Genetics of Inflammatory Bowel Diseases

Abstract

In this Review, we provide an update on genome-wide association studies (GWAS) in inflammatory bowel disease (IBD). In addition, we summarize progress in defining the functional consequences of associated alleles for coding and non-coding genetic variation. In the small minority of loci where major association signals correspond to non-synonymous variation, we summarize studies defining their functional effects and implications for therapeutic targeting. Importantly, the large majority of GWAS-associated loci involve non-coding variation, many of which modulate levels of gene expression. Recent expression quantitative trait loci (eQTL) studies have established that expression of the large majority of human genes is regulated by non-coding genetic variation. Significant advances in defining the epigenetic landscape have demonstrated that IBD GWAS signals are highly enriched within cell-specific active enhancer marks. Studies in European ancestry populations have dominated the landscape of IBD genetics studies, but increasingly, studies in Asian and African-American populations are being reported. Common variation accounts for only a modest fraction of the predicted heritability and the role of rare genetic variation of higher effects (i.e. odds ratios markedly deviating from one) is increasingly being identified through sequencing efforts. These sequencing studies have been particularly productive in very-early onset, more severe cases. A major challenge in IBD genetics will be harnessing the vast array of genetic discovery for clinical utility, through emerging precision medicine initiatives. We discuss the rapidly evolving area of direct to consumer genetic testing, as well as the current utility of clinical exome sequencing, especially in very early onset, severe IBD cases. We summarize recent progress in the pharmacogenetics of IBD with respect of partitioning patient responses to anti-TNF and thiopurine therapies. Highly collaborative studies across research centers and across subspecialties and disciplines will be required to fully realize the promise of genetic discovery in IBD.

Genome-wide association studies (gwas) in IBD

The first GWAS using genome-wide SNP chips were published in 2005^{1, 2}. The conceptual basis of GWAS is that most complex (i.e., not single-gene Mendelian) genetic disorders are polygenic, being driven by multiple, common genetic polymorphisms. A key methodologic advance enabling GWAS was the development of genome-wide SNP chips that typed a relatively limited number of selected polymorphisms (less than 1 million SNPs for European ancestry individuals) that comprehensively assayed common genetic variation genome-wide. These SNP chips could efficiently test common human genetic variation in a highly accurate and reproducible manner. Furthermore, a variety of analytic approaches were developed (e.g. principal components analysis) which allowed for precise matching of cases and controls; this assured that allele frequency differences would be driven by the disease trait, and not subtle population differences between cases and controls.

The first GWAS in IBD identified many new loci, which were consistently replicated between different studies. Because of the large number of genetic markers and thus statistical tests performed, strict statistical thresholds (p-values less than 5 × 10^-8) are required for proof of genome-wide significance. The early GWAS identified the most significant loci; as the sample sizes increased through meta-analyses and inclusion of additional, independent cohorts³, the number of significant loci progressively increased. A major concept recognized early on was the extent to which general genomic regions of association overlapped between distinct chronic immune mediated diseases. This observation formed the basis for the development of the Immunochip⁴, a cost-effective custom chip with a dense map of markers clustered in 186 regions (i.e., fine-mapping regions) demonstrating genome-wide significant evidence for association in at least one autoimmune or chronic inflammatory disease. In addition to the fine-mapping loci, additional custom content, including genomic regions just missing genome-wide significance by GWAS was added. A meta-analysis combining GWAS with Immunochip data identified 163 loci associated with IBD. Of these loci, 110 confer risk to both IBD subtypes, whereas 30 and 23 loci were unique to Crohn's disease and ulcerative colitis, respectively. These 163 loci explain 13.6% of CD and 7.5% of UC total disease variance, respectively⁵. More recently a trans-ethnic analysis identified an additional 38 new IBD loci; a striking overlap of directionality of odds ratios is observed between European and Asian ancestry cohorts⁶. A summary of these loci are provided in Table 1.

Table 1. Summary of significant IBD, CD and UC loci identified by GWAS.

