Inflammatory eye reactions with bisphosphonates and other osteoporosis medications: what are the risks?

Abstract

Inflammatory eye reactions (IERs) are rare but have been associated with medications to treat osteoporosis. The aim of this review is to summarize the current literature on the association between IERs and specific medications to treat osteoporosis (bisphosphonates, selective estrogen receptor modulators, strontium, denosumab and teriparatide). We cover the known epidemiology, potential pathogenic mechanisms and a resume of unanswered questions. Briefly, this review highlights that none of the existing randomized clinical trials were powered to identify these rare adverse events, and the majority of the information available is from spontaneous case reports and case series reporting associations between bisphosphonates and IERs. No case reports describe IERs after other anti-osteoporosis medications. Importantly, some case reports describe recurrence of the IER after affected patients were rechallenged with the same or another bisphosphonate, and that no reported cases resolved without discontinuation of the bisphosphonate. However, three large population-based cohort studies have shown conflicting results between osteoporosis treatments and IERs, but overall these studies suggest that IERs may actually be part of underlying inflammatory disease processes that also cause osteoporosis, rather than due to the medications used to treat osteoporosis themselves. There are no clear pathogenic mechanisms for how bisphosphonates could potentially cause IERs. However, the drug is secreted into the tears by the lacrimal gland and could cause irritation to the mucous membranes with subsequent release of inflammatory mediators, similar to the systemic response typically seen after infusion of bisphosphonates. However, in summary it is still not known whether there is a true causal association between bisphosphonates or other anti-osteoporosis medications and IERs, or whether it is confounding by indication and is actually due to underlying inflammatory diseases that cause both osteoporosis and IERs.

Introduction

This review aims to summarize the literature up to mid-2014 on the association between inflammatory eye reactions and medications to treat osteoporosis. Inflammatory eye reaction (IER) is a blanket term used to mean any of the following eye conditions: conjunctivitis, uveitis, scleritis, episcleritis and keratitis. The medications to treat osteoporosis discussed in this review are bisphosphonates (both nitrogen-containing such as alendronate, risedronate and ibandronate and non nitrogen-containing bisphosphonates such as etidronate), selective estrogen receptor modulators (raloxifene), strontium salt of ranelic acid (strontium ranelate), monoclonal receptor activator of nuclear factor κ-B ligand (RANKL) inhibitor (denosumab) and a recombinant form of parathyroid hormone (teriparatide). We cover what is known on the epidemiology and potential pathogenic mechanisms, and highlight the unanswered questions on the association between IERs and osteoporosis medications. The majority of the literature assesses the association between IERs and bisphosphonates, although other osteoporosis medications have also been studied somewhat. However, the epidemiology presented should be seen in the context of length of time the drug has been on the market and how many people have used it. To find the studies for this review we carried out a thorough literature search and are confident that we have fairly and comprehensively summarized the literature, although we did not carry out a formal systematic review and evidence synthesis.

Epidemiology

Ocular side effects of bisphosphonates and other treatments for osteoporosis have been reported for two decades, but are a rare adverse event. The randomized controlled trials of alendronate [Serrano et al. 2013], risedronate [Wells et al. 2008], zoledronate [Serrano et al. 2013], raloxifene [Olevsky and Martino, 2008], teriparatide [Brixen et al. 2004], strontium [Neuprez et al. 2008] and denosumab [Silva-Fernandez et al. 2013] have not reported IERs as an adverse reaction occurring in more than 1% of patients treated with either the active compound or placebo. However, none of these trials were powered to identify rare adverse events such as IERs.

