Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

1 Introduction

Pharmacogenomics is an evolving research discipline within ophthalmology. Open-angle glaucoma (OAG) and age-related macular degeneration (AMD) are common, worldwide causes of visual loss that are largely treated with pharmacologic therapies. Pharmacogenetic associations have been reported for both OAG (Schwartz et al. 2008; Moroi et al. 2009) and exudative AMD (Schwartz and Brantley 2011). At this time, genetic data are not typically used to make routine clinical decisions. However, as additional clinical trial results are collected, the potential benefits of pharmacogenomics knowledge in the care of patients with ophthalmic diseases will be better understood.

There are at least two potential roles for applying pharmacogenomics in the treatment of ophthalmic diseases. First, more therapies specifically targeted for individuals may lead to improved treatment outcomes and reduce patient exposure to inefficacious medications. Second, pharmacogenomics may lead to the development of novel therapies for these diseases.

2 Open-Angle Glaucoma

Control of intraocular pressure (IOP) is generally effective in delaying the progression of optic neuropathy and visual loss in patients with OAG, normal tension glaucoma (NTG), ocular hypertension, and related disorders (Costagliola et al. 2009a, b). Two of the major categories of medications used to lower IOP are β-adrenergic antagonists and prostaglandin analogues, both of which have a considerable rate of nonresponse. Pooled data from multiple randomized clinical trials reported a nonresponse rate of 28 % with the β-blocker timolol maleate and 18 % with the prostaglandin analogue latanoprost (Camras and Hedman 2003). There is currently no way to identify these nonresponders prior to initiation of therapy. This “trial and error” strategy may lead to extra office visits and exposure to additional medications for some patients.

2.1 β-Adrenergic Antagonists

The β-adrenergic antagonists include several nonselective agents (β1- and β2-blockers), such as timolol, and one β1-selective agent, betaxolol hydrochloride. The nonselective agents are generally more effective in IOP reduction than betaxolol (Allen et al. 1986). The high nonresponse rate associated with topical betaxolol is similar to the high nonresponse rate associated with systemic β1-blockers used to treat systemic hypertension and other diseases (Materson et al. 1993).

Most pharmacogenomic studies of β-adrenergic antagonists have focused on the polymorphisms in adrenergic receptor genes (Table 32.1). The β1-adrenergic receptor (β1-AR) gene contains two well-characterized single nucleotide polymorphisms (SNPs): Ser49Gly and Arg389Gly (Maqbool et al. 1999; Mason et al. 1999). The β2-adrenergic receptor (β2-AR) contains four common SNPs: Gly16Arg, Cys19Arg, Gln27Glu, and Thr164Ile (Green et al. 1993, 1994; Parola and Kobilka 1994; Liggett 2000). The α-adrenergic receptor (α-AR) contains multiple subtypes, many of which have well-characterized polymorphisms, including Del 301–303 in α 2B -AR and Del 322–325 in α 2C -AR (Flordellis et al. 2004). Furthermore, timolol is metabolized by cytochrome P40 2D6 (CYP2D6), and polymorphisms in CYP2D6 are associated with the efficacy of oral timolol in the treatment of systemic hypertension (McGourty et al. 1985).

Table 32.1 Pharmacogenomics of β-adrenergic antagonists
Full size table

The relationship between adrenergic receptor genotype and presence of OAG or NTG has been evaluated in several studies. In a Japanese study of 211 OAG patients, 294 NTG patients, and 240 controls, there was a significant association between NTG and the β 1 -AR Arg389Gly polymorphism (Inagaki et al. 2006). A US series of 299 OAG patients and 284 controls found no associations between β 2 -AR polymorphisms and clinical status (McLaren et al. 2007). In a Japanese series of 92 untreated NTG patients, associations were reported between diurnal IOP measurements and β 1 -AR Ser49Gly, α 2B -AR Del 301–303, and α 2C -AR Del 322–325 (Gao et al. 2010).