Chr	Position (Mb)	SNP	Source5, 6	Type	Candidate Genes
1	1.24	rs12103	Jostins et al.,	IBD	TNFRSF18,TNFRSF4,(30)
1	2.5	rs10797432	Jostins et al.,	UC	TNFRSF14,MMEL1,PLCH2,(8)
1	8.02	rs35675666	Jostins et al.,	IBD	TNFRSF9,(6)
1	20.15	rs6426833	Jostins et al.,	UC	(9)
1	22.7	rs12568930	Jostins et al.,	IBD	(3)
1	63.05	rs1748195	Liu et al.,	CD	USP1
1	67.68	rs11209026	Jostins et al.,	IBD	IL23R,IL12RB2,(4)
1	70.99	rs2651244	Jostins et al.,	IBD	(3)
1	78.62	rs17391694	Jostins et al.,	CD	(5)
1	92.55	rs34856868	Liu et al.,	IBD	BTBD8
1	101.47	rs11583043	Liu et al.,	UC	SLC30A, EDG1
1	114.3	rs6679677	Jostins et al.,	CD	PTPN22,DCLRE1B,(7)
1	120.45	rs3897478	Jostins et al.,	CD	ADAM30,(5)
1	151.79	rs4845604	Jostins et al.,	IBD	RORC,(14)
1	155.67	rs670523	Jostins et al.,	IBD	UBQLN4,RIT1,MSTO1,(28)
1	160.85	rs4656958	Jostins et al.,	IBD	CD48,SLAMF1,ITLN1,CD244,F11R,USF1,SLAMF 7,ARHGAP30,(8)
1	161.47	rs1801274	Jostins et al.,	IBD	FCGR2A,FCGR2B,FCGR3A,HSPA6,FCGR3B,FCRL A,(9)
1	169.52	rs6025	Liu et al.,	IBD	SELP, SELE, SELL
1	172.85	rs9286879	Jostins et al.,	CD	FASLG,TNFSF18,(0)
1	186.88	rs10798069	Liu et al.,	CD	PTGS2, PLA2G4A
1	197.6	rs2488389	Jostins et al.,	IBD	C1orf53,(2)
1	198.6	rs7555082	Liu et al.,	CD	PTPRC
1	200.09	rs2816958	Jostins et al.,	UC	(3)
1	200.87	rs7554511	Jostins et al.,	IBD	KIF21B,(6)
1	206.93	rs3024505	Jostins et al.,	IBD	IL10,IL20,IL19,IL24,PIGR,MAPKAPK2,FAIM3,RA SSF5,(3)
2	25.12	rs6545800	Jostins et al.,	IBD	ADCY3,(6)
2	27.63	rs1728918	Jostins et al.,	CD	UCN,(23)
2	28.61	rs925255	Jostins et al.,	IBD	FOSL2,BRE,(1)
2	43.81	rs10495903	Jostins et al.,	IBD	(5)
2	61.2	rs7608910	Jostins et al.,	IBD	REL,C2orf74,KIAA1841,AHSA2,(6)
2	62.55	rs10865331	Jostins et al.,	CD	(3)
2	65.67	rs6740462	Jostins et al.,	IBD	SPRED2,(1)
2	102.86	rs917997	Jostins et al.,	IBD	IL1R2,IL18RAP,IL18R1,IL1R1,IL1RL1,IL1RL2,(3)
2	145.49	rs11681525	Liu et al.,	CD	-
2	160.79	rs4664304	Liu et al.,	IBD	MARCH7, LY75, PLA2R1
2	163.1	rs2111485	Jostins et al.,	IBD	IFIH1,(5)
2	191.92	rs1517352	Jostins et al.,	IBD	STAT1,STAT4,(2)
2	198.65	rs1016883	Jostins et al.,	UC	RFTN2,PLCL1,(7)
2	199.7	rs17229285	Jostins et al.,	UC	0
