LKM Immunofluorescence Is Associated with DILI, Especially after Metamizole Intake

A frequent side effect of drugs is hepatotoxicity leading to drug-induced liver injury (DILI) in susceptible individuals. DILI occurring independently from drug dose or duration is referred to as idiosyncratic DILI in contrast to intrinsic DILI [1]. When suspecting idiosyncratic DILI the causative agent can be difficult to identify as patients often use multiple medications or might not remember all agents taken in the past and biomarkers for the diagnosis of DILI are missing. DILI has been described for a broad spectrum of substance including drugs, herbal, and dietary supplements. Studies regarding the aetiology of acute liver failure identified drugs as the most frequent cause of acute liver failure in the USA and European countries [2]. For example, two of the most common prescribed drugs causing DILI include amoxicillin/clavulanate or diclofenac [3‒5]. Regarding the pathomechanism of DILI, different underlying mechanisms have been proposed such as immune-mediated responses, reactive drug metabolites or mitochondrial injury [6]. Also, as most drugs associated with DILI interact or are metabolized by cytochrome P450 isoforms a crucial role of the cytochrome P450 enzyme family in the occurrence of DILI has been discussed. Interestingly, CYP P450 isoforms 2D6 and 2C9 are known targets for liver/kidney microsomal (anti-LKM) autoantibodies. Previous studies described the occurrence of anti-LKM-antibodies in DILI implicating a role in drug-induced liver injury and hypersensitivity reactions against certain drugs [7, 8] as well as in patients with hepatitis C infection [9]. However, in the routine diagnostic work-up of patients with liver disease, anti-LKM-antibodies are currently only of diagnostic value in the diagnosis of autoimmune hepatitis (AIH) as anti-LKM-1 is an autoantibody associated with and well characterized in AIH type 2 recognizing a specific aminoacid region in CYP2D6 [10]. The typical immunofluorescence pattern in these patients is characterized by staining of the third portion of the proximal tubules on rodent kidney section and a homogenous staining of the hepatocyte cytoplasm on rodent liver section [11].

In the diagnostic work-up of patients with liver injury, it might be difficult to distinguish between AIH and drug-induced autoimmune-like hepatitis (DI-ALH), which is considered a drug-induced liver injury with laboratory and/or histological findings that are indistinguishable from AIH [12]. Yet, as distinguishing DI-ALH from AIH is crucial for the subsequent therapy, it is important to define biomarkers for the distinction of these two similarly appearing but very different entities of liver injury. Therefore, as antigens of anti-LKM-antibodies also include members of the CYP enzyme family, we therefore were interested, whether the determination of autoantibodies is of diagnostic value when suspecting DI-ALH. We therefore screened our Hannover liver injury database consisting of 63,300 cases [13] for LKM-immunofluorescence titre positivity in order to analyse the underlying etiologies of LKM-immunofluorescence positivity.

Here, 63,300 cases of our local Hannover liver injury database were screened for high LKM-immunofluorescence positivity (titre ≥1:160). As reported previously [13], we used the following strategy for case-inclusion: AST or ALT >3 upper limit of normal (ULN) or AP or TBI >2 ULN at any time point during hospitalization. We identified 82 cases with a positive high LKM-titre with elevated liver enzymes treated in our hospital between January 2008 and July 2021. Cases with previous liver transplantation, polytoxicomania, and acute viral hepatitis were excluded. Cases with a recurrent hepatitis such as AIH flares or cases with a drug re-exposure were pooled and analysed as one event. After application of those criteria, 64 patients with high titre LKM-immunofluorescence positivity formed the study population (Fig. 1).

