Punctate Inner Choroidopathy and Multifocal Choroiditis with Panuveitis Share Haplotypic Associations with IL10 and TNF Loci

Abstract

Purpose.: The white-dot syndromes are a heterogenous group of chorioretinal disorders that have many common clinical features. Whether these disorders represent distinct clinical entities or different manifestations of the same disease warrants further interrogation. Two white-dot syndromes were investigated, with closely overlapping phenotypes—multifocal choroiditis with panuveitis (MFCPU) and punctate inner choroidopathy (PIC)—for differences in clinical course and genotype frequency at IL10 and TNF loci, known to be associated with noninfectious uveitis.

Methods.: Twelve polymorphisms were genotyped, spanning the TNFA and IL10 genomic regions, in 61 patients with MFCPU or PIC and 92 population controls from the United Kingdom and Republic of Ireland.

Results.: There were clear differences in clinical course between patients with MFCPU and PIC which had prognostic significance. However, both patient groups demonstrated similar associations with the IL10 haplotype, IL10htSNP2(−2849)AX/htSNP5(+434)TC and negative associations with the TNF haplotype, LTA+252A/TNFhtSNP1(−308)G/TNFhtSNP2(−238)G/TNFhtSNP3(+488)A/TNFd3.

Conclusions.: Despite clear differences in clinical course and outcome, MFCPU and PIC may still represent two manifestations of the same disease, given their similar genetic associations with IL10 and TNF loci, which are known to be associated with noninfectious uveitis and autoimmunity, in general. Definitive proof will necessitate genomewide sequence analysis. However, the data also support the notion that epigenetic factors have a strong effect on clinical phenotype.

The white-dot syndromes encompass a heterogenous group of chorioretinal disorders, characterized by white dots in the posterior pole of various distributions and morphologies. ^{1 –3} On the one hand, the diversity of clinical presentation suggests distinct clinical entities; on the other, collectively, these disorders may represent a spectrum of disease, the expression of which varies as a consequence of the robustness of host immunity. ^4,5 In support of the latter, these disorders have many clinical features in common, including female preponderance, unexplained visual field defects usually contiguous with the blind spot, photopsias, and reduced electroretinographic amplitudes. ⁶  

Central to this debate, two white-dot syndromes demonstrate closely overlapping phenotypes: both multifocal choroiditis with panuveitis (MFCPU) and punctate inner choroidopathy (PIC) are characterized by multiple small (50–300 μm), indistinct, yellow-gray lesions that pathologically represent inflammatory microgranulomata at the chorioretinal interface and evolve into punched-out, discrete atrophic scars during inactive phases of disease. ⁷ In addition, both disorders are frequently associated with peripapillary atrophy and complicated by choroidal neovascularization (CNV). Where they differ is in the degree of associated anterior chamber and vitreous activity: signs of intraocular inflammation, particularly cellular infiltration of the chambers, are marked in patients with MFCPU, but are only minimal or absent in PIC. Other investigators distinguish the disorders by their clinical course: MFCPU is often chronic and relapsing, whereas PIC is recurrent with phases of complete remission. Since both disorders may look identical on clinical examination during quiescent phases of disease, either due to remission or effective disease control with immunosuppressants, the two disorders may well represent two ends in the severity spectrum of a single disease. 

Collectively, the white-dot syndromes have been linked to a variety of infectious etiologies, since they are often preceded by prodromal illness or are epidemiologically associated with specific infections. In the Ohio–Mississippi valley, presumed ocular histoplasmosis (POHS) is associated with infection with the endemic Histoplasma capsulatum. ⁸ Furthermore, lymphocyte stimulation by H. capsulatum antigens correlates with disease activity in patients with POHS. ⁹ Histoplasmosis antigens and organisms have been detected in the affected eyes of POHS patients. ^10,11 Though PIC is clinically indistinguishable from POHS, ⁸ no such association is found with PIC or cases indistinguishable from POHS in regions where H. capsulatum is not found. ¹² Hence, the evidence linking PIC to a direct infectious etiology is speculative, although it remains possible that another unknown organism is responsible. Similarly, evidence linking MFCPU to a direct infectious etiology is not overwhelming: MFCPU was associated with Epstein-Barr virus in one study, ¹³ but these findings were not replicated subsequently. ^14,15 Other white-dot syndromes are known to be precipitated by vaccination. ^16,17 If we assume that the white-dot syndromes share a similar immunopathogenesis, given their overlap in phenotype, then preceding infection may be the drive systemically to precipitate innate immune activation and autoinflammation or a breakdown in tolerance to elaborate autoimmunity. ¹⁸  

