Age and Dose Are Major Risk Factors for Liver Damage Associated with Intravenous Glucocorticoid Pulse Therapy for Graves' Orbitopathy

Abstract

Background:

High-dose intravenous glucocorticoid (ivGC) pulse therapy, which is commonly used for Graves' orbitopathy (GO), has been associated with acute liver damage (ALD), resulting in a fatal outcome in a few cases. No certain risk factors for ALD have been established. Consequently, a large retrospective cohort study was performed.

Methods:

The relationship between ALD and several potential risk factors was assessed in 1076 consecutive patients with GO given ivGC. ALD was defined as an increase of alanine aminotransferase ≥300 IU/L.

Results:

Fourteen cases of ALD were recorded, resulting in a morbidity of 1.3%. Thirteen patients recovered and one died, resulting in a mortality of 0.09%. There was a significant, positive correlation of ALD with age and methylprednisolone acetate (MPA) cumulative dose, and ALD was more common (relative risk [RR]=2.8; p=0.05) in patients aged ≥53 years (9/420; 2.14%) than in those aged <53 years (5/656; 0.76%). In patients aged ≥53 years, there was a significant positive correlation of ALD with MPA cumulative dose, and with MPA dose per infusion. Thus, the frequency of ALD in this age group was greater (RR=3.48; p=0.04) in patients with a MPA dose per infusion ≥0.7 g (5/111, 4.5% vs. 4/308, 1.29%). Regardless of age, no cases of ALD were observed for MPA doses per infusion <0.57 g.

Conclusions:

Age and MPA dose are significant risk factors for ALD, with the following practical implications. First, the total MPA cumulative dose should not exceed 8.5 g (the average dose in patients without ALD). Second, in patients aged ≥53 years, selection and observation should be quite strict. However, being aged ≥53 years should not be seen as an absolute contraindication to ivGC, especially in patients with severe GO, considering that the risk of ALD, although statistically significant, was relatively low. Third, the MPA dose should not exceed 0.57 g per infusion, a measure to be applied regardless of age.

Introduction

Graves' orbitopathy (GO) is a unique eye syndrome that occurs in ∼25% of patients with Graves' disease, and is severe enough to require major treatment in 2–3% of them (1 –6). High-dose intravenous glucocorticoids (ivGC) pulse therapy is a well-established treatment modality for active GO, taking advantage of the immune suppressive and anti-inflammatory actions of glucocorticoids (GC) (1 –7).

Having been originally used and still being commonly used in several autoimmune diseases, especially of the nonorgan-specific type, ivGC has been widely employed for GO over the last 30 years (1 –7). Several studies and meta-analyses have shown a greater effectiveness and a lower prevalence of common side effects of GC if administered intravenously, compared with the oral route of administration (7,8). ivGC has been associated with acute liver damage (ALD) (9 –17), which has not been observed with oral GC. However, the lack of association of ALD with oral GC may be due to likelihood of under-reporting. In a few cases, ALD has resulted in a fatal outcome (9 –17). The first two cases of lethal ALD were reported in one patient with systemic lupus erythematosus and in one with dermatomyositis (9), followed by four additional cases in patients with GO (10,11). In a recent, large epidemiological study, ALD morbidity in patients treated with ivGC was estimated to be ∼1% (16).

Besides its frequency, an important issue concerning ivGC-associated ALD is that no certain risk factors have been identified, and because of this, it is quite difficult to make an evidence-based selection of patients to be treated. This is quite relevant, especially regarding GO, considering that there are no alternative medical treatments as effective as ivGC (1 –7). Based on three possible pathogenetic mechanisms (toxic, autoimmune, viral), the authors and others have introduced empirically a series of preventive measures over the years, including a reduction of the GC dose, frequency and number of pulses, and the administration of oral GC after and/or during ivGC to prevent immune rebound, and no cases of lethal ALD were reported since the last observation in 2003 (7,14), although ALD morbidity was apparently not reduced (16).