2	204.59	rs3116494	Liu et al.,	UC	ICOS, CD28, CTLA4
2	219.14	rs2382817	Jostins et al.,	IBD	SLC11A1,CXCR2,CXCR1,PNKD,ARPC2,TMBIM1, CTDSP1,(8)
2	228.66	rs11178120 3	Liu et al.,	IBD	CCL20
2	231.09	rs6716753	Jostins et al.,	CD	SP140,(5)
2	234.145	rs12994997	Jostins et al.,	CD	ATG16L1, INPP5D, (7)
2	241.57	rs3749171	Jostins et al.,	IBD	GPR35,(12)
2	242.74	rs35320439	Liu et al.,	CD	PDCD1, ATG4B
3	18.76	rs4256159	Jostins et al.,	IBD	0
3	46.46	rs11301008 1	Liu et al.,	UC	FLJ78302, LTF, CCR1, CCR2, CCR3, CCR5
3	48.96	rs3197999	Jostins et al.,	IBD	MST1,PFKFB4,MST1R,UCN2,GPX1,IP6K2,BSN,IP 6K1,USP4,(56)
3	53.05	rs9847710	Jostins et al.,	UC	PRKCD,ITIH4,(8)
3	101.57	rs616597	Liu et al.,	UC	NFKBIZ
3	141.11	rs724016	Liu et al.,	CD	-
4	3.44	rs2073505	Liu et al.,	IBD	HGFAC
4	26.13	rs4692386	Liu et al.,	IBD	-
4	38.33	rs6856616	Liu et al.,	IBD	-
4	48.36	rs6837335	Jostins et al.,	CD	TXK,TEC,SLC10A4,(3)
4	74.85	rs2472649	Jostins et al.,	IBD	CXCL5,CXCL1,CXCL3,IL8,CXCL6,PF4,CXCL2,PF4V 1,(3)
4	102.86	rs13126505	Jostins et al.,	CD	(1)
4	103.51	rs3774959	Jostins et al.,	UC	NFKB1,MANBA,(2)
4	106.08	rs2189234	Liu et al.,	UC	-
4	123.22	rs7657746	Jostins et al.,	IBD	IL2,IL21,(2)
5	0.59	rs11739663	Jostins et al.,	UC	SLC9A3,(8)
5	10.69	rs2930047	Jostins et al.,	IBD	DAP,(2)
5	38.87	rs395157	Liu et al.,	IBD	OSMR, FYB, LIFR
5	40.38	rs11742570	Jostins et al.,	IBD	PTGER4,(1)
5	55.43	rs10065637	Jostins et al.,	CD	IL6ST,IL31RA,(1)
5	71.69	rs4703855	Liu et al.,	IBD	-
5	72.54	rs7702331	Jostins et al.,	CD	(4)
5	96.24	rs1363907	Jostins et al.,	IBD	ERAP2,ERAP1,LNPEP,(2)
5	130.005	rs4836519	Jostins et al.,	IBD	(1)
5	131.19	rs2188962	Jostins et al.,	IBD	IRF1,IL13,CSF2,SLC22A4,IL4,IL3,IL5,PDLIM4,SLC 22A5,ACSL6,(8)
5	134.44	rs254560	Jostins et al.,	UC	(6)
5	141.51	rs6863411	Jostins et al.,	IBD	SPRY4,NDFIP1,(5)
5	150.27	rs11741861	Jostins et al.,	IBD	TNIP1,IRGM,ZNF300P1,(8)
5	158.8	rs6871626	Jostins et al.,	IBD	IL12B,(3)
5	172.32	rs564349	Liu et al.,	IBD	C5orf4, DUSP1
5	173.34	rs17695092	Jostins et al.,	CD	CPEB4,(2)
5	176.79	rs12654812	Jostins et al.,	IBD	DOK3,(17)
6	0.38	rs7773324	Liu et al.,	CD	IRF4, DUSP22