Case reports and case series

Much of the information available is from spontaneous case reports and case series on associations between bisphosphonates and IERs; a comprehensive summary of these is given in a review by Pazianas and colleagues [Pazianas et al. 2013]. There have been approximately 30 case reports and case series describing 101 patients with IERs, and overall it seems the onset of IERs after bisphosphonate exposure is within a few hours of exposure up to more than 3 years, with an estimated median of 3 weeks [ADRAC, 2004]. The IER may be unilateral or bilateral. Clinical features from these case reports and case series which support a causative link between bisphosphonates and IERs include recurrence of the IER after affected patients were rechallenged with the same or another bisphosphonate, and that no reported cases resolved without discontinuation of the bisphosphonate [Pazianas et al. 2013]. Originally it was thought that only nitrogen-containing bisphosphonates were associated with IERs; however, it is now believed that IERs can be associated with both classes of bisphosphonates. No case reports or case series on associations between raloxifene, strontium, denosumab or teriparatide and IERs have been identified.

Population-based cohort studies

Over the past few years, three large cohort studies have published information on the association between osteoporosis treatments and IERs [French and Margo, 2008; Etminan et al. 2012; Pazianas et al. 2013]. Population-based studies are vital as there are many diseases that have IERs and osteoporosis as part of their clinical phenotype such as rheumatoid arthritis, inflammatory bowel disease or sarcoidosis [Afshari et al. 2001]. This means the IERs may actually be part of the underlying disease process that also causes osteoporosis rather than due to the medications used to treat it.

Two of the cohort studies just looked at bisphosphonates only [French and Margo, 2008; Etminan et al. 2012], whilst the most recent [Pazianas et al. 2013] looked at other treatments for osteoporosis too, and in summary the results are contradictory.

The study from the USA [French and Margo, 2008] used the national Veterans Health Administration (VHA) pharmacy and clinical databases for 2006 which contains information on approximately 5.7 million veterans. Within this cohort, 35,252 new prescriptions for bisphosphonates (oral or parenteral) were issued during 2006. During the same time period, 3736 new diagnoses of IERs were identified by International Classification of Diseases 9 (ICD-9) codes (uveitis, scleritis or iridocyclitis). The 30-day and 6-month rates of IERs were age-adjusted to the national age distribution of the overall population of veterans. In addition, data on inflammatory diseases were recorded along with coprescription of other medications implicated as causes of IERs (rifabutin, trimethoprim-sulfamethoxazole, diethylcarbamazine, metipranolol and cidovir). Results showed that 6 months after a new prescription for bisphosphonates, there had been 7.9 new cases of IERs per 10,000 individuals. This rate did not differ significantly from rates of IERs in veterans not prescribed bisphosphonates, with a relative risk of 1.23 [95% confidence interval (CI) 0.85–1.79]. Nearly 43% of patients with IERs after bisphosphonate use had a systemic inflammatory disease commonly associated with IERs including rheumatoid arthritis, ankylosing spondylitis, psoriasis, inflammatory bowel disease, systemic lupus erythematosus or sarcoidosis.

The study from Canada [Etminan et al. 2012] used the British Columbia Linked Health Database and looked at bisphosphonates and IERs. It was based on all residents of British Columbia who had visited an ophthalmologist between 2000 and 2007 (n = 989,591) and who has at least 1 year of prescription drug information. IERs were defined as either uveitis or scleritis recorded in the database. People were excluded from this study if they had received more than one prescription for bisphosphonate (n = 55,444) because the investigators wished to focus on first-time users of bisphosphonates. Cohort members were then followed until uveitis or scleritis developed, death, termination of health coverage or end of the study. In total, this study was based on 10,827 first-time bisphosphonate users and 923,320 participants never exposed to bisphosphonates. After adjustment for age, gender, some diseases (ankylosing spondylitis, diabetes, inflammatory bowel syndrome, systemic lupus erythematosus, multiple sclerosis, psoriasis, rheumatoid arthritis and sarcoidosis) and nonsteroidal anti-inflammatory (NSAID) medication use, results showed first-time users of bisphosphonates were at a higher risk of scleritis compared with nonusers [relative risk (RR) 1.38, 95% CI 1.26–1.50]. Without adjustment for NSAIDs (but with adjustment for the other variables above), first-time users of bisphosphonates also had an increased risk of uveitis (RR1.45, 95% CI 1.25–1.68) ; after adjustment for NSAID use, the association with uveitis was no longer present. This may suggest the association between uveitis and bisphosphonates is due to confounding by indication.

