Diagnosing Wilson Disease in Acute Liver Failure: Comparison of Existing and Experimental Biomarkers

Abstract

INTRODUCTION:

In acute liver failure due to Wilson disease (ALF-WD), early and correct diagnosis is critical. Readily available biochemical parameters, e.g., ALT:AST ratio provide acceptable diagnostic accuracy. Recently, accurate nonceruloplasmin bound copper (ANCC), reflecting bioavailable serum copper and relative-ANCC (RelANCC) were developed. We investigated the diagnostic use of ANCC and RelANCC, and the validity of current diagnostic markers in ALF-WD.

METHODS:

Serum samples and data were collected through the US ALF Study Group registry. Etiologies included ALF-WD (n = 23), acetaminophen (n = 11), autoimmune (n = 10), drug induced liver injury (n = 13), HBV/HDV (n = 10), and other (n = 3). Standard clinical and biochemical parameters, serum copper (s-Cu), ANCC, and RelANCC were measured. Receiver operating characteristic analysis was performed.

RESULTS:

Both ANCC (cut-off ≥483 μg/L) and s-Cu (≥1369 μg/L) provided excellent diagnostic information for ALF-WD (area under the receiver operating curve [AUROC] 0.94 and 0.89). Low ALP (≤45 U/L), and ALT (≤52 U/L) individually provided diagnostic information for ALF-WD (AUROC’s > 0.89). Combining ALT with either s-Cu or ANCC further improved diagnostic accuracy, with the ANCC-ALT score being more accurate, AUROC 0.99, sensitivity 91% and specificity 100%. Adding ALP: total bilirubin ratio, AST:ALT ratio or serum zinc to ANCC did not improve accuracy. Median RelANCC was significantly higher in patients with WD (64%) than in non-WD ALF patients (22%) (P = 0.0001).

DISCUSSION:

Standard biochemistry provides excellent diagnostic information for identifying WD in the setting of ALF. Combining ALT with specific copper measurements significantly improved diagnostic use, with ANCC outperforming s-Cu. RelANCC may also be useful for WD diagnosis. We suggest adding s-Cu to the clinical algorithm in ALF.

Graphical Abstract

INTRODUCTION

Wilson disease (WD) is a rare, autosomal recessive disorder with a prevalence of 1:30.000 (1). In WD, dysfunction of the ATP7B protein, inhibits hepatobiliary excretion of copper (Cu) and impairs incorporation of Cu into ceruloplasmin (Cp) (2,3). Hepatic Cu is in equilibrium with serum copper (s-Cu), and therefore hepatic Cu retention leads to an excess of circulating nonceruloplasmin bound copper (NCC). NCC represents the bioavailable fraction of Cu which in WD is responsible for pathologic extrahepatic Cu deposition. Approximately 5% of patients with WD present with acute liver failure (ALF-WD) (4). Although WD is rare, it is the cause of approximately 3%–5% of all ALF cases (5–7). Correct and timely recognition of WD as the underlying cause of ALF is critical, as the transplant-free survival for patients with ALF-WD is particularly low (8).

The diagnosis of ALF-WD is challenging. The Leipzig scoring system was not designed to be used in the ALF setting (9). Cu measurements and genetic testing for ATP7B mutations take time. Urinary Cu measurements may be challenged by reduced kidney function in ALF (10,11). Kayser-Fleischer rings, low s-Cp and Coombs’ negative hemolytic anemia are less reliable in patients with ALF with cholestasis (12). Therefore, clinicians are typically left with standard routine biochemistry testing strategies to secure a timely and reliable diagnosis of ALF-WD. ALP:total bilirubin (tBili), AST:ALT ratios, and hemoglobin levels at presentation, provide accurate and readily available diagnostic information (12,13). However, some question their diagnostic accuracy (14,15). In pediatric patients with ALF, serum zinc (s-Zn) levels were significantly lower in ALF-WD (16). Another potentially useful marker is NCC, which has classically been calculated from s-Cu and s-Cp; however, the calculated NCC (cNCC) is unreliable, limiting its use for the diagnosis of ALF-WD (17,18).

