Interferon-induced thyroid dysfunction in chronic hepatitis C

Interferon-induced thyroid dysfunction in chronic hepatitis C

Journal of Gastroenterology and Hepatology

Early View (Articles online in advance of print) May 2009

Khaleel M Jamil,* Peter J Leedman, †,‡,§ Nickolas Kontorinis,* Lorenzo Tarquinio,* Saroja Nazareth,* Marion McInerney,* Crystal Connelly,* James Flexman, ¶ Valerie Burke, ‡ Cecily Metcalf** and Wendy Cheng*

*Departments of Gastroenterology and Hepatology, ¶ Microbiology, † Endocrinology and Diabetes, **Pathology, § Royal Perth Hospital, Centre for Medical Research, Western Australian Institute for Medical Research, and ‡ School of Medicine and Pharmacology, University of Western Australia, Perth, Western Australia, Australia

Correspondence to Dr Khaleel Jamil, Department of Gastroenterology, University Hospital Lewisham, Lewisham High Street, London SE13 6LH.

Email: khaleel.jamil@nhs.net

ABSTRACT

Background: Treatment of chronic hepatitis C with interferon is known to be associated with thyroid dysfunction (TD) in 5–14% of patients. We studied the incidence, types, outcome and risk factors predictive of thyroid dysfunction.

Methods: A retrospective analysis was performed on all patients treated with interferon alpha (IFN) or pegylated interferon alpha (PEG-IFN) ± ribavirin (RBV), who developed abnormal thyroid function tests (TFTs). These cases were compared with treatment-matched controls to identify factors predictive of thyroid dysfunction. Statistical methods consisted of: χ2 test, Fischer's exact test, Welch's t-test, and multivariate analysis.

Results: From a total of 511 patients, 45 cases with TD were identified (8.8%). Pegylated interferon alpha was associated with higher rates of TD than IFN (14.1% vs 6.0%, P = 0.0029). Female sex (OR 5.6, 95% CI 1.1–7) and Asian ethnicity (OR 2.7, 95% CI 1.4–22) were independent predictors of developing TD. Cytology was obtained in 13 patients: benign follicular pattern (8); thyroiditis (3); and normal (2). Thyroid peroxidase (TPO) antibodies (P = 0.004) and earlier onset of dysfunction (P = 0.03) were associated with need for treatment. Sixteen patients had persistent TD by the end of the follow-up period, predicted by female sex, non-Asian ethnicity, prior history of TD and TPO antibodies.

Conclusions: Pegylated interferon alpha, female sex and Asian ethnicity are independent risk factors for TD. Thyroid peroxidase antibodies and earlier TD within the course of IFN are associated with the requirement for treatment. Thyroid function tests should be monitored during and after IFN-based therapy. The most common cytological finding is a benign follicular pattern.

Introduction

There are approximately 260 000 people in Australia living with Hepatitis C (HCV) infection, and it is predicted that by 2020 this may rise to 836 000.1 Standard treatment combines pegylated interferon-α (PEG-IFN) and ribavirin (RBV) with response rates of 50–80% depending on viral genotype.2 Side effects frequently occur during therapy, requiring dose reductions or treatment cessation which has been shown to reduce sustained virological response (SVR) rates.3 Thyroid dysfunction (TD), the commonest autoimmune disorder associated with IFN therapy, occurs in 4–18% of patients. Previous studies have shown IFN related TD is associated with female gender,4–6 the presence of thyroid autoantibodies,6–8 and an oriental origin.9,10 The spectrum of IFN-associated TD includes destructive thyroiditis, Graves' thyrotoxicosis and hypothyroidism.11 To date, there is little data on the long-term course and outcome of patients who develop TD during IFN-based therapy. In this study we investigated the incidence, risk factors, patterns of disease, and outcome of TD associated with IFN-based therapy for chronic HCV infection.
Discussion

