65:M30 Ratio in Healthy Controls, Patients with NASH, or various stages of ALD

K18 measurements can identify necrosis vs. apoptosis^{7, 8}. Thus, K18 could help differentiate mechanisms of cell death and potentially help direct therapy. Patients with NASH had lower necrosis compared to apoptosis, and patients with alcohol abuse had numerically increasing necrosis as severity of liver disease increased (Suppl. Fig5a). A stepwise increase in necrosis was seen in alcohol abusing patients ranging from AUD patients with elevated ALT levels to severe AAH (Suppl. Fig5b). There was a significant trend for an increase in M65:M30 with increasing MELD (p<0.01). We also found that the severe AAH group was significantly different from AUD with no liver injury (p=0.001) and AUD with minimal liver injury (p=0.007) with regard to the M65:M30 ratio. Notably, severe AAH patients showed a shift toward necrotic type of hepatocyte cell death compared to both liver disease and drinking control groups--NASH, AUD.

K18M65:ALT Ratio and K18M30:ALT Ratio in AAH

As shown in Figure 1, the ALT level did not increase with severity of alcohol-induced liver injury. Thus, we postulated that the M65:ALT or the M30:ALT ratio might be a diagnostic marker for severe AAH. Most patients with severe AAH had an M65:ALT ratio >25, while most NASH patients had M65:ALT ratio that was below this cut off (Fig3a). Indeed, almost half of the patients with severe AAH had a M65:ALT ratio >100. Patients with severe or moderate AAH had significantly higher K18M65:ALT ratios compared to patients with the NASH or AUD. When grouped together, K18M65:ALT ratio was significantly higher in AAH patients than in NASH patients, suggesting this ratio may help to distinguish AAH from NASH (Fig3b). Importantly, severe AAH patients had higher ratios than moderate AAH patients (Fig3a). Using ROC curve, the K18M65:ALT ratio distinguished AAH from NASH (Fig3c) with a sensitivity of 0.971 and specificity of 0.829. Similar results were observed when the K18M30:ALT ratio was used (Fig3d–f). Thus, a very elevated K18 M65:ALT or K18M30:ALT ratio may be a biomarker for severe AAH.

Fig. 3:

K18M65:ALT and K18M30:ALT (postulated diagnostic biomarkers) across different groups. Fig. 3a: K18M65:ALT across all groups. Fig. 3b: K18M65:ALT in NASH vs. all AAH patients (moderate and severe). Fig. 3c: ROC curve for K18M65:ALT between NASH and AAH patients. Fig. 3d: Ratio of K18M30:ALT across all groups. Fig. 3e: K18M30:ALT in NASH vs. all AAH patients (moderate and severe). Fig. 3f: ROC for K18M30:ALT between NASH and AAH patients. Data are presented as mean ± SD. Statistical significance was set at p<0.05. ***p<0.001, and ****p<0.0001.

Biomarkers of Survival

We constructed ROC curves for K18M65 and K18M30 to predict three-month mortality (Fig4a,b). We compared these biochemical tests to standard- biomarkers of survival, the DF, MELD, ABIC, and GAHS (Fig4c–f). Both K18 biomarkers performed better than traditional biomarkers with regard to AUC. When we added readily-available covariates including age, sex, INR, Total bilirubin, and WBC count, there was moderate improvement in the K18 predictive ability (AUC) (Suppl. Fig6a–b).

Figure 4:

ROC curves for predicting 90-day mortality based on the K18 prognostic markers (K18M65, K18M30), four commonly used static biomarkers (DF, MELD, ABIC, GAHS). ROC curves for: K18M65 (Fig. 4a); K18M30 (Fig. 4b); DF (Fig. 4c); MELD (Fig. 4d); ABIC (Fig. 4e); GAHS (Fig. 4f).

We next performed Kaplan-Meier survival plots for K18M65 and K18M30 (Fig5a–b). Both K18 biomarkers outperformed the standard biomarkers, and these data suggest that the K18 may be a better indicator for mortality in patients with severe AAH (Fig5c–f). When we added the above noted covariates in the K-M curves for K18 evaluation, there was significant improvement in the K18 predictive ability (Suppl. Fig6c–d).

Figure 5:

Kaplan-Meier (K-M) plots for predicting 90-day mortality based on K18 prognostic markers (K18M65, K18M30) and four commonly used static biomarkers (DF, MELD, ABIC, and GAHS). K-M plots for: K18M65 (Fig. 5a); K18M30 (Fig. 5b); DF (Fig. 5c); MELD (Fig. 5d); ABIC (Fig. 5e); GAHS (Fig. 5f).