Fortunately, new methods for the determination of NCC have been developed, including Accurate NCC (ANCC), which is based on the direct measurement of the Cu-containing form of Cp by strong anion exchange chromatography coupled to triple quadruple inductively coupled plasma mass spectrometry (SAX-ICP-MS/MS) (19). These newly implemented NCC methods have proven valuable in the diagnosis of WD without ALF (20–22).

The aim of the present study was to investigate the accuracy of biochemical screening for adult ALF-WD in the largest to date ALF-WD cohort. A novel use of relative ANCC (RelANCC)—the ratio of s-Cu to ANCC - and establishment of a cutoff for RelANCC in the setting of ALF-WD was also explored.

METHODS

Study population

The United States Acute Liver Failure Study Group (ALFSG) was a North American registry of 3,364 adult patients with ALF admitted to 33 academic medical centers across Canada and the United States, between 1998 and 2019. We used data and serum samples available at day 1 or day 2 of study enrollment from all patients having a diagnosis of WD as the cause of their ALF from 2006 to 2019. Controls with ALF of other etiologies were selected randomly. Detailed clinical, laboratory, histological, and outcome data were reviewed in detail by the ALFSG Adjudication Committee.

The ALF-WD etiology was confirmed in accordance with the Leipzig diagnostic criteria including some, but not necessarily all, of the following: hepatic Cu (.250 μg/g dry weight); 24-hour urinary Cu (.100 μg/d); liver biochemistries, and positive rhodanine histochemistry on liver biopsy. Similarly, the etiology of ALF in other patients without WD were adjudicated by the same Adjudication Committee (23).

Each ALFSG study site provided standard of care and biochemical data from local laboratories, except for ANCC, s-Cu, and s-Zn, which were analyzed in a specialized laboratory and RelANCC that was calculated from the data.

Participating clinical centers received institutional review board approval before enrolling patients in the registry. Written informed consent was obtained from patients’ surrogates due to encephalopathy, per the review board approval. The study was performed in accordance with the Helsinki and Istanbul declarations. Inclusion criteria were any degree of hepatic encephalopathy plus evidence of moderately severe coagulopathy (international normalized ratio [INR] ≥1.5) and a presumed time course of the patient’s hepatic illness of less than 26 weeks. Serum samples were stored for up to 17 years at −80 degrees Celsius.

Measurement of ANCC and relative ANCC

Elemental Cu-protein speciation was developed for the quantitation of holo-ceruloplasmin using SAX-ICP-MS/MS (19). In brief, s-Cu containing proteins were separated using SAX and gradient elution. The eluting protein Cu content was measured on-line using ICP-MS/MS and quantified by comparison to Cu-EDTA standards using linear regression analysis. ANCC was calculated by subtraction of the ceruloplasmin-bound copper from the total s-Cu, and the RelANCC was calculated as the percentage ratio of total Cu to ANCC, $\left(RelANCC=\frac{ANCC}{s-Cu}\times 100\right)$. The determination of s-Zn was similarly measured by ICP-MS/MS on the same samples used for measurement of s-Cu using an analytically validated method.

Statistical analysis

Nonparametric tests were used for comparison of groups (Kruskal-Wallis followed by Mann–Whitney test where appropriate). The Spearman correlation coefficient analysis was used to assess correlations. Stepwise logistic regression was used to identify variables with independent diagnostic information. Based on the variables identified by the logistic regression analysis, a simple score for predicting a diagnosis of WD was derived.

We used receiver operator characteristics (ROC) analysis to evaluate the overall capability of different variables and models to predict WD diagnosis. Using the ROC curve, a cutoff point was selected for predicting WD and corresponding sensitivity, specificity, positive predictive value and negative predictive value was calculated. The DeLong AUC-ROC comparison test was used for ROC curve comparison using all variables vs Total s-Cu as Gold Standard. The results are expressed as the median and interquartile range unless otherwise specified. A two-tailed P-value of < 0.05 was considered statistically significant. All statistical analyses were performed using STATA Version 18.