In a multiracial US series of 48 normal volunteers, the β 1 -AR Arg389 homozygote genotype was associated with higher baseline IOP and a greater magnitude of response to treatment with betaxolol. There were no associations with β 1 -AR Ser49Gly (Schwartz et al. 2005). An Austrian series of 89 normal volunteers treated with timolol reported no association between IOP response and the homozygous haplotypes of β 2 -AR polymorphisms Arg16/Gln27, Gly16/Gln27, and Gly16/Glu27 (Fuchsjager-Mayrl et al. 2005). In contrast, in a US series of 210 patients with OAG or suspected glaucoma, β 2 -AR Gln27 homozygotes were more likely to have a 20 % or greater decrease in IOP following treatment with topical β-blockers (McCarty et al. 2008).

At least two series have studied associations between genotype and systemic toxicity to ophthalmic timolol. In a Finnish group of 19 OAG patients and 18 normal volunteers treated with timolol, β 1 -AR Ser49 homozygotes had higher systolic blood pressure and higher diastolic blood pressure than Gly49 carriers following a head-up tilt test. In this series, the investigators also studied polymorphisms of CYP2D6 and the α-subunit of G protein (GNAS1). Subjects determined to be poor CYP2D6 metabolizers (zero functional alleles) had less favorable pharmacokinetic and pharmacodynamic parameters than other subjects with aqueous timolol, but not with hydrogel timolol. The CC allele of the thymine-by-cytosine replacement at the base 393 (T393C) in GNAS1 was associated with a lower change in diastolic blood pressure from rest to maximum during exercise (Nieminen et al. 2005). Similarly, in a Chinese series of 133 OAG patients, systemic bradycardia associated with topical timolol was more common with the CYP2D6 Arg296Cys polymorphism (Yang et al. 2009).

2.2 Prostaglandin Analogues

Latanoprost is a highly selective agonist of the prostaglandin F2α (FP) receptor (Stjernschantz et al. 1995). A Japanese series of 100 normal volunteers treated with latanoprost reported that the SNPs rs3753380 and rs3766355 in the FP receptor gene were associated with the magnitude of response to therapy (Sakurai et al. 2007). Using pathway analysis, the following related polymorphisms were studied and reported to have no relationship with IOP reduction: prostaglandin transporter T396A (van der Zwaag et al. 2002); fatty acid amide hydrolase P129T (Sipe et al. 2002); matrix metalloproteinase-1 promoter-1607 insG (Rutter et al. 1998); matrix metalloproteinase-2 promoter C-1306T (Price et al. 2001); matrix metalloproteinase-3 promoter-1171 delA (Ye et al. 1995); and matrix metalloproteinase-9 promoter C-1562T and CA repeats (Zhang et al. 1999; St. Jean et al. 1995).

3 Age-Related Macular Degeneration (AMD)

Age-related macular degeneration (AMD) is the leading cause of irreversible visual loss among the elderly in Western nations (Resnikoff et al. 2004). Patients with AMD are typically classified as having exudative (neovascular) disease if there is evidence of choroidal neovascularization (CNV) and non-exudative (non-neovascular) if there is not (Ambati et al. 2003).

3.1 Non-exudative AMD

At this time, there is no proven effective therapy to reduce visual loss due to non-exudative AMD. The Age-Related Eye Disease Study (AREDS) reported that a specific combination of antioxidants and zinc reduces progression to advanced disease and visual loss in certain patients with non-exudative AMD (Age-Related Eye Disease Study Research Group 2001).

A polymorphism in the complement factor H gene (CFH) is strongly associated with AMD presence (Edwards et al. 2005; Haines et al. 2005; Klein et al. 2005; Hageman et al. 2005) and progression (Seddon et al. 2007). A second major risk locus includes the ARMS2 (age-related maculopathy susceptibility 2, also known as LOC387715) and HTRA1 (HtrA serine peptidase 1) genes and is also strongly associated with AMD (Jakobsdottir et al. 2005; Rivera et al. 2005; Yang et al. 2006; Dewan et al. 2006). Because the two genes are in strong linkage disequilibrium and their effects are statistically indistinguishable, it has yet to be determined if it is ARMS2 or HTRA1 that is responsible for the association with AMD.