Discussion

The current project was undertaken as part of an NIAAA U01 consortium study to identify novel biomarkers and unique drug targets/therapies for AAH. This study demonstrates that traditional biomarkers of liver injury, AST and ALT, did not reflect severity of liver injury in the spectrum of moderate and severe AAH. The AST:ALT ratio modestly, but significantly increased with liver disease severity. However, both K18M65 and K18M30 increased with severity of liver disease as assessed by MELD and DF. All severe AAH patients had elevated K18M65 concentrations. K18 levels generally separated AAH patients from NASH patients, while AST and ALT did not. K18 levels did not appear to be more sensitive than AST and ALT at detecting early alcohol-induced liver injury. This was somewhat surprising, as we have previously utilized K18 (especially M65) to detect early liver injury in toxicant-associated steatohepatitis (TASH) that was present on liver biopsy but not detected by elevated AST and ALT levels¹⁶.

We observed a stepwise increase in necrosis in alcohol abusing patients ranging from AUD patients with elevated ALT levels to severe AAH. Studies from Jaeschke’s group⁷ also showed a similar pattern. We had observed relatively more necrosis than apoptosis in AAH and TASH, compared to greater apoptosis in NASH¹⁷. Of interest, Mueller and coworkers observed an increase in M30 in AUD patients undergoing alcohol withdrawal¹⁸. It is possible that beneficial physiological apoptosis is replaced by a more toxic necrotic process with advanced AAH.

Liver biopsies are not standard-of-practice to diagnose AAH in many regions, including the U.S.; the large multi-center STOPAH trial did not require a liver biopsy¹⁹. A potential problem in treating patients with presumed AAH is excluding those with decompensated alcohol-associated cirrhosis without active hepatitis, as these patients would likely not benefit from anti-inflammatory agents such as steroids or IL-1 receptor antagonists, but would incur their side effects. Bissonnette and co-workers evaluated several potential biomarkers for AAH in patients undergoing liver biopsy to diagnose suspected alcoholic hepatitis¹³. Keratin 18 (M65) was the most useful biomarker evaluated, and an upper cutoff of 2000 IU/L had a predictive value of 91%, while a lower cutoff point of 641 IU/L had a negative predictive value of 88%. Of patients not having AAH on biopsy, 95% had only alcohol associated cirrhosis. These investigators then studied a validation cohort with similar results. In our study, 61% of severe AAH patients had K18M65 values >2000 IU/L, and none had values <641 IU/L. Importantly, the degree of fibrosis on biopsy-documented ALD has recently been reported not to correlate with plasma K18 concentrations²⁰. Thus, K18 appears to reflect hepatocyte cell death, but not the degree of fibrosis. A small study of alcoholic cirrhosis and AAH patients showed associations between increased K18 levels and increased severity of liver disease, and with patient death⁷. A large study of acute-on-chronic liver failure also showed increased K18 levels with worsening liver disease²¹. Both studies showed greater necrotic/non-apoptotic cell death with increased liver injury and/or patient death.

Our study also supports the use of the K18M65:ALT ratio in the diagnosis of severe AAH. The ALT level increases with severity of liver injury in most liver diseases, but not in AAH⁸. Our AUD subjects had higher levels of ALT than did our severe AAH patients. Moreover, NASH patients can have histologically similar/identical findings to alcoholic steatohepatitis, but there were clear differences in both the K18M65:ALT and the K18M30:ALT ratios in these two disease processes.

We were able to evaluate K18M65 and K18M30 as prognostic biomarkers for 90-day mortality, which is increasingly being used as a primary endpoint for drug therapy for severe AAH and is the endpoint for the current multi-center NIAAA AlcHepNet study. We showed that these K18 biomarkers performed better than the four most commonly used static biomarkers (DF, MELD, ABIC, GAHS) to predict 90-day mortality. The K18 markers along with the other covariates, including age, sex, INR, total bilirubin, and WBC count, were even better predictors for 90-day mortality. Kaplan-Meier plots further indicated the superior performance of these K18 biomarkers to predict mortality. One potential reason that a more robust predictive response was not seen with all biomarkers was the fact that we evaluated an extremely sick group of AAH patients (mean MELD >26, and three patients with a MELD >35 survived). However, this likely reflects the patient population that will be seen in future AAH trials; those who are more severely ill and more survivors on aggressive therapies. Thus, new biomarkers are needed in these patient populations.

This study has some limitations. Future studies will be required to validate K18 as a diagnostic/prognostic biomarker. Liver biopsies were not performed as they are not the standard of care in the U.S.. Blood samples were stored at −80°C for up to four years, but stability concerns are unlikely. Lastly, follow-up K18 analyses were not performed as the goal of this study was to develop an admission diagnostic/prognostic biomarker.

In summary, AST and ALT concentrations may be helpful in determining early stages of AAH, but are not helpful in detecting disease severity. K18 levels correlated with MELD and DF, and non-apoptotic cell death (M65:M30 ratio) increased with severity of AAH. A very elevated K18M65:ALT ratio may be helpful in the diagnosis of AAH. Finally, both the K18M65 and K18M30 outperformed the four most commonly used predictive static biomarkers for mortality in AAH (DF, MELD, ABIC, and GAHS). In an era of personalized medicine with multiple diverse therapeutic options, it will be helpful to have biomarkers that reflect the extent and mechanisms of cell death in AAH. Our results need to be validated in similar patients with severe AAH. We predict that such biochemical biomarkers will be critical for treatment decision making in future patients with severe AAH.