In order to identify ALD risk factors, a large retrospective cohort study was performed in patients with GO given ivGC and it was determined that age and GC dose play a major role as risk factors for ALD, thereby giving us some clues on how to manage ivGC treatment in candidate patients.

Patients and Methods

Study design and patients

The study design was to assess retrospectively the relation between ALD and several potential risk factors in a cohort comprising all consecutive GO patients given ivGC for a period of 15 consecutive years in the authors' GO clinic (a tertiary referral center). From January 1, 1998, to December 31, 2012, 1076 patients (287 males, 789 females; age 48.8±11.9 years, range 17–79 years) with moderate to severe, active GO were treated with high-dose ivGC as detailed below. Signed informed consent was obtained from all patients.

Methods

Approximately two weeks before treatment, all patients underwent the following evaluation: (a) routine blood tests, including, among others, total cholesterol, triglycerides, aspartate (AST) and alanine (ALT) aminotransferase, alkaline phosphatase (ALP), gamma-glutamyl transferase (γGT), and total and direct bilirubin; and (b) serum markers of exposure to hepatitis B virus (HBV) and hepatitis C virus (HCV). Liver ultrasound was performed routinely since 2004, whereas it was performed only in some patients up to 2003. Overall, it was available in 731 patients. Likewise, serum autoantibodies known to be associated with autoimmune hepatitis (18 –20), namely antinuclear (ANA), antimitochondrial (AMA), anticentromeric (ACA), antiliver-kidney-microsomal (LKM), and antismooth muscle (ASMA) antibodies, were measured routinely from 2004, whereas they were measured only in a few patients before then, and were therefore available in 635 patients.

The ivGC therapy protocols were: (a) 1998 (first semester): two methylprednisolone acetate (MPA) infusions on alternate days every other week, for a total of 16 infusions; MPA dose: 15 mg/kg of body weight for the first eight infusions (maximum 1 g) and 7.5 mg/kg of body weight for the last eight infusions; (b) 1998 (second semester)–2003: two MPA infusions on alternate days every other week, for a total of 12 infusions; MPA dose: 15 mg/kg for the first four infusions (maximum 1 g) and 7.5 mg/kg for the last eight infusions; (c) 2004–2012 (protocol a): one MPA infusion every week for a total of 12 infusions; MPA dose: 15 mg/kg of body weight for the first four infusions (maximum 1 g) and 7.5 mg/kg of body weight for the last eight infusions; (d) 2004–2012 (protocol b): one MPA infusion every week for a total of 12 infusions; MPA dose: 500 mg for the first four to six infusions and 250 mg for the last eight to six infusions. Sixty-nine patients were given the treatment twice for persistence or relapse of moderate to severe, active GO, not before six months from the previous course.

Beginning in 2004, after ivGC treatment with MPA, to avoid immune rebound, patients were given oral prednisone, starting at a dose of 40 mg every other day, which was tapered every 10 days to be withdrawn after 50 days. This type of treatment was given to 593 patients. Still, from 2004, patients who had positive tests for nonorgan-specific autoantibodies were also given oral prednisone 40 mg every other day starting on the day after the third infusion of MPA. This treatment was continued until the end of ivGC, after which the same protocol of oral prednisone used for the remaining patients was started.

AST, ALT, ALP, γGT, and total and direct bilirubin were measured every two weeks during ivGC in all patients, as well as every two weeks after ivGC, up to three months after the end of treatment. Then, they were measured monthly up to six months after the end of treatment.

ALD was defined as an increase of ALT ≥300 IU/L (21 –23). ALT, rather than AST, was chosen because it is more specific to the liver (22,23).