The study from Denmark [Pazianas et al. 2013] used the Danish National Prescription Database, which registers all prescriptions filled in Denmark, linked to the National Hospital Discharge Register, which covers all outpatient and inpatient diagnoses in the National Health Service, and included 88,202 people on osteoporosis medications between 1997 and 2007 (bisphosphonates: alendronate, clodronate, etidronate, ibandronate, or risedronate; raloxifene; strontium; and teriparatide). IERs were defined as those who filled a prescription for a topical eye steroid or had a new diagnosis of conjunctivitis, scleritis, episcleritis, keratitis or iridocyclitis using ICD-10 codes. The first-prescribed osteoporosis medication was used as the exposure of interest. Results showed that using unadjusted data alendronate and etidronate were associated with a higher number of people using topical eye steroids in the first year of treatment compared with the year prior to starting bisphosphonates (5.9% versus 5.1%, p < 0.05 for alendronate; 5.5% versus 4.7%, p < 0.05 for etidronate). However, no associations were seen after adjusting for underlying diseases, age and number of comedications. Furthermore, no difference was found in rates of IERs with bisphosphonate and nonbisphosphonate treatments after adjustment for underlying diseases and age. Overall, the incidence of IERs requiring topical steroids after initiation of anti-osteoporosis medication was low, and the greatest risk of IERs was seen amongst patients with seronegative rheumatoid arthritis, necrotizing vasculopathies and Sjogren’s syndrome, suggesting the underlying inflammatory disease is the most likely determinant.

All three of these large population-based cohort studies reported a low incidence of IERs associated with bisphosphonates or other osteoporosis medications, and a very low incidence of severe uveitis which is reassuring. Two of the studies concluded that the underlying inflammatory disease was the most likely determinant of the IER [French and Margo, 2008; Pazianas et al. 2013]. The other study which identified a small increased risk of scleritis in bisphosphonate users [Etminan et al. 2012], but not uveitis, was the smallest study in terms of numbers of bisphosphonate users and had more limited data on underlying diseases, perhaps meaning adjustment for confounders was less complete. However, this study arguably had the most robust data on eye disease as it was based on data collected directly from ophthalmologists, and was able to investigate scleritis and uveitis separately.

Pathogenesis

As discussed above, the difficulty with identifying whether medications for osteoporosis cause IERs is that there are many diseases that have IERs and osteoporosis as part of their clinical phenotype. This makes ‘confounding by indication’ an important consideration when reviewing the evidence and highlights the importance of identifying any pathogenic mechanism.

The potential pathogenesis of IERs after bisphosphonates is unknown, but the pattern suggests that the drug is secreted into the tears by the lacrimal gland and then causes transitory irritation to the mucous membranes [Fraunfelder and Fraunfelder, 2004; Peterson and Bedrossian, 2012]. This irritation may cause release of cytokines and other acute phase proteins in the eye, or cause activation of gamma delta T cells within the orbit. Toxicology studies have reported conjunctivitis and episcleral congestion in rabbits with larger doses of pamidronate than used therapeutically (100 times), and it was identified that pamidronate is secreted into tears [Mondelo et al. 1997]. Release of cytokines and synthesis of acute phase proteins is believed to be the possible mechanism for systemic bisphosphonate-induced inflammatory events, such as the influenza-like syndrome in the acute phase reaction seen in approximately one third of patients receiving their first dose of pamidronate, and a similar mechanism may cause IERs [Adami et al. 1987; Gallacher et al. 1989; Zojer et al. 1999]. However, no specific localization of cytokines and acute phase proteins to the tears or orbit has been demonstrated in response to bisphosphonates.