A subset of patients in the AREDS trials was evaluated with respect to polymorphisms in CFH and ARMS2. In this group from the AREDS, 264 of 876 AREDS category 3 and 4 patients (30.1 %) progressed to advanced AMD over 5 years. In these patients, nutritional supplementation with antioxidants and zinc was associated with a greater reduction in disease progression in patients with the CFH Y402H TT genotype (68 %) than in patients with the CC genotype (11 %). There were no significant differences in disease progression with respect to genotype at ARMS2 A69S (Klein et al. 2008).

3.2 Exudative AMD

A variety of pharmacologic treatments have demonstrated efficacy in the treatment of CNV secondary to exudative AMD. Photodynamic therapy (PDT) with verteporfin (Visudyne, Novartis, Basel, Switzerland) was reported to reduce the risk of visual loss in patients with predominantly classic CNV (Treatment of Age-related Macular Degeneration With Photodynamic Therapy (TAP) Study Group 1999). Intravitreal injection of VEGF inhibitors (Fig. 32.1) has now become standard treatment for exudative AMD (Kovach et al. 2012). At this time, four anti-VEGF agents are in clinical use. Pegaptanib (Macugen, Eyetech, Palm Beach Gardens, FL) (Gragoudas et al. 2004), ranibizumab (Lucentis, Genentech, South San Francisco, CA) (Rosenfeld et al. 2006; Brown et al. 2006), and aflibercept (Eylea, Regeneron, Tarrytown, NY) (Heier et al. 2011) are US FDA-approved for the treatment of exudative AMD. Bevacizumab (Avastin, Genentech, South San Francisco, CA) is FDA-approved for the systemic treatment of metastatic colorectal and other cancers (Yang et al. 2003), but is used extensively as an off-label intravitreal treatment of exudative AMD (Rosenfeld et al. 2005). Intravitreal triamcinolone acetonide (Fig. 32.2) has proven somewhat efficacious as an adjunctive therapy, especially when combined with PDT (Chan et al. 2009) or bevacizumab (Ahmadieh et al. 2011).

Fig. 32.1
figure 1

Photograph demonstrating the technique of intravitreal injection, left eye

Full size image
Fig. 32.2
figure 2

Fundus photograph, left eye, demonstrating the typical appearance of triamcinolone acetonide in the vitreous cavity following injection (Image courtesy of Harry W. Flynn, Jr., M.D.)

Full size image

Each of these agents has demonstrated efficacy, but there remains a persistent and unexplained variability among patients’ individual treatment response, especially with the anti-VEGF agents (Menghini et al. 2010). In addition, triamcinolone is associated with several adverse events, especially elevated IOP, which remains poorly understood (Smithen et al. 2004). PDT is delivered as an intravenous infusion combined with the use of a photoactivator. The anti-VEGF agents are delivered as intravitreal injections, which are associated with a low risk of endophthalmitis and other serious complications (Schwartz et al. 2009). Similar to medications used to treat OAG, there is currently no reliable way to identify nonresponders prior to treatment.

3.2.1 Photodynamic Therapy (PDT)

Most pharmacogenetic studies of PDT have focused on the AMD-associated variants CFH Y402H and ARMS2 A69S, although other genes have been studied, including C-reactive protein (CRP), VEGF (or VEGFA), pigment epithelium-derived factor (PEDF), and apolipoprotein E (APOE) (Table 32.2).

Table 32.2 Pharmacogenomics of photodynamic therapy
Full size table

In a series of 27 patients from England treated with PDT, those with the CFH Y402H CC genotype lost, on average, more letters of visual acuity than did patients with the CT genotype (Goverdhan et al. 2008). A subsequent series of 69 patients from the USA reported that mean visual acuity following PDT was worse in patients with CFH Y402H TT than with CT or CC, in patients with classic CNV but not occult CNV. There was no association seen between visual outcome and ARMS2 A69S (Brantley et al. 2009).