RESULTS

Patient characteristics

Seventy patients from the ALFSG cohort, 23 patients with WD (78%women) and 47 non-WD patients (51%women) were included (Table 1). The etiology of the non-WD ALF patients were acetaminophen (n = 11), autoimmune (n = 10), drug induced liver injury (n = 13), HBV/HDV (n = 10), Budd Chiari (n = 1), ischemia/shock (n = 1), and thyrotoxicosis (n = 1). For data analysis and presentation, patients with Budd Chiari, ischemia/shock, and thyrotoxicosis were grouped as Other.

Table 1.

Demographic data and select biochemistry by etiology of the ALF cohort

	WD	APAP	AIH	DILI	HBV/HDV	Other
n	23	11	10	13	10	3
Age	26 (8)	39 (19)	31 (10)	36 (17)	41 (13)	32 (23)
Sex (male)	5 (22%)	6 (55%)	5 (50%)	6 (46%)	5 (50%)	1 (20%)
ALT (U/L)	32 (43)	2,452 (2,984)	754 (640)	1,062 (2067.5)	1,218 (1479)	93 (1493)
AST (U/L)	180 (101)	3,072 (4,244)	466 (268)	1,066 (1777)	726 (525)	285 (3,873)
Alkaline phosphatase (U/L)	35 (40)	126 (60)	153 (64)	140 (107)	126 (18)	52 (41)
Albumin (g/dL)	2.5 (0.2)	2.8 (0.9)	3.0 (1.6)	2.5 (0.7)	2.7 (0.6)	3.8 (1.8)
Bilirubin (mg/dL)	34.8 (21.2)	5.4 (5.6)	26.2 (11.0)	19.7 (30.6)	26.5 (9.8)	26.2 (24.8)
Creatinine (mg/dL)	1.7 (2.9)	1.3 (1.0)	1.1 (0.5)	1.7 (1.5)	1.1 (1.8)	4.0 (2.3)
Sodium (mmol/L)	139 (6)	141 (11.5)	136 (7)	138 (6)	139 (11)	140 (9)
Hemoglobin (g/dL)	8.8 (2.5)	10.2 (4.1)	10.7 (3.1)	11.0 (2.4)	11.6 (2.1)	9.6 (3.6)
INR	2.8 (1.9)	3.3 (1.7)	3.1 (1.5)	2.9 (0.6)	3.8 (3.8)	2.4 (6)
MELD score	39.7 (20.2)	28.0 (13.8)	31.0 (8.9)	37.4 (15.8)	38.2 (22.1)	45.2 (13.4)
MELD-Na score	37.5 (16.8)	28.8 (11.4)	33.6 (13)	36.2 (15.0)	41.9 (22.8)	45.5 (13.1)
Ceruloplasmin (mg/dL)	21.2 (20.26)	20.0 (4.3)	16.7 (6.1)	17.9 (10.2)	15.6 (7.95)	19.7 (7.5)
Copper (μmol/L)	2,200 (1975)	767 (217)	751 (282)	759 (488)	702 (481)	837 (274)
ANCC (μmol/L)	1391 (1836)	144 (70)	196 (149)	165 (98)	141 (67)	154 (74)
RelANCC (%)	64 (36)	19 (8)	29 (7)	21 (8)	21 (16)	19 (17)
Zinc (μmol/L)	421 (251)	825 (592)	431 (192)	678 (505)	461 (147)	614 (133)

Data presented as median (interquartile range).

AIH, autoimmune hepatitis; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ANCC, accurate nonceruloplasmin bound copper; APAP, acetaminophen; DILI, drug induced liver injury; HBV, hepatitis B virus; HDV, hepatitis delta virus; INR, international normalized ratio; MELD, Model of end-stage liver disease; RelANCC, Relative ANCC; WD, Wilson disease.

The median model for end-stage liver disease (MELD) score for patients with ALF-WD was 40 and 32 for non-WD patients. MELD-Na was slightly but not significantly different from MELD within the cohort (Table 1). Seventeen patients with WD were transplanted and 5 died, whereas one survived without transplant.