6	3.42	rs13204048	Liu et al.,	CD	-
6	14.71	rs17119	Jostins et al.,	IBD	0
6	20.77	rs9358372	Jostins et al.,	IBD	(2)
6	21.42	rs12663356	Jostins et al.,	CD	(3)
6	31.27	rs9264942	Jostins et al.,	CD	HLA-C,PSORS1C1,NFKBIL1,MICB,(18)
6	32.595	rs6927022	Jostins et al.,	UC	HLA-DQB1,HLA-DRB1,HLA-DQA1,HLA-DRA,(12)
6	90.96	rs1847472	Jostins et al.,	IBD	(1)
6	106.43	rs6568421	Jostins et al.,	IBD	(2)
6	111.82	rs3851228	Jostins et al.,	IBD	TRAF3IP2,FYN,REV3L,(2)
6	127.45	rs9491697	Jostins et al.,	CD	(3)
6	128.24	rs13204742	Jostins et al.,	CD	(2)
6	138	rs6920220	Jostins et al.,	IBD	TNFAIP3,(1)
6	143.9	rs12199775	Jostins et al.,	IBD	PHACTR2,(5)
6	149.58	rs7758080	Liu et al.,	CD	MAP3K7IP2
6	159.49	rs212388	Jostins et al.,	CD	TAGAP,(5)
6	167.37	rs1819333	Jostins et al.,	IBD	CCR6,RPS6KA2,RNASET2,(3)
7	2.78	rs798502	Jostins et al.,	UC	CARD11,GNA12,TTYH3,(4)
7	17.44	rs1077773	Liu et al.,	UC	AHR
7	27.22	rs4722672	Jostins et al.,	UC	(14)
7	28.17	rs864745	Jostins et al.,	CD	CREB5,JAZF1,(1)
7	50.245	rs1456896	Jostins et al.,	IBD	ZPBP,IKZF1,(4)
7	98.75	rs9297145	Jostins et al.,	IBD	SMURF1,(6)
7	100.335	rs1734907	Jostins et al.,	IBD	EPO,(21)
7	107.45	rs4380874	Jostins et al.,	UC	DLD,(9)
7	116.89	rs38904	Jostins et al.,	IBD	(6)
7	128.57	rs4728142	Jostins et al.,	UC	IRF5,TNPO3,TSPAN33,(11)
7	148.22	rs2538470	Liu et al.,	IBD	CNTNAP2
7	26.88	rs10486483	Jostins et al.,	CD	(2)
8	27.23	rs17057051	Liu et al.,	IBD	PTK2B, TRIM35, EPHX2
8	49.13	rs7011507	Liu et al.,	UC	-
8	90.87	rs7015630	Jostins et al.,	CD	RIPK2,(4)
8	126.53	rs921720	Jostins et al.,	IBD	TRIB1,(1)
8	129.56	rs6651252	Jostins et al.,	CD	0
8	130.62	rs1991866	Jostins et al.,	IBD	(2)
9	4.98	rs10758669	Jostins et al.,	IBD	JAK2,(4)
9	93.92	rs4743820	Jostins et al.,	IBD	NFIL3,(2)
9	117.6	rs4246905	Jostins et al.,	IBD	TNFSF8,TNFSF15,TNC,(2)
9	139.32	rs10781499	Jostins et al.,	IBD	CARD9,PMPCA,SDCCAG3,INPP5E,(19)
10	6.08	rs12722515	Jostins et al.,	IBD	IL2RA,IL15RA,(6)
10	30.72	rs1042058?	Jostins et al.,	IBD	MAP3K8,(3)
10	35.295	rs11010067	Jostins et al.,	IBD	CREM,(3)
10	59.99	rs2790216	Jostins et al.,	IBD	CISD1,IPMK,(2)
10	64.51	rs10761659	Jostins et al.,	IBD	(3)