In a series of 110 patients from Japan, HTRA1 rs11200638 GG was associated with improved visual acuity outcomes and less risk of disease recurrence following PDT. In this study, the combination of CFH rs1410996 and rs2274700 was associated with a reduction in the time interval until disease recurrence following PDT. There was no association between PDT outcomes and CFH rs1061170 (Y402H) and rs800292; VEGF rs699947, rs1570360, and rs2010963; and PEDF rs12150053, rs12948385, rs9913583, and rs1136287 (Tsuchihashi et al. 2011).

Other series from Finland (Seitsonen et al. 2007), Israel (Chowers et al. 2008a), and Australia (Feng et al. 2009) reported no association between PDT outcomes and CFH Y402H. A subsequent series form Israel reported no associations between PDT and ARMS2 A69S or HTRA1 rs11200638 (Chowers et al. 2008b).

Polymorphisms in other genes have been linked to PDT outcomes. An Australian series of 273 patients reported an association between PDT outcomes and two of nine polymorphisms (rs2808635 GG and rs876538 AA) in CRP (Feng et al. 2009). In a series of 86 patients from Finland, two (rs699947 and rs2146323) of three VEGF polymorphisms were associated with PDT treatment outcomes, using a binary responder/nonresponder classification (Immonen et al. 2010).

A series of 90 Italian patients treated with PDT for classic CNV was screened for polymorphisms in genes related to coagulation, including factor V G1691A, prothrombin G20210A, factor XIII-A G185T, methylenetetrahydrofolate reductase (MTHFR) C677T, methionine synthase A2756G, and methionine synthase reductase A66G. Using a binary responder/nonresponder classification, responders were associated with prothrombin G20210A and MTHFR 677T carriers, and nonresponders were associated with factor XIII-A 185T carriers (Parmeggiani et al. 2007). The same group subsequently reported a series of 84 patients treated with PDT for occult CNV; in these patients, responders were associated with the combination carriers of factor V 1691A and prothrombin 20210A alleles, while nonresponders were associated with factor XIII-A G185T (Parmeggiani et al. 2008). The same investigators also reported a series of 234 patients treated with PDT for CNV secondary to pathologic myopia, rather than AMD. They reported an association between responders and carriers of MTHFR 677T allele and between nonresponders and factor XIII-A 185 GT/TT genotypes (Parmeggiani et al. 2010).

3.2.2 Anti-VEGF Agents

At this time, all reported pharmacogenetic studies of anti-VEGF agents have involved bevacizumab or ranibizumab. Again, most studies have focused on CFH Y402H and ARMS2 A69S (Table 32.3).

Table 32.3 Pharmacogenomics of anti-vascular endothelial growth factor therapy
Full size table

In a series of 86 US patients treated with bevacizumab, patients with the CFH Y402H CC genotype experienced less favorable visual results than other two genotypes (TC and TT), while there were no associations with ARMS2/LOC387715 (Brantley et al. 2007). Similar results were reported in a series of 197 patients from Austria, in which patients with CFH Y402H CC on average lost visual acuity, while patients with CC or TC on average gained visual acuity following treatment with bevacizumab (Nischler et al. 2011).