In support of this hypothesis, most evidence for the pathogenesis of uveitis comes from animal models, in which uveitis is induced, for example, by immunization of a susceptible animal with one of many retinal antigens, emulsified in heat-killed Mycobacterium butyricum and an adjuvant. This protocol will activate innate immune mechanisms that facilitate a proinflammatory antigen-specific T_H1/T_h17 adaptive immune response. Current evidence suggests that both T_H1 and T_H17 effector cells can independently induce tissue damage in mouse models of uveitis, largely through the production of the proinflammatory cytokine tumor necrosis factor (TNF)-α. ^{19 –21} Furthermore, differences between rodent strains in their susceptibility to the induction of uveitis is determined, in part, by their constitutive and stimulated levels of expression of TNFα and the counterregulatory interleukin (IL)-10, by Müller glia, RPE, microglia, and infiltrating T cells. ^{22 –25}  

If the variable susceptibility of different animal species and strains to the induction of uveitis is suggestive of a genetic predisposition to this disease, ²⁶ then evidence that uveitic patients may have a more general inherited predisposition to autoimmunity/autoinflammatory (AI) disease comes from the inheritance of uveitis in multiplex families ^27,28 and the association of uveitis with known monogenic (Blau syndrome) ²⁹ and polygenic (Crohn's disease) inflammatory disorders. ^30,31 In one study, 3% of PIC patients reported a personal history of AI disease, whereas 26% reported a family history of AI disease. ³² While associations between POHS and HLA-B7 and HLA-DR2 have been reported in a small number of patients, ^{33 –35} PIC and MFCPU have not been consistently associated with any HLA antigens. ^36,37 In contrast, the association between birdshot retinochoroidopathy (BRC) and HLA-A29 is one of the strongest described for any AI disease. ³⁸ Hence, if there is a common inherited predisposition to the white-dot syndromes, it is more likely to involve non-HLA genes, such as TNFA and IL10, since the levels of expression of these cytokines are key determinants of susceptibility to autoimmunity in animal models. 

Previously, we have shown that three haplotype-tagging single-nucleotide polymorphisms (htSNPs) in the IL10 gene: rs6703630, rs2222202, and rs3024490 are significantly associated with susceptibility to noninfectious uveitis, while an LTA+252AA/TNFhtSNP2GG haplotype (rs909253 and rs361525) is protective. ³⁹ In this study, using both genetic and clinical markers of disease course and severity, we sought to determine whether MFCPU and PIC, largely distinguished by the degree of associated clinically visible inflammation, do indeed represent two distinct clinical entities, or two extremes in severity of the same disorder, by addressing the following questions. First, are there clear differences between PIC and MFCPU, in clinical course and their association with polymorphisms that increase the risk and severity of uveitis in general; and secondly, are there differences between the two syndromes in their association with different polymorphisms within the same genes, IL10, TNFA, and the HLA genes, that might explain why patients with MFCPU mount a more clinically overt immune response than do others with PIC. 

Materials and Methods

Subjects

Subjects were recruited from four regional centers in Bristol (Bristol Eye Hospital), Aberdeen (Grampian University Hospitals), Dublin (The Royal Victoria Eye and Ear Hospital), and London (Moorfields Eye Hospital) as part of a larger study. ³⁹ Informed consent was obtained from all participants, after explanation of the nature and possible consequences of the study. Subjects were Caucasians of British or Irish descent for at least two generations. Ethical approval was given by each center, and the study adhered to the tenets of the Declaration of Helsinki. 

All subjects (61 patients, 92 controls) were given a full ophthalmic examination and categorized into three diagnostic groups: MFCPU, PIC, or control. MFCPU and PIC fall under the SUN (Standardized Uveitis Nomenclature) Working Group classification of posterior uveitis with clinical evidence of multifocal choroiditis. ⁴⁰ All affected individuals had multiple chorioretinal lesions in the posterior pole, including punched-out atrophic scars: those patients with significant vitritis, with or without anterior uveitis, at recruitment or during their follow-up (including disease onset, and SUN–defined recurrences and relapses ⁴⁰ ) were classified as having MFCPU; those patients with no evidence of anterior uveitis/vitritis, nor any history of such during their follow-up were classified as having PIC. Patients with transient posterior pole lesions (e.g., multiple evanescent white-dot syndrome or acute posterior multifocal placoid pigment epitheliopathy) were not included. Similarly, patients with a characteristic phenotype for BRC were excluded due to the strong prior association between BRC and HLA-A29. Control subjects were excluded if they had any history of AI disease (e.g., type 1 diabetes mellitus, ankylosing spondylitis, rheumatoid arthritis [RA], systemic lupus erythematosus [SLE], chronic obstructive pulmonary disease, ischemic heart disease, or neoplasia), or an eye-specific disorder with a significant genetic predisposition (age-related macular degeneration [AMD], glaucoma). 