Standard biochemistry

Standard laboratory values that were measured for all patients are shown in Table 1. Patients with ALF-WD had low levels of ALT, ALP, AST, and hemoglobin compared with patients with ALF without WD (Figure 1). Median tBili levels were higher in ALF-WD.

Figure 1.

Boxplot of serum values of candidate variables in patients with acute liver failure due to different etiologies. Note different units and ranges on Y-axes. Candidate values include s-Cu (a), ANCC (b), RelANCC (c) and ALT (d). Boxplot showing median values and interquartile range, whiskers showing minimum and maximum values. AIH, autoimmune hepatitis; ANCC, accurate nonceruloplasmin bound copper; APAP, acetaminophen; DILI, drug induced liver injury; HBV, hepatits B virus; HDV, hepatitis delta virus; RelANCC, Relative ANCC—the ratio of ANCC to s-cu; s-cu, total serum copper; WD, Wilson disease.

Copper related parameters

s-Cu and Cu related protein levels are shown in Table 1.

As expected, s-Cp varied widely in patients with ALF independent of etiology. The highest median s-Cp values were seen in patients with ALF-WD (21 mg/dL), whereas non-WD ALF patients had a median serum Cp value of 19 mg/dL (P = 0.006).

Ranges of s-Cu and ANCC for the separate patient groups with ALF are demonstrated in Figure 1. Subjects with WD had the highest levels for both, with none of the non-WD subjects having s-Cu level exceeding 1350 μg/L or ANCC above 343 μg/L.

There was a strong correlation between s-Cu and ANCC (r = 0.77, P = 0.0001) and this correlation was strongest in patients with WD (r = 0.90, P = 0.0001) and weaker for non-WD ALF patients (r = 0.52, P = 0.0003). Furthermore, in patients with WD, s-Cu correlated with both MELD score (r = 0.79, P = 0.001), INR (r = 0.62, P = 0.004), tBili (r = 0.77, P = 0.001), and AST (r = 0.47, P = 0.002). Likewise, ANCC positively correlated with both MELD score (r = 0.59, P = 0.006), INR (r = 0.67, P = 0.02), and tBili (r = 0.60, P = 0.003).

For non-WD ALF, s-Cu and ANCC only weakly correlated with tBili (r = 0.33, P = 0.02 and r = 0.45, P = 0.03, respectively) and inversely to INR (r = −0.41, P = 0.005 & r = −0.36, P = 0.02, respectively) but not to AST and ALT.

Median RelANCC was significantly higher in patients with WD than in non-WD patients, 64% compared with 22% (P = 0.0001). There was no significant difference in RelANCC between each of the non-WD etiologies for ALF (Figure 1, Table 1).

Standard laboratory s-Cu and s-Cp values were only available within the ALFSG database for 15 patients with WD and 5 non-WD patients. Consequently, standard cNCC values were only available for those same patients and thus direct comparison of cNCC to ANCC values was not possible.

Prediction of WD diagnosis (ROC analysis)

ROC analysis results are shown in Table 2 and Figure 2.

Table 2.

Summary of the diagnostic performance of biomarkers of Wilson disease in an acute liver failure setting

	Median WD-ALF	Median non-WD-ALF	AUC	Cut-off value	Sensitivity	Specificity	Correctly classified	LR+	LR−
s-Cu (μg/L)	2,201	759	0.89	≥ 1,369 ug/L	74%	100%	91.7%	—	0.26
ANCC (μg/L)	1394	163	0.94	≥ 483 ug/L	74%	100%	91.4%	—	0.26
RelANCC%	64	22	0.90	≥ 38.8%	83%	100%	94.3%	—	0.17
s-CU-ALTa			0.98	≥ 2	86%	98%	94%	38	0.15
ANCC-ALTa	19	−28	0.99	≥ 0.65	91%	100%	97%	—	0.10
Ceruloplasmin (mg/dL)	21	19	0.71	≥ 32 mg/dL	43%	98%	80%	20	0.58
Zinc (μg/L)	421	502	0.69	≤347 (ug/L)	35%	92%	74%	4	0.71
ALT (U/L)	32	1006	0.95	≤52 (U/L)	57%	100%	86%	—	0.43
AST (U/L)	180	778	0.87	≤236 (U/L)	81%	84%	83%	5	0.22
ALP (U/L)	35	131	0.89	≤45 (U/L)	61%	100%	87%	—	0.39
AST:ALT ratio	4.16	1.05	0.92	≥ 3.25	67%	98%	88%	29	0.34
ALP:tBili ratio	0.93	6.83	0.89	≤1.12	65%	98%	87%	30	0.36
ALP:tBili ratio < 1.12 and AST:ALT ratio > 3.25					52%	100%	84%	—	0.48