10	75.67	rs2227564	Jostins et al.,	IBD	(13)
10	81.03	rs1250546	Jostins et al.,	IBD	(5)
10	82.25	rs6586030	Jostins et al.,	IBD	TSPAN14,C10orf58,(4)
10	94.43	rs7911264	Jostins et al.,	IBD	(4)
10	101.28	rs4409764	Jostins et al.,	IBD	NKX2-3,(6)
10	104.23	rs3740415	Liu et al.,	IBD	NFKB2, TRIM8, TMEM180
11	1.87	rs907611	Jostins et al.,	IBD	TNNI2,LSP1,(17)
11	58.33	rs10896794	Jostins et al.,	IBD	CNTF,LPXN,(8)
11	60.77	rs11230563	Jostins et al.,	IBD	CD6,CD5,PTGDR2,(12)
11	61.56	rs4246215	Jostins et al.,	IBD	C11orf9,FADS1,FADS2,(12)
11	64.12	rs559928	Jostins et al.,	IBD	CCDC88B,RPS6KA4,TRPT1,FLRT1,(20)
11	65.65	rs2231884	Jostins et al.,	IBD	RELA,FOSL1,CTSW,SNX32,(22)
11	76.29	rs2155219	Jostins et al.,	IBD	(5)
11	96.02	rs483905	Jostins et al.,	UC	JRKL,MAML2,(2)
11	114.38	rs561722	Jostins et al.,	UC	FAM55A,FAM55D,(5)
11	118.74	rs630923	Jostins et al.,	IBD	CXCR5,(17)
12	6.49	rs7954567	Liu et al.,	CD	CD27, TNFRSF1A, LTBR
12	12.65	rs11612508	Jostins et al.,	IBD	LOH12CR1,(8)
12	40.77	rs11564258	Jostins et al.,	IBD	LRRK2,MUC19
12	48.2	rs11168249	Jostins et al.,	IBD	VDR,(8)
12	68.49	rs7134599	Jostins et al.,	IBD	IFNG,IL26,IL22,(1)
12	112.01	rs653178	Liu et al.,	IBD	SH2B3, ALDH2, ATXN2
12	120.15	rs11064881	Liu et al.,	IBD	PRKAB1
13	27.52	rs17085007	Jostins et al.,	IBD	(2)
13	40.86	rs941823	Jostins et al.,	IBD	(3)
13	43.02	rs9525625	Liu et al.,	CD	AKAP1, TFSF11
13	44.45	rs3764147	Jostins et al.,	CD	LACC1,(3)
13	99.95	rs9557195	Jostins et al.,	IBD	GPR183,GPR18,(6)
14	69.27	rs194749	Jostins et al.,	IBD	ZFP36L1,(4)
14	75.7	rs4899554	Jostins et al.,	IBD	FOS,MLH3,(6)
14	88.47	rs8005161	Jostins et al.,	IBD	GPR65,GALC,(1)
15	38.89	rs16967103	Jostins et al.,	CD	RASGRP1,SPRED1,(2)
15	41.55	rs28374715	Jostins et al.,	UC	ITPKA,NDUFAF1,NUSAP1,(8)
15	67.43	rs17293632	Jostins et al.,	IBD	SMAD3,(2)
15	91.17	rs7495132	Jostins et al.,	IBD	CRTC3,(3)
16	11.54	rs529866	Jostins et al.,	IBD	SOCS1,LITAF,RMI2,(10)
16	23.86	rs7404095	Jostins et al.,	IBD	PRKCB,(5)
16	28.595	rs26528	Jostins et al.,	IBD	RABEP2,IL27,EIF3C,SULT1A1,SULT1A2,NUPR1,(9)
16	30.47	rs11150589	Jostins et al.,	UC	ITGAL,(20)