All patients managed at the four regional centers had routine diagnostic and pretreatment investigations as part of the larger study. ³⁹ Patients' clinical courses were documented according to SUN guidelines ⁴⁰ for mode of onset, attack duration, course of an attack, ocular remission or relapse, from disease onset to a common census date 5 years from the start of the study (more fully described elsewhere ³⁹ ). Furthermore, complications and surgical procedures were documented in a standardized pro forma. 

Genotype associations were determined for three parameters of disease severity:

1.  
   Ocular remission according to SUN guidelines. ⁴⁰
2.  
   Maintenance immunosuppression was defined as the most recent combination of immunosuppressants to consistently control disease activity for at least 3 months, with no increase in immunosuppression during this period (more fully described elsewhere ³⁹ ).
3.  Visual outcome, assessed by  
   •  
     Visual acuity (VA) at the census date,
   •  
     Change in VA from disease onset to the census date (defined, according to SUN guidelines, as a decrease in Snellen VA of >3 lines ⁴⁰ ).

Genotyping

HtSNPs in the IL10 and TNF genomic regions were selected and genotyped using published methods, ³⁹ and Hardy-Weinberg probabilities were calculated for the larger cohort. ³⁹ The TNFd microsatellite polymorphism was genotyped as previously described. ⁴¹ HLA class I (A, B, and C) and II (DRB1, DRQA1, and DRQB1) typing was performed with sequence-specific primers (SSP-PCR) at medium resolution. ⁴²  

Statistical Analyses

Demographic information, clinical course parameters, patient, and control genotype distributions were compared between dichotomous groups using the two-tailed χ² test or Fisher's exact test, where appropriate (SPSS 14.0; SPSS UK Ltd., Woking, UK, and Epi Info 6.04d; Centers for Disease Control and Prevention, Atlanta, GA). The Bonferroni correction was applied to genotypic data to adjust for number of comparisons (n = total number of loci). 

Distributions of ordinal phenotypic characteristics were compared using the Kruskal-Wallis nonparametric test, while normally distributed continuous data were compared between groups using the t-test. Snellen VAs were converted to logMAR for analyses. 

Modeling of haplotypes, genotypes and allelic associations was performed using the UNPHASED application. ⁴³  

Results

Differences in Clinical Course between MFCPU and PIC Patients Have Prognostic Implications

The demographics of the control and white-dot syndrome groups combined were not significantly different for age and sex. In both patient groups, there was a similarly high prevalence of AI disease in patients' self-reported personal medical history and family history (Table 1). However, patients with PIC and MFCPU differed significantly in age, but not in sex preponderance (Table 1), since patients with MFCPU were generally older. ⁴⁴  

Table 1.

 

Demographic Information on Recruited Subjects

Uveitis Classification	Number in Group	Age at Recruitment (y)		Sex		Patients with a Personal History of AI Disease	Patients with a Family History of AI Disease
		Mean	Range	Male	Female
MFCPU	30	57*	15–89	10	20	4	5
PIC	31	40*	26–64	5	26	4	8
Combined white-dot group	61	48	15–89	15	46	8†	13‡
Healthy controls	92	49	21–89	27	65

*  Mean ages of MFCPU and PIC patients were significantly different (t-test, P < 0.001).

†  Of eight patients with AI or inflammatory disease other than uveitis, two had psoriasis, two had asthma, one had rheumatoid arthritis (RA), one had AI hemolytic anemia, one had seronegative spondyloarthropathy, and one had both cold agglutinin disease and primary hyperparathyroidism.

‡  Of 13 patients with a self-reported family history of AI or inflammatory disease, five reported RA, two reported an undefined arthritis, one reported MS, one reported ankylosing spondylitis, one reported psoriasis, and one reported juvenile arthritis in the family history. In addition, there were two fraternal sisters recruited to the study, both affected by MFCPU.