Formula for ANCC-ALT 5 0.012 × ANCC (μg/L) 2 0.03 × ALT (U/L). Formula for s-Cu-ALT 5 0.003 × s-Copper (μg/L) 2 0.02 × ALT (U/L).

^aP , 0.03 (DeLong AUC-ROC comparison).

ALF-WD, acute liver failure due to Wilson disease; ALT, alanine aminotransferase; ALP, alkaline phosphatase; AST, aspartate aminotransferase; ANCC, accurate nonceruloplasmin bound copper; AUC, area under the curve; LR−, negative likelihood ratio; LR1, positive likelihood ratio; RelANCC, Relative ANCC; s-Cu, total serum copper; tBili, total serum bilirubin.

Figure 2.

Receiver operating characteristic (ROC) curve for identifying Wilson disease etiology in patients presenting with acute liver failure, based on candidate variables. Total serum Cu (a), ANCC (b), s-Cu-ALT (c) and ANCC-ALT (d) each provided relevant diagnostic information. ANCC, accurate nonceruloplasmin bound copper; ANCC-ALT, combined score of ANCC and ALT; s-Cu-ALT, combined score of s-copper and ALT.

Standard biochemistry and ratios

Of the standard laboratory parameters, ALT provided the strongest diagnostic information (area under the receiver operating curve [AUROC] 0.95). A cutoff of ≤52 (U/L) yielded a sensitivity of 57% and a specificity of 100% (Table 2, Figure 2).

Low serum ALP also provided independent diagnostic value (AUROC 0.89) A ≤45 (U/L) cutoff provided sensitivity of 61% and specificity of 100%. AST, Hemoglobin, and s-Zn yielded AUROC’s of 0.87, 0.79, and 0.69, respectively. Combining values into an AST:ALT ratio or an ALP:tBili ratio did not improve the diagnostic capabilities (Table 2).

Sequentially screening patients with an optimized ALP:tBili ratio of < 1.12 and then screening with an optimized AST:ALT ratio of >3.25 did improve specificity to 100% but with a sensitivity of just 52%.

Copper parameters and combined models

s-Cu (cut-off ≥1369 μg/L) was quite effective in diagnosing ALF-WD with an AUROC of 0.89 (sensitivity 74% and specificity 100%). However, ANCC (cutoff ≥483 μg/L) was superior with an AUROC of 0.94 (sensitivity 74% and specificity 100%) (Table 2, Figure 2). RelANCC had an AUROC of 0.90 (sensitivity 83% and specificity 100%).

Using a weighted s-Cu and ALT or ANCC and ALT and combining them into one score using logistic regression analysis (s-Cu-ALT 5 0.003 × s-Copper (μg/L) − 0.02 × ALT (U/L); ANCC-ALT 5 0.012 × ANCC (μg/L) − 0.03 × ALT (U/L)), improved the diagnostic accuracy with AUROC’s of 0.98 and 0.99, respectively (Table 2, Figure 2). Adding s-Cp to any of the models did not improve the diagnostic yield for ALF-WD, nor did RelANCC instead of ANCC.