In this study of a large Australian cohort of patients with hepatitis C, the prevalence of biochemical TD during IFN-based treatment therapy was 8.8%. The prevalence of significant TD requiring treatment was 3.9% of which 70% was due to hypothyroidism. Females were at highest risk of developing TD. These findings are consistent with previously published data from Australian and worldwide cohorts.6,12,13
Pegylated IFN was associated with significantly more TD than standard interferon (14% vs 6%). Interferon has a direct inhibitory effect on synthesis and secretion of thyroid hormones.14,15 Thus, it is plausible that the increased half-life resulting from pegylation enhances this effect. However, clinicians may have been more alert to IFN-induced TD in latter times, and therefore been more proactive in detecting TD in the era of RBV and PEG-IFN. Such a recall bias would suggest the associations were coincidental rather than causal. In a recent meta-analysis comparing PEG-IFN with standard IFN, no difference in TD was found.6 Further analysis of other cohorts will be required to validate these observations.

In this study the addition of RBV to IFN did not increase the rate of TD. However, previous studies have shown that RBV modifies T-helper cell responses, inducing Th-1 activity and activating CD8+ lymphocytes.16–18 This mechanism could lead to immune-mediated thyroid destruction. In addition, a previous study associating increased hypothyroidism with RBV postulates an alternative mechanism through apoptosis of thyroid follicular cells,19 as ribavirin has been shown to regulate cellular differentiation, proliferation, and apoptosis in in vitro models.20,21 Results are conflicting, and the role of RBV remains unclear.

There was a trend towards higher TD as treatment duration increased, supporting the idea of a dose-related effect of the drugs. This would explain why genotype 3 was associated with a lower rate of TD, as these patients were exposed to shorter treatment.

The key factors predictive of IFN-associated TD were female sex, Asian ethnicity and a prior history of thyroid disease. One previous study found an association between Asian ethnicity and TD,9 but the number of Asian patients was small, making definitive interpretation difficult. Consistent with our data, several studies have shown pre-existing thyroid autoimmunity predicted hypothyroidism.7,22 In the present study there was a trend towards TD in patients with a family history of autoimmune disease. Taken together, these results suggest there are predisposing genetic factors that alter patients' susceptibility to development of TD.23

We found that TPO antibodies (especially high titre) were strongly associated with TD requiring treatment as opposed to biochemical TD alone. Furthermore, TPO antibodies were found more frequently in patients who developed hypothyroidism, and in patients who had persistent TD at the end of follow-up. These findings support the strong foundation for an immunological basis for the development of the IFN-associated TD. As well as the direct effects on thyrocytes, IFN activates lymphocytes leading to increased cytokine production, and the induction of thyroid autoantibodies.11 Furthermore, the hepatitis C virus itself has been hypothesized to induce thyroid autoantibody production,24 consistent with the observation that background thyroid disease is common in hepatitis C.5 In the present study, other thyroid autoantibodies were not predictive of the need for treatment, or of persistence of TD.

This is the first study to report thyroid cytological findings in IFN-induced thyroid disease. A benign follicular pattern was seen in both hyper- and hypothyroidism, but was more commonly associated with the former. In contrast, as expected, a lymphocytic thyroiditis was only seen in hyperthyroidism. These data support the concept of an immune-mediated IFN-induced thyroiditis in these patients. However, the histopathology numbers are small and the findings cannot be generalised on the basis of this. Further study in this area is indicated in order to determine if thyroid histology may be a useful additional prognostic indicator in these patients.

Hypothyroidism is the commonest form of TD associated with anti-HCV therapy. In this study, 93% of treated hypothyroid patients remained hypothyroid by the end of follow-up. Conversely, two of the six treated hyperthyroid patients became hypothyroid after antithyroid medication and only one remained hyperthyroid by the end of the follow-up period (13 months). There was a biphasic pattern of biochemical hyper- then hypothyroidism in six patients, typical of an inflammatory thyroiditis induced by IFN. This pattern has been reported previously,22,25,26 associated with induction of a destructive thyroiditis resulting in a transient subclinical thyrotoxicosis. Indeed, it is possible that substantial numbers of patients with IFN/RBV induced hypothyroidism follow this pattern, with the window of transient hyperthyroidism being missed both clinically and biochemically.