The outcome of the study was the relation between ALD and the following potential risk factors: sex, age, cumulative MPA dose, MPA dose per infusion (the ratio between the MPA dose and the number of infusions), obesity (a body mass index [BMI] ≥30 kg/m²), BMI, clinically overt previous liver diseases of any type (comprising viral hepatitis of any type, toxic hepatitis, alcohol-related acute or chronic liver disorders), previous exposure to HBV or HCV, positive tests for nonorgan-specific autoantibodies, administration of oral prednisone after ivGC, administration of oral prednisone during ivGC, liver steatosis, diabetes, hypertension, hyperlipemia, total cholesterol values, and triglycerides values.

Statistical analyses

Descriptive data are presented as mean±standard deviation (SD). When appropriate, the following tests were performed: (a) simple regression analysis, (b) multiple regression analysis (only within parameters that were significant by simple regression analysis and that were independent from one another), and (c) chi-square test.

Cut-off values were established based on the following formula: [(mean values in the unaffected population×SD in the affected population)+(mean values in the affected population×SD in the unaffected population)]/(SD in the affected population+SD in the unaffected population).

The relative risk (RR) was calculated based on the following formula: incidence in the exposed population for a given parameter/incidence in the nonexposed population for the same parameter.

Results

A total of 14 cases of ALD were recorded, resulting in a morbidity of 1.3%. Demographical features of patients with ALD are shown in Table 1. As reported above, ALD was defined as an increase of ALT ≥300 IU/L, but also AST, which is less specific to the liver (22,23), was increased in all 14 patients with ALD, and no cases of isolated AST increase were recorded. Thirteen of the 14 patients with ALD recovered either spontaneously (eight patients) or following intramuscular or oral treatment with GC for autoimmune hepatitis (five patients). One patient died because of liver failure, resulting in a mortality of 0.09% and, limited to patients with ALD, in a mortality of 7.1%. Some of these patients, including the one who died, were reported previously (11,13,16).

Table 1.

Demographical Features of 14 Patients with Graves' Orbitopathy who Developed ALD During or Following Intravenous Glucocorticoid Pulse Therapy

ID	Sex	Age (years)	Year	ALD outcome
PMR	F	45	1998	Recovery
SE	F	30	2002	Recovery
LI	F	47	2002	Death
GR	F	55	2002	Recovery
LF	F	43	2003	Recovery
PA	F	54	2003	Recovery
BAM	F	77	2005	Recovery
ZP	F	76	2006	Recovery
FL	M	55	2006	Recovery
PL	F	75	2007	Recovery
MR	F	53	2008	Recovery
AR	F	78	2009	Recovery
LP	M	61	2010	Recovery
GM	M	46	2012	Recovery

ALD, acute liver damage.

No significant correlation was found between ALD and sex, obesity, BMI, previous liver diseases of any type, previous exposure to HBV or HCV, positive tests for nonorgan-specific autoantibodies, administration of oral prednisone after ivGC, administration of oral prednisone during ivGC, liver steatosis on ultrasound, diabetes, hypertension, hyperlipemia, total cholesterol values, and triglycerides values.

As shown in Table 2, by simple regression analysis, a significant correlation of ALD with age, MPA cumulative dose, and MPA dose per infusion was observed. Because the MPA cumulative dose and MPA dose per infusion were not independent variables, separate multiple regression analyses were performed for each of these two variables with age. A significant correlation with ALD was confirmed only for age and MPA cumulative dose, which were both greater in patients with ALD (Table 2).

Table 2.

Correlation Between ALD and Age, MPA Dose, and MPA Dose per Infusion, by Simple and Multiple Regression Analyses

	ALD	No ALD	Simple regression	Multiple regression
Age (years)	56.8±14.8; range 30–78	48.6±11.9; range 17–79	p=0.01	p=0.01 with each of the other parameters^a
MPA cumulative dose (g)	10.2±6.2; range 3.8–30	8.5±2.6; range 1.2–30	p=0.01	p=0.02^a
MPA dose per infusion (g)	0.73±0.13; range 0.57–1	0.67±0.11; range 0.41–1	p=0.03	p=NS^a

Multiple regression analyses were performed separately for age and MPA cumulative dose and for age and MPA dose per infusion.