Use of zinc

s-Zn levels were significantly different between groups (P < 0.007). Patients with WD had the lowest median s-Zn levels in the cohort, but they were not statistically significantly different from levels in patients with autoimmune hepatitis (P > 0.6) or HBV/HDV (P> 0.7) (Table 1). s-Zn levels alone did not provide relevant statistical power for an effective diagnosis of ALF-WD with an AUROC of 0.69 (sensitivity 35% and specificity 92%) (Table 2).

Proposed algorithm to detect WD in ALF

In patients with ALF with low ALT, total s-Cu should be measured and the s-Cu-ALT score calculated. If the score is ≥ 2, assume WD. Apply the New Wilson Index prognostic score and transfer patient to a transplantation center for treatment (24). At the same time, perform genetic testing to support the diagnosis going forward.

DISCUSSION

In the setting of WD in ALF, there is great urgency to establish the diagnosis to begin specific treatments aimed at Cu reduction whereas evaluation and preparation for liver transplant is considered. In addition, a diagnosis of ALF-WD permits family screening to identify others at risk. Previous use of the ALFSG registry data demonstrated a high diagnostic accuracy for ALF-WD using the ALT:AST and ALP:tBili-ratios (12). Since then, additional patients with ALF-WD entered the registry, and new biochemical markers of bioavailable Cu were developed (16,19). We were therefore able to explore the diagnostic use of the above parameters for WD. We compared those parameters with a new ANCC assay in the setting of ALF.

Of 70 patients with ALF included from the ALFSG cohort, 23 had ALF-WD. In accordance with previous studies, we found a high proportion of women with ALF-WD, and patients with ALF-WD had lower ALT, ALP, AST, and hemoglobin levels and higher tBili levels compared with non-WD ALF. We also showed that ALT as a single biochemical marker represented one of the best standard diagnostic markers for ALF-WD (Table 2, Figure 2). Using the new assay for bioavailable copper, ANCC performed well in identifying ALF-WD (AUROC 0.94). However, a simple combined score of ANCC and ALT improved accuracy even further (AUROC 0.99). Using the more widely available s-Cu rather than ANCC also improved performance (AUROC 0.98), but with a slightly lower sensitivity.

As previously reported (12), s-Cp levels in ALF-WD were elevated above typically lower values seen in stable WD. Thus, it is important to recognize that s-Cp has no diagnostic value in ALF-WD.

In this study, the currently used ALT:AST and ALP:tBili ratio did not increase the diagnostic capability compared with the use of ALT alone—in contrast to previous findings (12). In addition, sequentially combining those ratios as previously proposed did not perform better than ALT alone, despite optimized cutoffs for this cohort (Table 2). Thus, a further simplification of routine testing for WD diagnosis in the setting of ALF might be possible. This conflict may be partly explained by lower acetaminophen hepatoxicity frequently in the current cohort compared with the earlier study (23% vs 37%), these patients present with the highest levels of ALT, AST, and ALP.

Another proposed scoring system in pediatric patients using a summarized score based on ALT, AST, AST:ALT ratio, ALP, and ALP:tBili ratio failed to yield any additional diagnostic information compared with ALT alone in the present cohort (13). This difference may be due to the conduct of our studies in adult patients.

Similarly, low s-Zn did not provide diagnostic accuracy in this adult cohort, in contrast to a previous study in pediatric patients (25).

Although low ALT is a known characteristic of ALF-WD some patients in the other category with Budd-Chiari, ischemia/shock or thyrotoxicosis also had relatively low ALT compared with other non-WD ALF patients. These patients were distinguished as non-WD by other Cu parameters. Although low ALT thus seems to be an appropriate marker of WD in the setting of ALF, we caution that this study only included few patients with ALF etiologies that also may present with low ALT. We therefore advocate the combination of low ALT and high copper values to diagnose WD in ALF.

Other similar direct copper measurements exist, using different methodology. To our knowledge, only one has been implemented in the clinic—exchangeable serum copper (CuEX) (26). CuEX is determined by complexing NCC with EDTA followed by ultrafiltration of the sample to isolate the NCC. This method has been criticized in recent studies, which instead support the ANCC methodology (22,27,28). Another limitation of the CuEX methodology is its lack of freeze stability (29,30), making it unsuitable for the purpose of this present study.