Medical reports were analysed regarding the underlying causes for liver enzyme elevation, laboratory values, histological results, and medication. As we screened for high LKM-titres, the other antibody titres for autoimmune-mediated liver diseases (ANA-titres, AMA-titres, LKM-titres) were also considered positive when ≥1:160. During the evaluation period of this study, immunofluorescence testing was performed by experienced technicians within our routine clinical laboratory using the methodology published in the guidelines by the Committee for Autoimmune Serology of the International Autoimmune Hepatitis Group [14]. Briefly, 5 µm sections of snap frozen rodent kidney, liver, and stomach were used for immunofluorescence and incubated with patient sera in dilution series starting from 1:40 to 1:320. After incubation with a secondary fluorochrome-labelled antibody, tissue sections were analysed using a fluorescence microscope (Olympus BX60, Olympus Corp. Tokyo, Japan). Patients with typical immunofluorescence pattern [11] were subsequently tested for LKM 1–3 by immunoblot or ELISA. All cases of the study cohort were re-evaluated for the probability of DILI based on laboratory patterns, analysing of the medical reports and RUCAM scoring as well as histories of previous and current medication potentially causing DILI. DILI categorisation was performed by calculating the R ratio value (ALT/ULN ÷ AP/ULN) with R ≤2 defined as cholestatic, R ratio between 2 and 5 as mixed and R ≥5 defined as hepatocellular injury according to the current guidelines [15]. Whenever applicable, the RUCAM score was calculated for causality assessment of DILI with categorisation the suspicion of DILI to highly probable (score >8), probable (6–8), possible (3–5), unlikely (1–2), and excluded (score ≤ 0) [16]. Analogous to the studies of Weber et al. [17] for metamizole-associated liver injury one point was given for known hepatotoxicity. Patients were referred to 2 different groups: metamizole-associated DILI and a non-metamizole-associated liver injury group with positive LKM-immunofluorescence, which also included non-DILI patients. We used the following definition for Hy’s law: ALT or AST >3 combined with elevation of TBI >2 without AP elevation (AP <2 ULN) [15].

Quantitative variables were tested for normal distribution (Kolmogorov-Smirnov test) when n ≤ 30 and were analysed parametrically (unpaired t test) and non-parametrically (Mann-Witney-U test). In cases with large sample sizes (n > 30 for each group) no pre-testing was performed and an unpaired t test was used because mean values for large sample sizes are approximately normally distributed according to the central limit theorem. All quantitative values are reported as ULN. Medians are reported as median + interquartile range. Means are reported as means + SD of ULN. Homogeneity of variances was tested for t tests. When equal variances were given, the unpaired t test was applied while in cases with unequal variances, Welch’s t test was used. Frequencies were analysed by application of the Chi-squared test or the exact fisher test for cases with expected cell frequencies below 5. p values <0.05 were considered as significant. Cramer-V >0.1 were considered to be relevant. Statistical analyses were performed with SPSS 28 (IBM Corp. Released 2021. IBM SPSS Statistics for Windows, Version 28.0. Armonk, NY: IBM Corp). For graphs and figures, Microsoft excel and Microsoft Power Point (2021, Version 2108, Microsoft Coorporation) were used. All analyses were in accordance with our institutional data protection board.

We screened our database and identified 64 patients with a positive high LKM immunofluorescence. Based on the medical histories, we retrospectively re-evaluated potential causes for liver injury in those patients. After retrospective re-evaluation, liver injury in 53 of the 64 patients was classified as drug induced (82.8%) according to available medical histories. The remaining 11 patients were classified as AIH (n = 6, 9.3%), AIH DD DILI (n = 2, 3.1%), alcoholic liver disease (n = 2, 3.1%), and acute or acute-on-chronic liver failure of unknown reason (n = 1, 1.5%). Notably, in 62% of the cases with liver injury induced by drugs (n = 33 of 53) we identified metamizole as causing drug. In the other 20 patients, in which drugs were most likely causative for liver injury, only in 5 patients a possible causing drug could be identified (once each: nitrofurantoin, terbinafine, amoxicillin/clavulanate, pantoprazole, and simvastatin). In the 15 remaining patients’ drugs were most likely causative but remained unknown. Based on this re-evaluation, we compared metamizole-associated liver injury (n = 33) with non-metamizole-associated liver injuries (n = 31) for analysis of baseline characteristics. Patients with metamizole-associated liver injury were predominantly female (n = 27, 81.8%) and median age was 38 (24.5–49.5) years compared to 36 years (29–50) in the non-metamizole-associated liver injury group, respectively. Almost all cases with metamizole-associated liver injury showed a predominantly hepatocellular pattern (n = 32, 97%) with 19 patients (57.6%) fulfilling Hy’s law criteria. Re-exposure to metamizole was documented in 5 patients and resulted in liver enzyme elevation in all patients. In 1 patient, the increase in liver enzymes was accompanied by agranulocytosis. Maximum ALT levels were 67.5 ± 40.4 ULN and significantly higher in patients with metamizole-associated liver injury compared to 42.1 ± 38.2 ULN in patients with non-metamizole-associated liver injury (p = 0.012). There was no difference in maximum TBI between both groups (17.2 ± 10.2 vs. 17.3 ± 10.9, p = 0.972). We also assessed the initial diagnoses made by the treating physicians. Whereas about half of the patients with metamizole-associated liver injury were initially correctly diagnosed with DILI (n = 19), in about 20% of the patients the initial diagnosis could not distinguish between DILI and AIH (n = 7, 21.2%). Initial diagnoses in patients with metamizole-associated liver injury and non-metamizole-associated liver injury are shown in Figure 2. Detailed baseline characteristics are shown in Table 1.