Although the two patient groups had been followed up for an equivalent length of time by the census date (on average, 9.9 [MFCPU] vs. 8.1 [PIC] years), there were clear clinical differences between them in mode of onset, attack duration, and episodes of remission (Table 2). MFCPU patients more frequently experienced chronic inflammation, requiring systemic maintenance immunosuppression, whereas PIC patients experienced recurrent bouts of disease—either new areas of choroiditis or recurrent CNV—and were more likely to achieve complete remission (quiescence off all medication). Similarly, MFCPU patients were significantly more likely to require higher levels of immunosuppression (Mann–Whitney test, P = 0.003). 

Table 2.

 

Comparison of Clinical Course of MFCPU and PIC Patients

Clinical Course	MFCPU (n = 30)	PIC (n = 31)	P
Eyes affected
Left only	16.7 (5)	29.0 (9)	NS
Right only	0 (0)	12.9 (4)	NS
Both eyes	83.3 (25)	58.1 (18)	0.031
Onset
Sudden	13.3 (4)	61.3 (19)	<0.001
Insidious	73.3 (22)	29.0 (9)	0.001
Missing data	13.3 (4)	9.7 (3)
Duration of attack
Limited	3.3 (1)	19.4 (6)	NS
Limited then persistent	0 (0)	3.2 (1)	NS
Persistent	96.7 (29)	67.7 (21)	0.006
Missing data	0 (0)	9.7 (3)
Course of attack
Acute	6.7 (2)	9.7 (3)	NS
Recurrent	13.3 (4)	48.4 (15)	0.005
Chronic	80.0 (24)	32.2 (10)	<0.001
Missing data	0 (0)	9.7 (3)
Remissions
Achieved remission	5 (12)	74.2 (23)	0.007
Missing data	0 (0)	3.2 (1)
Relapses on maintenance treatment (Rx)
Patients requiring Rx	93.3 (28)	51.6 (16)	<0.001
Missing data	0 (0)	3.2 (1)
Patients relapsing despite Rx	53.6 (15/28)	25.0 (4/16)	NS
Patients achieving control on Rx	96.4 (27/28)	93.8 (15/16)	NS
Patients uncontrolled despite Rx	3.6 (1/28)	6.2 (1/16)	NS
Length of follow-up, y
Mean time since onset at census date	9.9	8.1	NS
Range	2.8–18.4	0.5–37.4

 Data are the percentage of the group (number of patients). NS, nonsignificant (P > 0.05).

Although the VA of the worse eye of both patient groups were comparable at presentation and at the census date, MFCPU patients were more likely to lose vision in both eyes (Table 3). In addition, MFCPU patients tended to experience more dramatic reductions in acuity (Fig. 1). 

Table 3.

 

Change in Visual Acuity of MFCPU and PIC Patients up to the Census Date

	Patients with VA <LogMAR 0.4 in Worse Eye at Presentation	Patients with VA <LogMAR 0.4 in Worse Eye at Census Date	Drop in VA >3 LogMAR Lines in One Eye	Drop in VA >3 LogMAR Lines in Both Eyes*
MFCPU	50 (15)	63.3 (19)	56.7 (17)	20 (6)
PIC†	48.4 (15)	58.1 (18)	51.6 (16)	0 (0)

 Data are the percentage of the group (number of patients).

*  Data were missing for one PIC patient at onset and two at the census date.

†  Difference between PIC and MFCPU groups is significant (Fisher exact test, P = 0.011).

Figure 1.

 

Change in logMAR visual acuity of the worse eye in (A) MFCPU and (B) PIC patients from disease onset to the census date, a date common to all patient groups and regional centers and occurring after study recruitment was complete, 5 years from the start of the study (more fully described in Atan et al. ³⁹ ).

Complications were prevalent in both patient groups. In concordance with the published literature, ⁴⁴ CNV was significantly more frequent in PIC, whereas cataract, epiretinal membrane formation, cystoid macular edema, and glaucoma were more frequent in MFCPU (Fig. 2, Table 4). Consequently, two thirds of the patients in each group needed a surgical procedure. Photodynamic therapy (PDT), intravitreal triamcinolone, and anti-vascular endothelial growth factor (VEGF) agents were more frequently administered to PIC patients (to treat CNV), whereas cataract extraction was the most frequent procedure undergone by the MFCPU patients (Fig. 3, Table 5). 

Figure 2.

 

A comparison of the frequencies of the most common complications to affect (A) MFCPU and (B) PIC patients, as a function of the total number of occurrences of each complication. CNVM, choroidal neovascular membrane; ERM, epiretinal membrane; CMO, cystoid macular edema; Other, including retinal detachment (see Table 4 for full list).

Table 4.