The release of the inflammatory mediators interleukin 6 (IL-6) and tumor necrosis factor (TNF-α) in response to bisphosphonates has been demonstrated both in vitro and in vivo with maximum levels seen 24 hours and 48 hours post pamidronate treatment, respectively [Sauty et al. 1996; Thiebaud et al. 1997].

As analogs of pyrophosphate, the bisphosphonates share several homologies with nonpeptide gamma delta T cell ligands that activate gamma delta T cells [Kunzmann et al. 1999, 2000] and there is evidence that bisphosphonates could activate these cells. Firstly, gamma delta T cells recognize naturally occurring small nonpeptide antigens and have high specificity for derivatives of pyrophosphate [Tanaka et al. 1994, 1995]. Secondly, synthetic nitrobisphosphonates inhibit farnesyl pyrophosphate synthase leading to isopentenyl pyrophosphate (IPP) accumulation and IPP is a potent trigger of gamma delta T-cell activation [Tanaka et al., 1994, Van Beek et al., 1999, Thompson et al. 2002]. See Figure 1.

Figure 1.

Model of the molecular mechanism underlying the acute phase reaction to bisphosphonates in vivo. Following administration of bisphosphonates (BP), cells such as peripheral blood mononuclear cells (PBMCs) internalize the BP, resulting in inhibition of farnesyl diphosphate synthase (FPPS) and the accumulation of isoprenoid lipid substrates such as isopentenyl pyrophosphate (IPP). IPP is then presented to gamma delta T cells (γδ T cell) causing their activation and the release of pro-inflammatory cytokines such as tumor necrosis factor alpha (TNFα), interferon gamma (IFNγ), and interleukin 6 (IL-6) characteristic of the acute-phase reaction. In addition, BPs may directly activate γδ T cells by acting as a nonpeptide ligand homolog and/or a phosphoantigen. Adapted from [Thompson and Rogers, 2006].

These mechanisms could help explain immediate IERs seen within 48 hours. It is more difficult to explain reactions occurring after weeks or months, but accumulation of bisphosphonates within the eye over time, perhaps influenced by physical and chemical properties of the drug and by individual patient-related factors including genetic predisposition, may explain delayed IERs.

Unanswered questions

The main question that is still unanswered is whether there is a true causal association between bisphosphonates or other anti-osteoporosis medications and IERs, or whether it is confounding by indication and is actually due to underlying inflammatory diseases that cause both osteoporosis and IERs. Alternatively, it is possible that bisphosphonates or other anti-osteoporosis medications are necessary but not sufficient to cause IERs on their own, and only do so in the presence of underlying inflammatory diseases or other factors that we have not yet identified such as particular medications or comorbid diseases. Careful epidemiological study could help by increasing the evidence base, although on its own would be insufficient to establish a causal relationship between osteoporosis medications and IERs. Standard methodologies are needed to classify people into users of bisphosphonates or other anti-osteoporosis medications, and standard definitions of IERs. Standardization of exposure and outcome across studies would also help produce a cohesive body of knowledge. In addition, further study of nonbisphosphonate medications is needed, particularly using population-based cohorts. Finally, there is no information available on potential pathogenic mechanisms of nonbisphosphonate medications.

Conclusion

The incidence of IERs after initiation of anti-osteoporosis treatments is low (approximately 8 per 10,000 bisphosphonate users) and the incidence of severe uveitis is very low. There are numerous case reports and case series reporting IERs after bisphosphonates, but none reporting IERs after other anti-osteoporosis medications. The highest quality epidemiological evidence currently available comes from three population-based cohort studies that report contradictory results. A potential pathogenic mechanism for which there is some evidence is direct irritation to the mucous membrane causing release of cytokines and other acute phase proteins in the eye, or causing activation of gamma delta T cells within the orbit. However, it is still not known whether there is a true causal association between bisphosphonates or other anti-osteoporosis medications and IERs, or whether it is confounding by indication and is actually due to underlying inflammatory diseases that cause both osteoporosis and IERs.