We also addressed the question, whether patients with liver injury and positive LKM-antibodies also develop other autoantibodies associated with autoimmune mediated liver disease. High titres for ANA were detectable in 25% (n = 16) of all patients. Notably, ANAs were almost exclusively detected in patients with liver injury induced by drugs (15/16). AMA or SMA was less frequent, only 1 patient of the total cohort presented with high AMA and 5 patients with high SMA. Furthermore, we analysed the expression pattern of LKM-antibodies regarding the LKM-subgroups LKM-1, LKM-2, and LKM-3. Although all patients of our cohort were positive for LKM immunofluorescence, only in 5, 1, and 6 patients respective antibodies of already established LKM-subgroups were detectable. Thereby, LKM-1 was almost only detected in patients with AIH (4/5), the presence of LKM-2 and LKM-3 could not be attributed to one specific cause of liver injury. Additionally, antibodies directed against CYP 1A2, which can be found in patients with diyhdralazine-induced hepatitis [18] were detected in 3 patients with DILI, one caused by pantoprazole and in 2 other patients with DILI of unknown cause. Detailed autoantibody profiles are shown in Tables 2 and 3.

Liver biopsy was performed in 52 of the 64 patients (81.3%) (29 in patients with metamizole-associated liver injury and 23 in non-metamizole-associated liver injury). In patients with metamizole-associated liver injury, infiltration with lymphocytes was described in 88.9% (24/27) followed by neutrophilic granulocytes and eosinophilic granulocytes with 66.7% and 54.2%, respectively. Plasma cells were observed in 63.6% and mostly described as isolated (36.4%) and moderate (22.7%) while severe infiltration was only seen in one biopsy (4.5%, 1 of 22 biopsy results with information on plasma cells). Ceroid macrophages were reported in 47.8% (11/23) in the metamizole-associated liver injury group. Steatosis was described in 34.8% of the metamizole-associated DILI biopsies and median necrosis was described with 50% (27.5–60). The histological pattern was similar in patients with non-metamizole-associated liver injury. Biopsy results are shown in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000545507).

Of the 53 patients in the total LKM⁺ liver injury cohort 13 patients required liver transplantation or died, 1 patient died after liver transplantation. Prednisolone was administered in half of the patients (27/53), but did not affect survival as approximately 30% (8/27) of the patients treated died or required liver transplantation compared to 19% (5/26) of those without treatment (p = 0.38). Of the 33 patients with metamizole-associated liver injury 24 (72.7%) patients recovered, while 8 (24.2%) patients required liver transplantation and 1 (3%) patient died. Prednisolone was administered in half of the patients (18/33) with metamizole associated liver injury and did not influence survival as 27% of the patients with or without steroids (5/18 vs. 4/15) died or required liver transplantation (p = 1.0).

In the metamizole liver injury group, aminotransferase levels and INR were significantly higher than in patients that died or required transplantation at both the initial determination and their maximum. Hy’s law was fulfilled in 77.8% of the patients that underwent liver transplantation or died while only in 50% of the group of patients that recovered (p = 0.24). Kidney function at the initial investigation did not differ between the two groups. However, we observed significantly higher urea and creatinine levels at their maximum in patients that required liver transplantation or died (p < 0.001 both). Results are shown in Table 4.