 

Frequency of Complications in MFCPU and PIC Patients

Complication	MFCPU	PIC*	P
Choroidal neovascular membrane	23.3 (7)	86.7 (26)	<0.001
Cataract	70.0 (21)	16.7 (5)	<0.001
Epiretinal membrane	53.3 (16)	3.3 (1)	<0.001
≥1 episode of:
Acute CMO (lasting <6 months)	40.0 (12)	3.3 (1)	<0.001
Chronic CMO (lasting >6 months)	33.3 (10)	0 (0)	0.001
Glaucoma	33.3 (10)	0 (0)	0.001
Ocular hypertension or steroid response	36.7 (11)	16.7 (5)	NS
≥1 episode of:
Acute papillitis (lasting <6 months)	23.3 (7)	3.3 (1)	NS
Chronic papillitis (lasting >6 months)	13.3 (4)	0 (0)	NS
Enlarged blind spot	10.0 (3)	0 (0)	NS
Macular scar or central scotoma	10 (3)	6.7 (2)	NS
Retinal detachment/retinal break	6.7 (2)	3.3 (1)	NS
Neovascularization (of the optic disc, iris or elsewhere), with or without vitreous hemorrhage	6.7 (2)	0 (0)	NS
Nonglaucomatous optic atrophy	6.7 (2)	0 (0)	NS
Band keratopathy	6.7 (2)	0 (0)	NS
Posterior synechiae/iris bombe/cyclitic membrane	6.7 (2)	0 (0)	NS
Visual field loss	6.7 (2)	0 (0)	NS
Retinoschisis	3.3 (1)	0 (0)	NS
Branch retinal vein occlusion	3.3 (1)	0 (0)	NS
RPE detachment	3.3 (1)	0 (0)	NS
Macular ischaemia	3.3 (1)	0 (0)	NS
Subretinal fibrosis	0 (0)	3.3 (1)	NS

 Data are the percentage of the group (number of patients).

*  Data were missing for one PIC patient.

Figure 3.

 

Frequency of each type of surgical procedure undergone by (A) MFCPU and (B) PIC patients, as a function of the total number of occurrences of each procedure. IV, intravitreal; PI, peripheral iridectomy/iridotomy; PDT, photodynamic therapy.

Table 5.

 

Surgical Procedures Required by MFCPU and PIC Patients

Surgical Procedure	MFCPU	PIC*	P
Cataract extraction	63.3 (19)	3.3 (1)	<0.001
Photodynamic therapy	0 (0)	50.0 (15)	<0.001
Intravitreal steroid	3.3 (1)	40.0 (12)	0.001
YAG or surgical capsulotomy	26.7 (8)	0 (0)	0.005
Intravitreal anti-VEGF agent	0 (0)	20.0 (6)	0.02
Argon/xenon laser to posterior segment	13.3 (4)	16.7 (5)	NS
Vitrectomy	13.3 (4)	6.7 (2)	NS
Trabeculectomy	6.7 (2)	0 (0)	NS
Cyclodiode laser	6.7 (2)	0 (0)	NS
Peripheral iridectomy/iridotomy	6.7 (2)	0 (0)	NS
Other (exenteration for second diagnosis)	3.3 (1)	0 (0)	NS

 Data are the percentage of the group (number of patients).

*  Data was missing for one PIC patient.

Susceptibility and Resistance to the White-Dot Syndromes Are Conferred by IL10 and TNF Haplotypes

Susceptibility to the white-dot syndromes, as a combined group, was associated with one IL10 locus: IL10htSNP5 (P _c < 0.012). In addition, there was a relationship between the IL10htSNP2 locus and the white-dot syndromes that lost significance after correction for multiple comparisons (Table 6). Although the IL10htSNP5 locus was significantly associated with MFCPU but not PIC, there were no significant differences in genotype or allele frequencies between the MFCPU and PIC groups at any IL10 or TNF locus (Table 6). 

Table 6.