Identification of markers of therapeutic outcome in patients receiving IFN treatment for hepatitis C would be useful clinically. It has been postulated that development of TD is indicative of a more aggressive immune response to IFN therapy and as a consequence these patients are more likely to clear the virus.9,13 Our findings do not support this theory as there was no association between SVR and development of TD.

Results

Study population

From a total of 511 patients, 45 cases with TD were identified (8.8%). The mean follow-up period was 40 months (range 4–113). Pegylated interferon alpha therapy was associated with significantly more TD than standard IFN (14% vs 7%, P = 0.038). RBV was not associated with TD as illustrated in Table 1.

Predictors of thyroid dysfunction

The baseline characteristics and treatment data for the cases and treatment-matched controls are illustrated in Table 2. The mean treatment duration was approximately 36 weeks for cases and approximately 31 weeks for controls. The mean age for men and women was similar (∼43 years) with an approximately equal sex ratio in the cases. The vast majority of cases had Metavir fibrosis scores of 1–2 (∼65%). In the univariate analysis female sex, Asian ethnicity and previous history of thyroid abnormality were significantly associated with development of TD. There were trends towards higher TD in patients with non-genotype 3, with a family history of auto-immune disease, and with longer treatment duration but these did not reach statistical significance. In the multivariate analysis, female sex (OR 2.75, 95% CI 1.07–7.06) and Asian ethnicity (OR 5.66, 95% CI 1.43–22.4) were independent predictors of TD. However, TD was not associated with age, weight, fibrosis score, ANA, ASMA, AMA, or SVR.

Type of thyroid dysfunction

Twenty-three patients (51%) developed biochemical hyperthyroidism, and 22 (49%) hypothyroidism (Table 3). The mean time to development of abnormal TFT was 21.4 weeks (SD 13). Co-existing diabetes was predictive of hypothyroidism (P = 0.049).

The presence of TPO antibodies was associated with hypothyroidism, although this did not reach statistical significance (P = 0.11). Cytology was obtained in 13 patients: a benign follicular pattern was more common in hyperthyroid patients compared with hypothyroid (100% vs 44%); lymphocytic thyroiditis was found in 3 hypothyroid and no hyperthyroid patients; cytology was normal in 2 cases.

Severity and treatment of thyroid disease

After assessment by an endocrinologist, treatment was given in 20 cases (Table 4). The other cases did not require treatment as these patients had mild sub-clinical TD. Six patients (30%) were treated for hyperthyroidism with carbimazole or radio-iodine, which rendered two hypothyroid by the end of the follow-up period. The remaining 14 patients (70%) were treated for hypothyroidism with thyroxine replacement; five of whom had treatment-induced thyroiditis characterised by transient symptomatic and biochemical hyperthyroidism followed by hypothyroidism. Female sex, TPO antibodies and an earlier onset of TD within the IFN therapy were significantly associated with need for treatment (Table 5). Sixteen patients had persistent TD at the end of follow-up (mean follow-up period in this group was 42 months, range 7–113 months): 13 of the 14 hypothyroid patents (93%) had persistent TD, compared with 3 of the 6 (50%) hyperthyroid patients (P = 0.06). Particularly high titre TPO antibodies (>725) were invariably associated with hypothyroidism and the need for long term thyroxine (100%).

Patterns and outcome of TD, and association with SVR

In the univariate analysis persistence of TD was predicted by female sex, non-Asian ethnicity, prior history of TD, TPO antibodies, and a shorter duration of antiviral therapy. All but female sex reached statistical significance in the multivariate analysis (Table 6). Persistence of TD was not associated with age, family history of autoimmune disease or timing of onset of TD.