16	50.66	rs2066847	Jostins et al.,	CD	NOD2,ADCY7,(5)
16	68.58	rs1728785	Jostins et al.,	UC	ZFP90,(6)
16	86	rs10521318	Jostins et al.,	IBD	IRF8,(4)
17	25.84	rs2945412	Jostins et al.,	CD	LGALS9,NOS2,(3)
17	32.59	rs3091316	Jostins et al.,	IBD	CCL13,CCL2,CCL11,(4)
17	37.91	rs12946510	Jostins et al.,	IBD	IKZF3,ZPBP2,GSDMB,ORMDL3,GSDMA,(12)
17	40.53	rs12942547	Jostins et al.,	IBD	STAT3,STAT5B,STAT5A,(13)
17	54.88	rs3853824	Liu et al.,	CD	-
17	57.96	rs1292053	Jostins et al.,	IBD	TUBD1,RPS6KB1,(9)
17	70.64	rs7210086	Jostins et al.,	UC	(3)
17	76.74	rs17736589	Liu et al.,	UC	-
18	12.8	rs1893217	Jostins et al.,	IBD	(6)
18	46.39	rs7240004	Jostins et al.,	IBD	SMAD7,(2)
18	56.88	rs9319943	Liu et al.,	CD	-
18	67.53	rs727088	Jostins et al.,	IBD	CD226,(2)
18	77.22	rs7236492	Liu et al.,	CD	NFATC1, TST
19	1.12	rs2024092	Jostins et al.,	CD	GPX4,HMHA1,(20)
19	10.49	rs11879191	Jostins et al.,	IBD	TYK2,PPAN-P2RY11,ICAM1,(25)
19	33.73	rs17694108	Jostins et al.,	IBD	CEBPG,(8)
19	46.85	rs4802307	Jostins et al.,	CD	(9)
19	47.12	rs1126510	Jostins et al.,	UC	CALM3,(14)
19	49.2	rs516246	Jostins et al.,	CD	DBP,SPHK2,IZUMO1,FUT2,(22)
19	55.38	rs11672983	Jostins et al.,	IBD	NLRP7,NLRP2,KIR2DL1,LILRB4,(15)
20	30.75	rs6142618	Jostins et al.,	IBD	HCK,(10)
20	31.37	rs4911259	Jostins et al.,	IBD	DNMT3B,(8)
20	33.8	rs6088765	Jostins et al.,	UC	PROCR,UQCC,CEP250,(8)
20	43.06	rs6017342	Jostins et al.,	UC	ADA,HNF4A,(9)
20	44.74	rs1569723	Jostins et al.,	IBD	CD40,MMP9,PLTP,(11)
20	48.95	rs913678	Jostins et al.,	IBD	CEBPB,(5)
20	57.82	rs259964	Jostins et al.,	IBD	ZNF831,CTSZ,(5)
20	62.34	rs6062504	Jostins et al.,	IBD	TNFRSF6B,LIME1,SLC2A4RG,ZGPAT,(23)
21	16.81	rs2823286	Jostins et al.,	IBD	0
21	34.77	rs2284553	Jostins et al.,	CD	IFNGR2,IFNAR1,IFNAR2,IL10RB,GART,TMEM50 B,(6)
21	40.46	rs2836878	Jostins et al.,	IBD	(3)
21	45.62	rs7282490	Jostins et al.,	IBD	ICOSLG,(9)
22	21.92	rs2266959	Jostins et al.,	IBD	MAPK1,YDJC,UBE2L3,RIMBP3,CCDC116,(8)
22	30.425	rs2412970	Jostins et al.,	IBD	LIF,OSM,MTMR3,(8)
22	39.69	rs2413583	Jostins et al.,	IBD	ATF4,TAB1,APOBEC3G,(16)
22	41.87	rs727563	Liu et al.,	CD	TEF, NHP2L1, PMM1, L3MBTL2, CHADL