Drug-induced liver injury (DILI) is a rare event in the general population. The determination of the real incidence of DILI is difficult, but several studies in different countries estimated an incidence from 2 to 3 cases/100,000 individuals annually in western countries. On the other hand, the incidence of DILI is increasing in hospitalized patients, especially in patients with elevated liver enzymes [19]. However, the diagnosis of DILI is somehow challenging as it can mimic almost every other liver condition and as established pathognomonic biomarkers are missing it is still a diagnosis of exclusion [20]. One phenotype of DILI is termed DI-ALH, as it can present with laboratory and histological findings very similar to idiopathic AIH. However, the distinction of AIH and DI-ALH is crucial for the correct management of patients, as usually AIH requires long-term immunosuppression, whereas DI-ALH can be managed without long-term immunosuppression. In the late 1980s and 1990s, several studies reported the presence of LKM-antibodies in patients with DILI caused by halothane, dihydralazine, tienilic acid, and anticonvulsants [18, 21, 22]. Furthermore, case reports described AIH specific LKM-1 in AIH following haematopoietic stem cell transplantation [23, 24]. However, up to date the diagnostic value of LKM-antibodies in case of liver injury is only well-established in AIH. Therefore, we aimed to analyse the prevalence of LKM-antibodies in patients with DILI and its diagnostic potential regarding the distinction of DI-ALH from idiopathic AIH. In this study, we made several important findings.

• Positive LKM immunofluorescence occurs in patients with DILI,

• histological features of DILI with LKM-antibodies are consistent with idiopathic AIH, therefore matching criteria for the diagnosis of DI-ALH,

• in case of LKM+ DI-ALH, metamizole is the most frequent causing drug,

• LKM+ DI-ALH are mostly negative for known LKM-subtypes LKM-1, LKM-2, LKM-3 as well as AMA, SLA and SMA, and

• patients at risk for liver transplantation or death have more impaired coagulopathy and higher aminotransferases at first determination than those who recover from metamizole-associated LKM+DILI.