Methods

Patients

A case control study was performed between the period 1994 and 2006 at a tertiary hospital liver clinic. The patient cohort was acquired from a database of all patients with HCV infection treated with IFN-based therapy (including standard IFN monotherapy, IFN+RBV combination therapy, and PEG-IFN+RBV combination therapy). Patients had thyroid function tests (TFTs) measured at baseline and at 3-monthly intervals during treatment. Cases of TD were identified by development of abnormal serum thyroid stimulating hormone (TSH: TSH-3 ADVIA Centaur System; two-site sandwich immunoassay using direct chemiluminometric technology, normal: 0.40–4.00 mU/L) ± abnormal free thyroxine (fT4: ADVIA Centaur System; competitive immunoassay using direct chemiluminometric technology, normal: 10–23 pmol/L) levels on treatment. These cases were compared with randomly selected controls to identify factors predictive of thyroid dysfunction. The control group was selected from the cohort of HCV treated patients who did not develop TD, using computer generated randomly assigned numbers. As it was found that PEG IFN treated patients developed more TD than standard IFN, controls were treatment matched before randomisation.

Data collection

Demographic, epidemiological and clinical data including gender, ethnicity, weight, age at treatment, co-morbidity (including autoimmune disease), family history of autoimmune disease (defined as having a first or second degree relative diagnosed with autoimmune disease), HCV genotype, liver histology, type, duration and outcome of anti-HCV treatment were analysed. A previous history of thyroid disease was defined as having been diagnosed and treated for TD prior to antiviral therapy.

Laboratory data collected included presence of auto-antibodies (anti-nuclear antibody (ANA), anti-smooth muscle antibody (ASMA), anti-mitochondrial antibody (AMA)). Specialised TFTs (including free tri-iodothyronine (fT3: ADVIA Centaur System; competitive immunoassay using direct chemiluminometric technology, normal: 3.5–6.5 pmol/L), thyroid peroxidase antibody (TPOAb: Immulite 2000; solid phase 2 site sequential chemiluminescent immunometric assay, normal: <35 kU/L), anti-TSH receptor antibody (anti-TSH Ab: The Brahms human TRAB test; Manual Radioimmunoassay using I125 normal: <1 U/L), and anti-thyroglobulin antibody (TGAb: Immulite 1000; solid phase 2 step chemiluminescent enzyme immunoassay, normal: <40 kU/L) were measured only in patients who developed TD. Sustained virological response was defined as the absence of detectable serum HCV ribonucleic acid 24 weeks after the completion of antiviral therapy. The follow-up period was measured from the end of antiviral therapy.

Types, patterns and severity of thyroid dysfunction

Biochemical hypothyroidism was defined by a TSH above the upper limit (4 mU/L) or a fT4 below the lower limit (10 pmol/L) of the laboratory normal ranges. Biochemical hyperthyroidism was defined by a TSH below the lower limit (<0.40 mU/L) or a fT4 above the upper limit of the laboratory normal range (>23 pmol/L). Patients with symptomatic TD were referred on to an endocrinologist for further evaluation and treatment, which for some included fine needle aspiration of the thyroid and a thyroid 99Tc Technecium uptake scan to differentiate thyroiditis from Graves' disease, as well as specific therapy for hypothyroidism or hyperthyroidism. Transient dysfunction was defined by normalisation of TFTs without the need for ongoing treatment by the end of follow-up. Conversely, persistent TD was characterised when ongoing treatment was required, or TFTs remained abnormal by the end of the follow-up period.

Statistical analysis

Descriptive statistics are shown as mean ± SD as appropriate. Comparisons between groups were made using the χ2 test or Fischer exact test for qualitative data. Comparisons of means were calculated using the unpaired t-test. All tests were two tailed. Logistic regression models were used to examine predictors of TD. A two sided P-value < 0.05 was considered to be statistically significant.