Parentheses refer to the number of additional genes in the locus.

Functional consequences of missense alleles associated to IBD

Because of the highly correlated nature of human genetic polymorphisms, association signals often span broad genomic regions including multiple genes and transcribed sequences. Thus far, it is often been difficult to assign the likely “causal gene” driving the association signal. The presence of maximal association signals corresponding to missense alleles within a gene in the region, especially when linked to functional studies demonstrating altered gene function, provides strong support for that gene in contributing to disease pathogenesis. Importantly, these studies provide critical, often unexpected insight into disease mechanisms. For example, the finding that Crohn's disease associated missense alleles in the bacterial peptidoglycan sensing leucine rich repeat domain of NOD2^{7, 8} result in impaired activation of NF-κB^{8, 9} supported the general concept that deficiencies of innate immune cell function represent a central factor in Crohn's disease, distinguishing it from ulcerative colitis.

ATG16L1 and sequence annotation

The association of the Thr300Ala polymorphism within ATG16L1 to Crohn's disease in 2007 represents one of the key advances of the GWAS era^{10, 11}, propelling an intense interest in the role of autophagy in Crohn's disease. However, the functional impact of the alanine allele associated to Crohn's disease defied elaboration for many years following the initial genetic association. Recently, it was discovered that the alanine polymorphism introduces a consensus caspase 3 or 7 cleavage sequence, resulting in more efficient degradation of the Crohn's disease associated ATG16L1 allele. These results clearly establish that the CD risk allele is correlated with impaired autophagy. Therefore, active efforts to induce autophagy to treat a variety of human disease, including Crohn's disease, are ongoing¹². These studies also highlight the necessity of more broadly considering the impact of genetic variation on covalent modifications (e.g., gain or loss of glycosylation¹³, phosphorylation, sumolyation, as well as alterations which alter proteolytic susceptibility such as with ATG16L1) in defining the functional impact of nonsynonymous genetic variation^{14, 15}.

Uncommon protective, loss-of-function allele in IL23R

While there are multiple, independently acting risk alleles in the IL23R gene region, the most significant association is at Arg381Gln, where the minor glutamine allele (observed in approximately one out of seven European ancestry individuals) confers a two to three-fold protection against developing IBD¹⁶. This protective effect is greater in Crohn's disease than in ulcerative colitis, similar to many IBD-general loci⁵. Importantly, multiple groups have established that the protective glutamine allele is a loss-of-function allele^17-19, resulting in multiple functional alterations including decreased numbers of IL-23 dependent CD4+ Th17 and CD8+ Tc17 cells¹⁹. Thus far, no data that carriage of the protective glutamine allele incurs an increased risk of complications (e.g. infectious diseases) have thus far been reported. Taken together, this would indicate that decreasing IL-23 signaling, such as through monoclonal antibody blockade of anti-p40²⁰ or anti-p19²¹ (the two subunits of the IL-23 cytokine) may be beneficial in the treatment of IBD. Because IL23R is specific to IL-23 signaling, and not shared with the related IL-12 pathway, it is theoretically possible that selected targeting of the IL-23 pathway through blocking p19 (IL-23 specific) may either be more effective or be associated with fewer side effects than anti-p40 blockade, which blocks both IL-12 and IL-23 signaling.

Crohn's disease associated FUT2 null alleles associated with non-secretor status

Approximately 20% of European ancestry cohorts are homozygous carriers for W134X in FUT2, and this allele represents a Crohn's disease-specific association²². FUT2 encodes for a fucosyltransferase 2 enzyme which creates a soluble precursor for the ABO blood group antigens²³. Homozygous carriers of the FUT2 null allele do not express ABO antigens in saliva. Taken together, these studies highlight the potential importance of the mucus layer in Crohn's disease.

Rare missense alleles in PRDM1 and CARD9

Targeted sequencing within GWAS signals to identify rare missense alleles has proved quite productive in identifying likely causal alleles with functional effects. Rare mutations in PRDM1 (PR domain containing 1, with ZNF domain) have been identified, which increase T cell proliferation and cytokine secretion with activation²⁴. In a separate study, resequencing of CARD9 identified a rare splice variant which is highly protective (odds ratio of 0.3) and results in skipping of exon 11 and premature truncation close to the final splice junction with exon 12. If this rare protective splice variant results in a loss-of-function, this would be consistent with the finding that the common IBD risk allele at CARD9 is associated with higher gene expression²⁵. Taken together, this would indicate that, analogous to IL23R, CARD9 loss-of-function alleles are protective against IBD, and so therefore approaches to block CARD9 to treat IBD may be effective.