 

Associations between Polymorphic Loci and Susceptibility to the White-Dot Syndromes

Locus	MFCPU		PIC		Combined		Comparison of Genotype Frequencies: MFCPU versus PIC
	P uncorr	P c	P uncorr	P c	P uncorr	P c	P uncorr	P c
IL10htSNP1 (−3545) rs1800890	0.905	NS	0.128	NS	0.454	NS	0.231	NS
IL10htSNP2 (−2849) rs6703630	0.091	NS	0.016	NS	0.009	NS	0.595	NS
IL10htSNP3 (−1082) rs1800896	0.046	NS	0.843	NS	0.117	NS	0.339	NS
IL10htSNP4 (−819) rs1800871	0.224	NS	0.838	NS	0.182	NS	0.611	NS
IL10htSNP5 (+434) rs2222202	<0.001	<0.012	0.79	NS	<0.001	<0.012	0.117	NS
IL10htSNP6 (+504) rs3024490	0.119	NS	0.128	NS	0.033	NS	0.522	NS
IL10htSNP7 (+1847) rs3024493	0.641	NS	0.502	NS	0.587	NS	0.582	NS
LTA+252 rs909253	0.803	NS	0.020	NS	0.211	NS	0.044	NS
TNFhtSNP1 (−308) rs1800629	0.220	NS	0.316	NS	0.263	NS	0.343	NS
TNFhtSNP2 (−238) rs361525	0.031	NS	0.843	NS	0.182	NS	0.063	NS
TNFhtSNP3 (+488) rs1800610	0.325	NS	0.838	NS	0.711	NS	0.246	NS
TNFd UniSTS 256848	0.148	NS	0.284	NS	0.349	NS	0.089	NS

 Two-sided P values are calculated by χ² analysis with Bonferroni correction for the number of loci. NS, not significant; P > 0.05; P _uncorr, two-sided uncorrected probability; P _c, Bonferroni- corrected probability (n = 12) (P _c = P × 12).

We next modeled associations between the white-dot syndromes and loci, IL10htSNP2 and IL10htSNP5, using UNPHASED, since the results above suggested that susceptibility to the white-dot syndromes might be determined by both loci. Indeed, the association between the IL10htSNP2 locus and the white-dot syndromes was significant after conditioning on IL10htSNP5 genotypes (P = 0.041). In particular, one haplotype was significantly associated with disease in all patient groups: IL10htSNP2(−2849)AX/IL10htSNP5 (+434)TC was more frequent in the combined (19.3%, P = 0.00064), MFCPU (21.1%, P = 0.00087), and PIC groups (17.7%, P = 0.0037) than in the controls (4.5%). Of all the uveitides investigated in the larger study, MFCPU and PIC were the only groups to demonstrate this haplotypic association with IL10 (data not shown). Further analyses revealed that the risk-conferring alleles were IL10htSNP2A and IL10htSNP5T. Both alleles were associated with the white-dot syndromes, with odds ratios of 2.5 and 3.0, respectively (Table 7). 

Table 7.

 

Associations between IL10htSNP2A and IL10htSNP5T with Susceptibility to the White-Dot Syndromes

Subject Group	IL10htSNP2A (−2849) rs6703630	P	Odds Ratio (95% CI)	IL10htSNP5T (+434) rs2222202	P	Odds Ratio (95% CI)
% Combined white-dot syndrome carriers	65.6 (40/61)	0.007	2.5 (1.3–4.8)	88.5 (54/61)	0.013	3.0 (1.2–7.5)
% MFCPU carriers	60.0 (18/60)	0.116	2.0 (0.8–4.5)	96.7 (29/30)	0.004	11.4 (1.5–88.2)
% PIC carriers	71.0 (22/31)	0.008	3.2 (1.3–7.6)	80.6 (25/31)	0.328	1.6 (0.6–4.5)
% Control carriers	43.5 (40/92)			71.7 (66/92)

 Data are the percentage of subjects. P values are two-sided. Significant results are in bold.

Allelic IL10 haplotypes occurring with >5% frequency were inferred using the application, PHASE. ⁴⁵ However, none were significantly associated with the white-dot syndromes, either combined or as two distinct groups. 

There were no associations between the TNF loci and the white-dot syndromes, combined or as two distinct groups (Table 6). Only one haplotype was negatively associated with disease in the combined group only, LTA+252A/TNFhtSNP1G/TNFhtSNP2G/TNFhtSNP3A/TNFd3 (0% vs. 7% of controls, P _c = 0.023). No TNF haplotypes were associated with either susceptibility or resistance to disease in the individual patient groups. 

HLA-DR2 Is More Frequent in Patients with PIC Than in Those with MFCPU

We next looked for an HLA association with disease in the combined and individual patient groups, with particular focus on HLA-B7 and HLA-DR2, because of their link to the phenotypically similar, POHS. ^{33 –35} HLA-B7 did not occur significantly more frequently in the combined (14.9% vs. 13.2% in controls, P = 0.72), MFCPU (13.5% vs. 13.2% in controls, P = 0.86), or PIC groups (16.1% vs. 13.5%, P = 0.58). HLA-DRB1*15 (HLA-DR2) was more frequent in the PIC group (25.8% vs. 15.7% in controls, P _uncorr = 0.02) but not the MFCPU (11.5% vs. 16.1%, P _uncorr = 0.21) or combined groups (19.3% vs. 16.1%, P _uncorr = 0.28). 