MPA, methylprednisolone.

Based on these findings, a cut-off value at 53 years of age was established, which turned out to be significant. Thus, the frequency of ALD was significantly greater (RR=2.8; p=0.05) in patients aged ≥53 years (9/420; 2.14%) than in those aged <53 years (5/656; 0.76%), with a 64.2% sensitivity and a 61.2% specificity. In contrast, it was not possible to establish a significant cut-off value for MPA cumulative dose.

Analyses of risk factors were restricted to patients aged ≥53 years, within whom there was a significant correlation of ALD with MPA cumulative dose, MPA dose per infusion, and BMI by simple and multiple regression (Table 3). Because MPA cumulative dose and MPA dose per infusion were not independent variables, multiple regression analyses were performed separately for each of these two variables with BMI.

Table 3.

Correlation Between ALD and MPA Dose, BMI, and MPA Dose per Infusion, by Simple and Multiple Regression Analyses, in Patients Aged ≥53 Years

	ALD	No ALD	Simple regression	Multiple regression
MPA cumulative dose (g)	11.2±7.4; range 3.8–30	8.5±2.7; range 2–30	p=0.05	p=0.003^a
BMI (kg/m²)	24.0±3.1; range 19–29	26.9±4.5; range 17–54	p=0.05	p=0.03 (with MPA cumulative dose); p=0.02 (with MPA dose per infusion)^a
MPA cumulative dose per infusion (g)	0.75±0.14; range 0.57–1	0.67±0.11; range 0.43–1	p=0.03	p=0.01^a

Multiple regression analyses were performed separately for MPA cumulative dose and BMI and for BMI and ratio between MPA dose and number of MPA infusions.

As expected, MPA cumulative dose and MPA dose per infusion were greater in patients with ALD aged ≥53 years (Table 3), whereas BMI was lower (Table 3). Because many patients were given a fixed MPA dose not tailored to body weight or BMI, the possibility was considered that the lower BMI in patients with ALD aged ≥53 years reflected a greater ratio between the MPA cumulative dose and BMI, which was in fact the case. Thus, the ratio between the MPA cumulative dose and BMI was significantly greater (p=0.002) in patients with ALD aged ≥53 years (0.47±0.32 g/[kg/m²], range 0.20–1.3] than in those without ALD (0.32±0.11 g/[kg/m²], range 0.07–1.3).

Among patients aged ≥53 years, significant cut-off values could not be established for the MPA cumulative dose, BMI, and ratio between MPA cumulative dose and BMI. In contrast, a cut-off value for MPA dose per infusion was established at 0.7 g, which turned out to be significant. Thus, the frequency of ALD in patients aged ≥53 years was significantly greater (RR=3.48; p=0.04) in those with a MPA dose per infusion ≥0.7 g (5/111; 4.5%) than in those with a MPA dose per infusion <0.70 g (4/308; 1.29%), resulting in a 55.5% sensitivity and in a 74.1% specificity. Regardless of age, no cases of ALD were observed for MPA doses per infusion <0.57 g.

Discussion

The use of ivGC for GO has been significantly limited by the fear for ALD, of which several cases have been reported, six with a fatal outcome (9 –17). The lack of known risk factors has clearly affected the possibility of selecting patients and/or tailoring ivGC treatment when necessary. To overcome this limitation, a large retrospective cohort analysis was performed in consecutive GO patients treated with ivGC over a period of 15 years. It was determined that age and MPA dose are significant risk factors for the development of ALD, which, as discussed below, has important implications in clinical practice.