CuEX on its own has limited diagnostic value for WD, however, when CuEX is normalized to s-Cu it is known as the Relative Exchangeable Copper which has shown promise as a diagnostic tool for WD (30,31). We therefore similarly normalized ANCC to s-Cu, to generate the RelANCC. In this study, a cutoff value for RelANCC of ≥38.8% was diagnostic for ALF-WD. However, RelANCC did not add any further diagnostic value above ANCC on its own, AUROC 0.90 vs 0.94 (Table 2). This could be attributed to the elevated ceruloplasmin levels observed in WD-ALF, which differ from the classical diagnostic setting of WD where values are consistently lower than the normal range.

The study is strengthened by the large size of the cohort and the thorough clinical characterization and adjudication of patient cases from the ALFSG.

The relatively small sample size of patients with ALF-WD remains a limitation of research in rare diseases. However, to our knowledge this is the largest cohort of adult ALF-WD studied to date and differences between patients with WD and non-WD ALF patients were robust. The generalizability of the present findings is limited by the relatively small control group and lack of external validation cohort. Future studies are needed to expand on this issue. Validating these findings in an independent cohort would require a significant international effort, including access to biobank samples and clinical data from multiple centers. The fact that our reported baseline data was in line with existing literature, and robustness of the identified differences does offer some assurance for the generalizability of the findings. Validity could have been improved by using an age, sex, and outcome matched control group; however, due to the low age range in the ALF-WD group, matching was deemed infeasible. This choice may have introduced selection biases.

Finally, the long storage time for some of the frozen patient samples could theoretically affect the measurement of s-Cu and ANCC; however, that effect would have been similar for the ALF-WD and non-WD specimens. This supported the ability to determine s-Cu/ANCC on properly prepared and stored serum specimens for many years after sample collection, an important finding for the use of stored serum samples from patient repositories.

The study adds to our understanding of the use of standard biochemistry in the diagnosis of ALF-WD—and supports a possibility for further simplification of current biochemical screening for ALF-WD. In addition, the study supports the clinical use of Cu measurements such as s-Cu and ANCC in combination with standard biochemistry to greatly improve sensitivity. The measurement of s-Cu and not ANCC is widely clinically available and the AUROC of the 2 tests when combined with ALT is not very different (0.98 vs 0.99)—as such we can recommend the use of s-Cu-ALT to screen for ALF-WD. The higher sensitivity and specificity of the ANCC-ALT score could be of significance given the low prevalence of ALF-WD. Assuming a 5% prevalence of ALF-WD in the ALF population, the positive predictive value for s-Cu-ALT would be ≈ 70% vs 100% for ANCC-ALT.

In conclusion, this study supports the use of standard biochemistry to diagnose ALF-WD but questions some previous models. In addition, the study demonstrates the value of s-Cu and ANCC in WD especially in combination with low ALT. The s-Cu-ALT and ANCC-ALT models, had the highest diagnostic capabilities in ALF-WD. We suggest that s-Cu be added to the clinical workup of patients with ALF where the etiology is unclear and ALT levels are low. Further adoption of the ANCC assay and its validation in different clinical settings such as ALF will establish its use in the clinical arena.

Study Highlights.

WHAT IS KNOWN

• Transplant-free survival in patients with acute liver failure (ALF) due to Wilson Disease (WD) is particularly low.

• The diagnosis of WD in the setting of ALF is challenging.

WHAT IS NEW HERE

• This study includes the largest to date cohort of patients with ALF due to WD.

• A combined score of alanine aminotransferase and serum copper has high diagnostic accuracy. Accurate Nonceruloplasmin bound copper further improves accuracy, but is not yet widely available.

ABBREVIATIONS:

ALP

alkaline phosphatase

AST

aspartate aminotransferase

AUC-ROC

area under the receiver operating characteristic curve

HBV

hepatitis B virus

HDV

hepatitis D virus

MS

mass spectrometry

WD-ALF

Wilson disease–associated acute liver failure