No combined HLA-TNF haplotypes inferred by PHASE were significantly associated with susceptibility or resistance to disease in either the combined or individual patient groups (data not shown). 

Severity of Disease Is Not Associated with Polymorphisms of the TNF and IL10 Loci

There were no associations between the TNF and IL10 polymorphisms investigated in this study and three parameters of disease outcome (ocular remission, maintenance immunosuppression and visual outcome) in either the combined or individual patient groups after Bonferroni correction (data not shown). 

Discussion

In this study, we chose to investigate two closely related white-dot syndromes, MFCPU and PIC, to determine whether they might represent two distinct clinical entities or two ends in the severity spectrum of a single disorder, using both genetic and clinical markers of disease. The most striking phenotypic similarity between the two syndromes is the appearance of their chorioretinal lesions. The most significant difference between them is their degree of associated anterior chamber and vitreous inflammation. In a British and Irish cohort, we found that the distinction between MFCPU and PIC has prognostic value: MFCPU is more frequently associated with a nonremitting clinical course, poor visual prognosis for both eyes, and normally requires higher levels of maintenance immunosuppression. Hence, a significant amount of anterior chamber/vitreous inflammation in a patient with characteristic punched-out chorioretinal scars should alert the physician to using a higher level of immunosuppression to control inflammation and protect the other eye from visually threatening complications. However, although these differences in clinical course clearly distinguish the two groups, they are still consistent with the notion that MFCPU and PIC represent two manifestations of the same disease. A similar approach has been adopted in multiple sclerosis, which has four internationally recognized subtype definitions, including relapsing-remitting, primary progressive, secondary progressive, and progressive relapsing, that have prognostic value. ⁴⁶  

Nevertheless, there were further differences that distinguish the two disorders. There was a significant difference in mean age between the two groups (although this hides the wide range in age of MFCPU patients) and the nature of associated complications experienced by patients. However, these differences may, again, simply reflect the manner in which a patient's age at onset and degree of associated inflammation influence the manifestation of disease. Since autoimmunity occurs more commonly in the elderly, ⁴⁷ they may be more likely, on average, to develop a greater degree of clinically visible inflammation. Furthermore, cataracts are more common in the elderly, patients with clinically overt intraocular inflammation, and patients on topical and/or systemic immunosuppression, which are all characteristic of MFCPU, whereas CNV complicates low-grade subclinical inflammation, as in PIC. Although CNV is traditionally regarded as a complication resulting from previous attacks of choroiditis that create breaks in Bruch's membrane, there is compelling evidence currently that CNV pathogenesis is a result of dysregulated immunity that leads to a para-inflammatory state in the local microenvironment and continued low-grade tissue dysfunction. ^48,49 For example, both monogenic disorders, like Sorsby's fundus dystrophy, ⁵⁰ or polygenic multifactorial ocular diseases, like AMD, ⁵¹ demonstrate genetic associations and defects that lead to dysregulated innate immunity and a para-inflammatory state. As a result of proinflammatory cytokine release, including the pivotal cytokine in retinal inflammation, TNFα, VEGF is upregulated through a pathway involving transcription factor, SP1, thereby linking inflammation with angiogenesis. ⁵² In addition, the ability of macrophages, a predominant feature of CNV, to liberate VEGF is dependent on their activation signals from IL10, TNF-dependent interferon-γ production and hypoxia. ⁵³ Hence, single or combined therapy with systemic immunosuppression, intravitreal triamcinolone, PDT with or without anti-VEGF agents, is effective in the treatment of inflammatory CNV. ^{54 –57} Previous studies have reported CNV in up to 69% of patients with PIC. ^{32,58 –60} We found that 86.7% of our PIC patients developed CNV, which is likely to reflect the referral pattern of patients to tertiary centers for PDT and other therapies on a clinical trial basis. 