Supplementary Material

1

Supplemental Figure 1: Aminotransferases and K18 in Early Liver Disease. Healthy Controls, AUD patients no liver injury and with minimal liver injury were grouped separately and compared regarding clinical markers of liver injury and hepatocyte death. Fig. S1a: ALT levels. Fig. S1b: AST levels. Fig. S1c: K18M65 levels. Fig. S1d: K18M30 levels. Data are presented as Mean ± SD. *p<0.05, and ***p<0.001.

Supplemental Figure 2: Association of clinical markers of acute alcoholic hepatitis. S1a: Association of AST and MELD in all AAH patients. Fig. S2b: Association of ALT and MELD in all AAH patients. Fig. S2c: Association of AST:ALT and MELD in all AAH patients. Fig. S2d: Association of AST and Maddrey DF in all AAH patients. Fig. S2e: Association of ALT and Maddrey DF in all AAH patients. Fig. S2f: Association of AST:ALT and Maddrey DF in all AAH patients. Data presented in black color depict moderate AAH patients and in red color are severe AAH patients. The Spearman correlation coefficient and the p-value are shown in each panel. The solid curve is the smoothing spline to capture the relationship between the two markers shown in each panel

Supplemental Figure 3: Association of K18 with the clinical indicators, MELD and Maddrey DF in AAH patients. Fig. S3a: Association of K18M65 and MELD in all AAH patients. Fig. S3b: Association in K18M30 and MELD in all AAH patients. Fig. S3c: Association of K18M65 and Maddrey DF in all AAH patients. Fig. S3d: Association of K18M30 and Maddrey DF in all AAH patients. Data presented in black depict moderate AAH patients and in red show severe AAH patients. The solid curve in each panel is the smoothing spline to capture the relationship between K18 and MELD (or Maddrey) score. Statistical significance was set at p<0.05.

Supplement Figure 4: Association of K18 protein fragments and clinical markers of liver injury, AST, ALT and AST:ALT of acute alcoholic hepatitis (AAH) in moderate and severe patients. Fig. S4a: Association of K18M65 and AST in all AAH patients. Fig. S4b: Association of K18M65 and ALT in all AAH patients. Fig. S4c: Association of K18M65 and AST:ALT in all AAH patients. Fig. S4d: Association of K18M30 and AST in all AAH patients. Fig. S4e: Association of K18M30 and ALT in all AAH patients. Fig. S4f: Association of K18M30 and AST:ALT in all AAH patients. Data presented in black color depict moderate AAH patients and in red color were severe AAH patients. The Spearman correlation coefficient and the p-value are shown in each panel. The solid curve is the smoothing spline to capture the relationship between K18 fragments and AST, ALT and AST:ALT, respectively.

Supplement Fig. 5: The M65:M30 Ratio in Liver Disease and in the Spectrum of AUD. Fig. S5a: Hepatocyte cell death ratio of K18M65:M30 across all the groups. Statistical significance was set at p<0.05. Fig. S5b: Boxplot for K18M65:M30 ratio in AUD (with and without liver injury), and moderate and severe AAH groups. The trend analysis showed significantly increasing trend of the M65:M30 ratio across the groups by severity in liver injury.

Supplement Fig. 6: Multivariate presentation of ROC curve and survival analyses (Kaplan-Meier (K-M) plots) for predicting 90-day mortality using K18M65 and K18M30 prognostic biomarkers in the severe arm patients. Covariates used are age, sex, INR, Total bilirubin, and WBC count. Fig. S6a: Multivariate K18M65. Fig. S6b: Multivariate K18M30. Fig. S6c: Multivariate K18M65. Fig. S6d: Multivariate K18M30.

Need to Know.

Background

Keratin 18 is released from dying hepatocytes. We investigated whether level of K18 is a better marker of hepatocyte death than standard biomarkers and might be used to identify patients with acute alcoholic hepatitis at risk for death within 90 days.

Findings

We found a stronger association between serum level of keratin 18 and amount of hepatocyte death and liver disease severity than for other biomarkers. The ratio of K18M65:M30 might be used as marker of hepatocyte death, and the ratio of K18M65:alanine aminotransferase might be used to distinguish patients with AAH from patients with non-alcoholic steatohepatitis. Serum levels of K18 might be used to identify patients with severe AAH at risk for death.

Implications for patient care

We identified blood biomarkers to identify patients with severe acute alcoholic hepatitis with a high risk of death.

Abbreviations

AUD

Alcohol use disorder

ALD

Alcoholic liver disease

ALT

Alanine aminotransaminase

AST

Aspartate aminotransferase

K18

Keratin 18

CS

Clinically significant

NCS

Clinically non-significant

LTDH

Lifetime drinking history

serum M65

Soluble K18 (indicative of necrosis)

serum M30

caspase-cleaved fragment of K18 (indicative of apoptosis)

Maddrey’s DF

Maddrey’s discriminant function

MELD

Model for end-stage liver disease

TBili

Total bilirubin