When looking into the underlying liver disease of our patients with positive LKM immunofluorescence, we observed that remarkably more than half of the liver injuries with LKM positive immunofluorescences could be attributed to drugs (82.8%, 53 of 64 patients). Moreover, in approximately half of the patients with LKM positive immunofluorescence and DI-ALH, we could identify metamizole as the most probable causing drug (n = 33 of 53). Metamizole, also known as dipyrone is a very potent antipyretic and analgetic drug, but serious side effects like agranulocytosis had led to its withdrawal from several countries [25]. On the other hand, it is still frequently used in Germany, Italy, Spain, Switzerland, and Brasil [25]. In Germany, the prescription of metamizole increased fifteen-fold between 1991 and 2018 [26]. Even though frequently used, only 1 case of DILI caused by metamizole was reported until 2002, yet metamizole-associated DILI has just recently regained attention [17, 27‒31]. Up to date, the pathogenesis of the liver injury induced by metamizole is still unclear, but several studies suggested immune-mediated mechanisms [17, 28]. Although the occurrence of ANA and AMA antibodies have been described in the context of DILI, LKM-antibodies have only in rare cases been linked to DILI, even though most of the drugs causing DILI are metabolized via cytochrome P450 isoform, a known target of autoimmunity in several chronic liver diseases including drug-induced hepatitis [7]. As metamizole is also metabolized by cytochrome P450 in the liver mainly by cytochrome P450 [25], it is conceivable that the metabolism of metamizole by CYP P450 or neoantigen formation with secondary loss of tolerance to endogenous cytochromes play a role in LKM immunofluorescence positivity and liver injury observed in this study. The liver injury due to metamizole has been associated with severe liver injuries that show characteristics of autoimmunity [17, 28]. Possible immune-mediated mechanism or immune-allergic mechanisms are also supported by a study of Federmann et al. who described a case of metamizole-associated DILI with development of an exanthema [27]. Preveden et al. [29] detected a SIRS like syndrome in a patient with metamizole-associated DILI. Krisai et al. [30] hypothesized that immunological mechanisms may play a role in metamizole-associated DILI because of positivity of a lymphocyte transformation test. Notably, even though half of the LKM+ DI-ALH could be attributed to metamizole, none were positive for the LKM-subgroups LKM-1 and LKM-2. As expected, LKM-1 and LKM-3 antibodies were almost exclusively detected in patients diagnosed with AIH. Moreover, in 15 of 20 patients with LKM+ DI-ALH not attributed to metamizole, the causing agent could not be identified by retrospective chart review. As metamizole is a very popular and frequently prescribed drug in Germany and liver injury caused by metamizole was neglected until 2020, it appears possible, that in these cases metamizole was taken but not reported by the patients and also not specifically excluded by the treating physicians. Based on this observation, it is conceivable that metamizole DI-ALH induces indeed a specific LKM-antibody, but of currently unknown epitope. Furthermore, DI-ALH with positive LKM immunofluorescence were not exclusively related to metamizole intake as nitrofurantoin, simvastatin, terbinafine, or amoxicillin/clavulanate also were observed as potential triggers. This is in line with other studies that described features of autoimmunity similar to AIH for nitrofurantoin, simvastatin, terbinafine, and amoxicillin/clavulanate as well as proton pump inhibitors [12, 32, 33]. The observation of LKM-antibodies reacting with unknown epitopes in liver disease other than AIH was also made by Muratori et al. [9] In this study, sera from HCV patients showing typical immunofluorescence patterns for LKM-1 bound to different polypeptides in immunoblot than sera from patents with confirmed AIH; however, a specific aminoacid sequence of these peptides were not identified in this study. Interestingly, albeit high LKM immunofluorescence was detected, high ANA-titres were only found in 25% of cases with metamizole-associated LKM+ DI-ALH which is less than reported by Weber et al. [17] (72%) or Sebode et al. [28] (50%). This might be explained by the threshold we used for ANA positivity of ≥1:160 in order to avoid overinterpretation of only slightly elevated cases. The hepatocellular biochemical injury pattern in the majority of patients and the predominance for female gender we found in this study is in line with previous studies [17, 28] and described for cases of DI-ALH [12]. Infiltration with lymphocytes was observed to a similar degree as of around 88% as in the study by Sebode et al. [28] (76%). While in the aforementioned study, eosinophilic infiltration was rarely seen and plasma cells were absent, we found eosinophilics and plasma cells in half of the studied liver biopsies, but eosinophilics and plasma cells mostly were not predominant and described as isolated in our cohort. However, as we selectively included patients with high titres of LKM-antibodies, we might have introduced selection bias regarding entities of liver injury with autoimmunity features. The use of steroids in the management of DI-ALH is controversially debated among experts [12]. According to EASL-guidelines, in suspected DI-ALH, the first treatment step is the withdrawal of the causing drug and in case of lack of improvement the use of corticosteroids is recommended. Several studies reported a more rapid response to glucocorticoids in AIH patients when IgG4 was elevated. Even though these studies were performed in a limited number of patients and the proportion of patients with elevated IgG4 levels was highly variable ranging from 3 to 35% among studies [34], IgG4 might serve as a useful marker in prediction of steroid response in patients with liver damage with features of autoimmunity. However, as clear indicators and precise treatment protocols for the use of steroids in DI-ALH patients are still missing, steroid treatment is basically up to the personal experience of the treating physicians. Also, in our cohort in about half of the patients the decision for steroid treatment was made, but we did not observe an impact of steroid treatment. Even though steroid treatment does not necessarily improve the clinical course of DI-ALH, the lack of treatment response to steroids in our patient cohort might also be attributed to the fact that as a tertiary transplant centre patients were referred after the right time point for prednisolone administration had already passed. However, when steroids had been implemented in the management of suspected DI-ALH physicians should make an attempt of withdrawal of steroids, as DI-ALH usually do not need continuous immunosuppression, but it has to be noted, that the distinction of DI-ALH and idiopathic AIH can be very difficult. Also in this study, 12% of the patients were initially diagnosed with AIH and the reanalysis of these cases revealed that metamizole-DI-ALH was the more likely diagnosis. This emphasizes that the diagnosis of DI-ALH can be challenging. In cases in which AIH is a possible differential diagnosis, prednisolone administration is an important part of the therapy. As it is difficult to distinguish AIH from DI-ALH, the administration of prednisolone should always be an individual decision. However, if DI-ALH is a possible differential diagnosis to AIH, early taper of immunosuppression should be discussed to avoid side effects of long-term immunosuppression in cases in which DI-ALH is possible.