We have found that patients with PIC and MFCPU have a high prevalence of AI disease in their personal (13%) and family histories (21%), suggesting a genetic predisposition to AI disease, in general. ³⁹ Furthermore, the white-dot syndromes were associated with an IL10 haplotype, IL10htSNP2(−2849)AX/htSNP5(+434)TC, which includes two htSNPs in the IL10 gene that, in a previous study, were associated with noninfectious uveitis, overall. ³⁹ Similarly, the LTA+252A/TNFhtSNP1(−308)G/TNFhtSNP2(−238)G/TNFhtSNP3(+488) A/TNFd3 haplotype was negatively associated with disease. It is possible that the number of patients included in these analyses was not large enough to detect a significant association with IL10htSNP6, or that this SNP has less influence on the pathogenesis of white-dot syndromes compared with noninfectious uveitis, in general. Yet, the IL10htSNP5 locus is associated with MFCPU, in particular, the combined white-dot syndrome group, and a larger cohort of patients with noninfectious uveitis, ³⁹ and this association remains through all allele, genotype, and haplotype analyses. Furthermore, two large genome-wide association studies (GWASs) have recently identified susceptibility loci for another uveitic syndrome, Behçet's disease, within IL10. The first identified rs1554286 in intron 3 (linked to IL10htSNP6) in a Japanese cohort, and rs1800871 in the IL10 promoter (IL10htSNP4) in Turkish and Korean patients, ⁶¹ whereas the second identified rs1518111 in intron 2 (also linked to IL10htSNP6) in patients from the Middle East, Europe, and Asia. ⁶² IL10htSNP5 and IL10htSNP6 are both located in intron 1 of IL10 and are separated by only 70 bp. Hence, the evidence from GWASs and our own work suggests a susceptibility locus for AI disease in intron 1 of the IL10 gene—IL10htSNP5 and/or IL10htSNP6—although the possibility remains that the functionally relevant SNP is in linkage disequilibrium with them both. Of further interest, we found that there were no significant differences between our MFCPU and PIC patients at any IL10 or TNF locus investigated in this study, suggesting that these patients had more in common with each other, at a genetic level, than with our control cohort. Hence, we cannot distinguish MFCPU from PIC, based on genotype at these loci. The relevance of the IL10 and TNF polymorphisms to noninfectious uveitis is particularly salient in the context of current concepts of immunoregulation. Defects and deficiencies in IL10-secreting Foxp3⁺ Tregulatory cells (Tregs) have been identified in various forms of noninfectious uveitis with unrestrained activity of proinflammatory cytokines, particularly TNFα, ⁶³ and it is interesting to note that anti-TNFα therapies promote restoration of the number of Treg cells and function together with resolution of the inflammatory disease. ⁶⁴ It is a short step to envisioning a scenario in which polymorphisms in the IL10 gene directly affect Treg function in susceptible individuals, thus promoting disease. 

Previous studies have shown that HLA-B7 and HLA-DR2 are more frequent in patients with POHS. ^33,34,36 In this study, there was no association between HLA-B7 and MFCPU or PIC, although there was an increased frequency of the HLA-DRB1*15 allele (HLA-DR2) in PIC patients (P _uncorr= 0.02). Hence, although polymorphisms of the IL10 and TNF genomic regions may determine susceptibility to autoimmunity in general, there may be differences between MFCPU and PIC patients at other loci to account for differences in the degree of associated inflammation and clinical course between the two syndromes. This question can only be addressed by a much larger study to investigate differences between patient groups across the whole genome, by direct sequencing or DNA microarray. Furthermore, the data also support the notion that epigenetic factors may have an effect. 

In summary, we have found that patients with MFCPU and PIC demonstrate clear differences in their clinical course that have prognostic value and that this should impact on their clinical management. Although we could distinguish MFCPU from PIC patients based on their clinical course, we did not find any differences in genotype frequencies between the two groups at loci within the TNF and IL10 genomic regions and did in fact find that they are associated with identical IL10 and TNF haplotypes. Based on these analyses, MFCPU and PIC may well represent two manifestations of the same disease. In addition, our data and results from other studies suggest a susceptibility locus for AI disease in intron 1 of the IL10 gene, and this work should direct further functional investigation of the transcriptional regulation of IL10 expression in health and disease. 

Footnotes

 Supported by the National Eye Research Centre and the Medical Research Committee of the Charitable Trusts for the United Bristol Hospitals.

Footnotes

 Disclosure: D. Atan, None; S. Fraser-Bell, None; J. Plskova, None; L. Kuffová, None; A. Hogan, None; A. Tufail, None; D.J. Kilmartin, None; J.V. Forrester, None; J.L. Bidwell, None; A.D. Dick, None; A.J. Churchill, None

The authors thank the Immunology and Immunogenetics Department at Southmead Hospital, North Bristol Health Care Trust for HLA typing and all the patients and control subjects for their participation.