ALD morbidity in the present cohort was 1.3%, similar to that reported recently in a smaller series of consecutive patients treated with ivGC (∼1%) (16), and previously in a series of ∼800 patients (∼0.9%) (11). ALD related mortality was 0.09%, greater than that reported in a recent series in which, however, no lethal cases were detected, presumably because of an insufficient number of subjects (n=376) (16). Mortality was lower than that reported in the previous series of ∼800 patients (0.3%) (11), in which, however, morbidity was likely underestimated, because liver enzymes were not measured systematically in all patients treated with ivGC, thereby explaining why, also limited to patients with ALD, the mortality we observed here (7.1%) was markedly lower than that observed in the previous series (42.7%).

Among the various parameters that were examined as potential risk factors, older age and a higher MPA cumulative dose were significantly and independently associated with ALD. Thus, in patients with ALD, age was significantly greater than in those without ALD, allowing us to establish a cut-off value at 53 years, above which RR=2.8 for ALD. Likewise, the MPA cumulative dose was significantly greater in patients with ALD, and although a cut-off value over which ALD was more significantly observed could not be established, the average dose in patients without ALD was 8.5 g, in line with previous observations suggesting no adverse, fatal events with MPA doses ≤8 g (7,14).

In patients aged ≥53 years, the MPA cumulative dose, MPA dose per infusion, and BMI appeared to be independent risk factors for ALD. Again, cut-off values in this subgroup of patients could not be established for MPA cumulative dose above which ALD was more frequent, but, as in the total cohort, the average cumulative dose in patients without ALD was 8.5 g. The same applies to the MPA dose per infusion, although no cases of ALD were detected for doses <0.57 g per infusion, which was also the case when this analysis was extended to the entire cohort. Considering that the MPA dose per infusion conferred a 3.48 RR for ALD in patients aged ≥53 years, this parameter seems to be a rather important risk factor, even though its sensitivity appears to be quite low (55.5%).

The relation between ALD and a low BMI reflected the fact that many patients were given a fixed MPA dose, not tailored based on body weight or BMI, as shown by the significantly greater ratio between MPA cumulative dose and BMI in patients with ALD.

These observations have important clinical implications that can be summarized as follows. First, the total MPA cumulative dose should not exceed 8.5 g, the average dose in patients without ALD. This is in line with a recent survey among members of the European Thyroid Association (14), as well as with a recent meta-analysis (7), where it was shown that fatal adverse events, including ALD, are not observed when cumulative MPA doses of ≤8 g are administered. Here, as mentioned above, although cases of ALD were observed also with doses <8.5 g, no lethal cases were observed below this dose. Second, particular attention should be paid to patients aged ≥53 years, in whom the risk of ALD is significantly increased. Thus, in this age group, selection of patients should be quite strict, limiting the therapy to those with more severe forms of the disease, and using MPA doses ≤0.57 g per infusion. On the other hand, being aged ≥53 years should not be seen as an absolute limitation to ivGC, especially in patients with severe GO, considering that the risk of ALD, although statistically significant, was relatively low (RR=2.8). The use of MPA doses ≤0.57 g per infusion should also be applied to patients aged <53 years, in whom, as reported above, no cases of ALD were reported for MPA doses <0.57 g per infusion. In this regard, the protocol proposed by Kahaly et al. (24), and then adopted by the European Group On Graves' Orbitopathy (EUGOGO) (25), which recommends the use of MPA starting doses of 0.5 g per infusion (six infusions of 0.5 g followed by six infusions of 0.25 g) should be adopted. Provided these recommendations are applied, the present findings indicate that, concerning the liver, ivGC is a relatively safe treatment that can be used for GO in clinical practice.