Another important finding of our study was that a quarter of patients with metamizole-associated DI-ALH required liver transplantation. This is higher compared to previous studies, where rates of transplantation were below 10% [17, 28]. The reason for the higher frequency of severe courses in our cohort remains unknown. Different referral strategies to tertiary centres cannot be excluded even though all studies were performed in Germany. Patients that required liver transplantation or died had significantly higher aminotransferases at initial investigation. In addition, impaired coagulation was also associated with transplantation and death and should therefore be regarded as a risk factor for bad outcome.

Our study has several limitations including the single centre and the retrospective study design which, despite using of causality assessments, carries the risk of uncertainty concerning the DI-ALH diagnosis we made after reanalysing the patient’s medical records. Furthermore, as we selected patients with positive LKM immunofluorescence, those patients with metamizole-associated DI-ALH or metamizole DILI and negative LKM immunofluorescence are missing. On the other hand, it is also possible that metamizole intake was not reported in some patients and therefore metamizole associated DI-ALH is a cause for some of the DILIs with an unknown drug in the non-metamizole liver injury group. Moreover, as we are a tertiary transplant centre it is possible that patients were referred to our centre after the right time for prednisolone administration was passed and that thus prednisolone was initiated too late to develop a curative effect. Furthermore, as distinguishing between AIH and DI-ALH, especially in patients with positive markers of autoimmunity may be challenging, decision of prednisolone administration should be individually determined and this study did not aim to question the use of steroids in managing DI-ALH. However, we showed in this study, that the pattern of LKM-antibodies might help in distinguishing idiopathic AIH from DI-ALH and that in cases of LKM positive DILI metamizole should be suspected as possible causative agent. Further studies should focus on the underlying mechanism of metamizole-associated DI-ALH. This is of particular interest because the drug is widely prescribed especially in Germany with millions of daily doses and nowadays is an indispensable component of pain medicine that helps avoiding opioids and therefore opioid related addiction.

DI-ALH, especially after metamizole administration, can be a reason for a positivity in LKM immunofluorescence tests mostly with negative LKM-ELISA or Western blots. It is important to establish a timely diagnosis as the outcome may be severe in more than 1 quarter of patients.

We are very grateful to Nicolas Simon of our Enterprise data warehouse for his support in data collection. We thank Steffi Loges and Nicole Henjes for their support regarding the autoantibody data.

This study was approved by the Ethics committee of Hannover Medical School (Approval No. Nr.10301_BO_K_2022) and was conducted in accordance with the World Medical Association Declaration of Helsinki. Written informed consent was obtained from all patients.

Elmar Jaeckel is a co-inventors of the patent for the use of anti-HIP1R/BSA antibodies for the diagnosis of AIH (European Patent [EP3701264 B1]), US Patent application (Application number: 16/754,006). Bastian Engel received a provision of laboratory tests free of charge for research-use for other projects from Euroimmun Medizinische Labordiagnostika AG and Werfen GmbH/Inova Diagnostics Inc. The other authors have no conflict of interests.

Bastian Engel was supported by the PRACTIS – Clinician Scientist program of Hannover Medical School (DFG, ME 3696/3).

Data collection, analysis, and writing of the manuscript: Kilian Bock, Ingmar Mederacke, and Young-Seon Mederacke. Review of the conception, design, and manuscript preparation: Kilian Bock, Young-Seon Mederacke, Bastian Engel, Elmar Jäckel, Heiner Wedemeyer, and Ingmar Mederacke.

All data generated or analysed during this study are included in this article and its online supplementary material files. Further enquiries can be directed to the corresponding author.