In a series of 156 US patients treated using an as-needed protocol with ranibizumab, the CFH Y402H CC polymorphism correlated with an increased number of injections performed (Lee et al. 2009). Similar results were reported in a series of 90 patients from Poland, in which CFH Y402H CC and ARMS2 A69S TT were associated with relatively less favorable visual outcomes following treatment with ranibizumab (Teper et al. 2010). In a series of 243 Swiss patients treated with ranibizumab and classified as poor responders vs. good responders, CFH Y402H CC was associated with poor responders and the combination of heterozygous genotypes at CFH Y402H and frizzled homolog 4 (FZD4) rs10896563 was associated with good responders; ARMS2, HTRA1, VEGFA, complement factor B (CFB), kinase insert domain receptor (KDR), and low-density lipoprotein receptor-related protein 5 (LRP5) were not associated with treatment outcomes (Kloeckener-Gruissem et al. 2011). A UK series of 104 patients treated with ranibizumab reported an association between visual acuity improvement of five letters or more and CFH Y402H CT compared with CFH Y402H TT; the investigators also reported nonsignificant trends towards more favorable outcomes associated with HTRA1 and VEGF polymorphisms (McKibbin et al. 2012).

In a series of 172 Australian patients treated with bevacizumab, ranibizumab, or a combination of the two medications, the APOE ε4 allele associated with better visual acuity outcomes compared with the APOE ε2 allele (Wickremasinghe et al. 2011). The Comparison of AMD Treatments Trials (CATT) was a prospective randomized clinical trial comparing ranibizumab to bevacizumab. A subset of 834 patients enrolled in CATT underwent pharmacogenetic testing. No statistically significant differences were found with respect to variants at CFH, ARMS2, HTRA1, and C3 (Hagstrom et al. 2013).

3.2.3 Triamcinolone Acetonide

In a pilot study of 52 US patients treated with intravitreal triamcinolone acetonide for a variety of retinal diseases, including wet AMD, there were no significant associations between visual outcome and six common polymorphisms of the glucocorticoid receptor gene (GR) (Gerzenstein et al. 2008). In another study of 102 Hungarian patients, the GR N363S polymorphism was associated with steroid-induced ocular hypertension upon treatment with topical corticosteroids (prednisolone acetate 0.5 %, flurometholone 0.1 %, or combined) after photorefractive keratectomy (Szabo et al. 2007).

Other genes have been investigated with respect to steroid-induced IOP elevation, including spliceosome proteins (Xu et al. 2003; Yan et al. 2010) and immunophilins (Zhang et al. 2008). In a US series of 197 OAG patients, 107 steroid responders, and 400 normal volunteers, there were no associations with polymorphisms in GR, the immunophilin FKBP4, or the spliceosomes SFRS3, SFRS5, or SFRS9 (Fingert et al. 2010).

4 Summary

At this time, pharmacogenomics is primarily useful as a research tool within ophthalmology. Pharmacogenomic testing of ophthalmic medications is not generally performed in clinical settings.

The Pharmacogenomics Knowledgebase (PharmGKB) states: “Timolol is a non-selective beta-adrenergic agonist applied to the eye to reduce intraocular pressure. It is metabolized via CYP2D6. The FDA recommends, but does not require, genetic testing prior to initiating or reinitiating treatment with Timolol (Istalol)” (http://www.pharmgkb.org/drug/PA451690, accessed 9/12/12). However, the FDA Table of Pharmacogenomic Biomarkers in Drug Labels, which lists medications that include pharmacogenomic information on their labels, includes the systemic beta-blockers carvedilol, metoprolol, and propanolol, but not timolol or any other ophthalmic medication (http://www.fda.gov/Drugs/ScienceResearch/ResearchAreas/Pharmacogenetics/ucm083378.htm, accessed 9/12/12).

Although several studies of glaucoma and AMD therapeutics have reported statistically significant associations between genotype and treatment response, none of these associations has been confirmed in large-scale clinical trials. In many instances, the data appear inconsistent among studies. This may be due to important underlying differences in baseline genetic characteristics between studies, particularly between studies from different continents. Different studies also used different enrollment criteria (normal volunteers vs. glaucoma patients, classic CNV vs. occult), different study endpoints (visual acuity, anatomic response, number of re-treatments required), and statistics (continuous outcomes vs. binary “good responder”/“poor responder” outcomes).

As data from clinical trials continues to be collected, various pharmacogenomic relationships may become clearer.