As discussed above, a relationship between GC dose and fatal ALD is already generally accepted (7,14). However, this conclusion was based only on case reports, as no cases of fatal ALD were recorded for MPA doses ≤8 g, as shown both in the above-mentioned meta-analysis (7) and survey (14). However, to the authors' knowledge, no cohort investigations were conducted before the present study, with the exception of another study from the authors' group (16), in which, however, risk factors could not be identified because of the smaller number and of the greater homogeneity of patients. In fact, to the authors' knowledge, a statistically significant relationship between GC dose and ALD has not been shown previously. In addition, even the above-mentioned meta-analysis (7) and survey (14) do not offer information on the role of the GC dose on mild, nonfatal liver damage. Thus, the present study provides, to the authors' knowledge, the first clear-cut evidence of a direct relationship between GC dose and ALD, the latter both severe and mild. Furthermore, to the authors' knowledge, no information was available before the present study on the role of age as a risk factor for ALD.

Obviously, this study carries the limitations of a retrospective investigation, which may affect its epidemiological significance. Thus, ivGC was given to patients using protocols that changed over the years and that differed concerning both the cumulative and the single MPA dose, which, also in view of the present findings, might have had an influence on the overall prevalence of ALD. Another possible limitation is the fact the evaluation of patients also changed over the years, and, as a consequence, some tests were not performed in all patients. For example, liver ultrasound was not performed in ∼30% of patients and, similarly, autoantibodies known to be associated with autoimmune hepatitis (18 –20) were not measured in ∼40% of patients. However, the number of patients in whom these tests were available was likely sufficient to exclude a major role of these parameters as ALD risk factors.

Footnotes

Acknowledgment

Supported by a Grant from MIUR (Ministero dell'Istruzione, dell'Università e della Ricerca Scientifica) (2004068078 to M.M.).

Author Disclosure Statement

Eleonora Sisti, Barbara Coco, Francesca Menconi, Marenza Leo, Roberto Rocchi, Francesco Latrofa, Maria Antonietta Profilo, Barbara Mazzi, Paolo Vitti, Claudio Marcocci, Maurizia Brunetto, and Michele Marinò declare that they do not have any commercial association that might create a conflict of interest in connection with this manuscript.

References

Bartalena

, Fatourechi

. 2014. Extrathyroidal manifestations of Graves' disease: a 2014 update. J Endocrinol Invest, 37:691–700.

Wiersinga

. 2007. Management of Graves' ophthalmopathy. Nat Clin Pract Endocrinol Meta, 3:396–404.

Perros

, Krassas

. 2009. Graves' orbitopathy: a perspective. Nat Rev Endocrinol, 5:312–318.

Stan

, Garrity

, Bahn

. 2012. The evaluation and treatment of Graves' ophthalmopathy. Med Clin North Am, 96:311–328.

Leo

, Menconi

, Rocchi

, Latrofa

, Sisti

, Profilo

, Mazzi

, Albano

, Nardi

, Vitti

, Marcocci

, Marinò

. 2015. Role of the underlying thyroid disease on the phenotype of graves' orbitopathy in a tertiary referral center. Thyroid, 25:347–351.

Bartalena

. 2013. Diagnosis and management of Graves' disease: a global overview. Nat Rev Endocrinol, 9:724–734.

Zang

, Ponto

, Kahaly

. 2011. Clinical review: intravenous glucocorticoids for Graves' orbitopathy: efficacy and morbidity. J Clin Endocrinol Metab, 96:320–332.

Riedl

, Kolbe

, Kampmann

, Krämer

, Kahaly

. 2015. Prospectively recorded and MedDRA-coded safety data of intravenous methylprednisolone therapy in Graves' orbitopathy. J Endocrinol Invest 2015 Jan 10 [Epub ahead of print].

Dourakis

, Sevastianos

, Kalipi

. 2002. Acute severe steathoepatitis related to prednisolone therapy. Am J Gastroenterol, 97:1074–1075.

10.

Weissel

, Hauff

. 2000. Fatal liver failure after high-dose glucocorticoid pulse therapy in a patient with severe thyroid eye disease. Thyroid, 10:521.

11.

Marinò

, Morabito

, Brunetto

, Bartalena

, Pinchera

, Marcocci

. 2004. Acute and severe liver damage associated with intravenous glucocorticoid pulse therapy in patients with Graves' ophthalmopathy. Thyroid, 14:403–406.

12.

Salvi

, Vannucchi

, Sbrozzi

, Del Castello

, Carnevali

, Fargion

, Beck-Peccoz

. 2004. Onset of autoimmune hepatitis during intravenous steroid therapy for thyroid-associated ophthalmopathy in a patient with Hashimoto's thyroiditis: case report. Thyroid, 14:631–634.

13.

Marinò

, Morabito

, Altea

, Ambrogini

, Oliveri

, Brunetto

, Pollina

, Campani

, Vitti

, Bartalena

, Pinchera

, Marcocci

. 2005. Autoimmune hepatitis during intravenous glucocorticoid pulse therapy for Graves' ophthalmopathy treated successfully with glucocorticoids themselves. J Endocrinol Invest, 28:280–284.

14.

Marcocci

, Watt

, Altea

, Rasmussen

, Feldt-Rasmussen

, Orgiazzi

, Bartalena

. 2012. Fatal and non-fatal adverse events of glucocorticoid therapy for Graves' orbitopathy: a questionnaire survey among members of the European Thyroid Association. Eur J Endocrinol, 166:247–253.

15.

Melamud

, Lurie

, Goldin

, Levi

, Esayag

. 2014. Methylprednisolone-induced liver injury: a diagnostic challenge. Isr Med Assoc J, 16:180–181.

16.

Sisti

, Coco

, Menconi

, Leo

, Rocchi

, Latrofa

, Mazzi

, Albano

, Vitti

, Marcocci

, Brunetto

, Marinò

. 2015. Intravenous glucocorticoid therapy for Graves' ophthalmopathy and acute liver damage: an epidemiological study. Eur J Endocrinol, 172:269–276.

17.

Le Moli

, Baldeschi

, Saeed

, Regensburg

, Mourits

, Wiersinga

. 2007. Determinants of liver damage associated with intravenous methylprednisolone pulse therapy in Graves' ophthalmopathy. Thyroid, 17:357–362.

18.

Dalekos

, Zachou

, Liaskos

, Gatselis

. 2002. Autoantibodies and defined target autoantigens in autoimmune hepatitis: an overview. Eur J Intern Med, 13:293–303.

19.

Strassburg

, Manns

. 2011. Therapy of autoimmune hepatitis. Best Pract Res Clin Gastroenterol, 6:673–687.

20.

Manns

, Vogel

. 2006. Autoimmune hepatitis, from mechanisms to therapy. Hepatology, 2:S132–144.

21.

Sheth

, Gordon

, Chopra

. 1997. Nonalcoholic steatohepatitis. Ann Intern Med, 126:137–145.

22.

Pratt

; Kaplan

. 2000. Evaluation of abnormal liver-enzyme results in asymptomatic patients. N Engl J Med, 342:1266–1271.

23.

Torezan-Filho

, Alves

, Neto

, Fernandes

, Strauss

. 2004. Clinical significance of elevated alanine aminotransferase in blood donors: a follow-up study. Liver Int, 6:575–581.

24.

Kahaly

, Pitz

, Hommel

, Dittmar

. 2005. Randomized, single blind trial of intravenous versus oral steroid monotherapy in Graves' orbitopathy. J Clin Endocrinol Metab, 90:5234–5240.

25.

Bartalena

, Krassas

, Wiersinga

, Marcocci

, Salvi

, Daumerie

, Bournaud

, Stahl

, Sassi

, Veronesi

, Azzolini

, Boboridis

, Mourits

, Soeters

, Baldeschi

, Nardi

, Currò

, Boschi

, Bernard

, von Arx

. 2012. Efficacy and safety of three different cumulative doses of intravenous methylprednisolone for moderate to severe and active Graves' orbitopathy. J Clin Endocrinol Metab, 97:4454–4463.