Fine-Mapping Non-Coding Genetic Variation: Integration With Expression and Epigenetic Data Implicates Cell Subtypes In Disease Pathogenesis

It has long been understood that disease-causing alleles are highly enriched for nonsynonymous, coding region variation that modulates protein structure and/or function²⁶. However, for 80-90% of GWAS-identified loci, the maximal association signals do not include a nonsynonymous variant, instead being confined to non-coding variation. Presumably, much of this non-coding variation exerts its pathogenic effects through modulation of gene expression. Expression quantitative trait loci (eQTL) involve DNA polymorphisms which are correlated with inter-individual variation in gene expression (typically mRNA) and early studies reported a marked enrichment of eQTLs overlapping GWAS loci. A particular value to such eQTL-GWAS mappings is that directionality of effect (disease-associated risk allele associated with increased or decreased expression) may be inferred. With the reporting of many more eQTL studies, including more well-powered and highly context specific studies (e.g. time-course expression analysis with cellular stimulation) it is now understood that the vast majority of genes, demonstrate eQTLs under certain conditions²⁷. This underscores the ubiquity of eQTLs, their highly context-specific nature and the potential value of studies focused on maximally relevant contexts, such as directly examining relevant disease tissues, as opposed to a reliance on in vitro systems. In this regard, transcriptome analyses of small mRNA amounts isolated from select immune cell subsets from primary intestinal tissues provides insight into differential gene expression for associated loci²⁸.

A major recent advance has been the demonstration of significant enrichment of GWAS signals with various epigenetic marks, including cell specific active enhancer (e.g. H3K27Ac) or promoter (e.g. H3K4me1) marks as defined by ChIP-Seq (chromatin immunoprecipitation sequencing). Importantly, recent studies have demonstrated enrichment of cell-specific H3K27Ac marks in patterns consistent with present understanding of disease mechanisms. For example, enrichment of precisely-mapped GWAS signals in cardiac disease, autoimmune and metabolic diseases has been observed within heart tissue, immune and liver cell enhancer marks, respectively²⁹. Within IBD, the greatest enrichment for cell specific enhancers has been observed within CD4+ T cell subsets, with lesser enrichment observed for CD8+, B cells and monocytes³⁰. Integrating epigenetic, intestinal expression and eQTL data provides improved capacity to predict likely IBD genes compare to gene association results alone³¹. Of note, while Crohn's disease demonstrated the greatest enrichment for immune cell epigenetic marks, ulcerative colitis GWAS signals were enriched within both immune cell and colonic mucosal marks³⁰. While ChIP-seq studies of colonic mucosa would involve a highly heterogeneous mixture of cells³⁰, a separate study demonstrated enrichment of ulcerative colitis GWAS signals within H3K27Ac marks in intestinal enteroids³², thereby implicating a primary pathogenic role for altered epithelial function in IBD.

Functional analyses of GWAS signals: variation in pattern recognition receptor signaling

The NOD2 associations to Crohn's disease highlight the importance of host genetics and inter-individual variability in responses to innate microbial stimulation, or pattern recognition receptor (PRR) signaling. Beyond the NOD2 associations which are associated with decreased cytokine secretion with PRR-stimulation^{9, 33}, inter-individual variability in host cytokine secretion from PRR-stimulated macrophages has been reported to a variety of IBD-associated alleles. PRR-initiated cytokine secretion was diminished in ICOSLG (ICOS ligand) risk allele carriers, similar to NOD2, and identified a novel role for ICOSLG signaling in innate immune cells³⁴. In contrast, IBD-associated polymorphisms in IRF5 (interferon regulatory factor 5) are a major determinant of inter-individual variability in PRR-stimulated cytokine secretion, with IRF5 risk alleles associated with increased cytokine secretion³⁵.

Other pathogenic mechanisms of non-coding variation

A functional methylome map of ulcerative colitis has been reported whereby intestinal biopsies from 10 monozygotic twin pairs (20 individuals) were analyzed for variable DNA methylation patterns. Altogether, 61 genomic regions with differential methylation patterns were identified that were also associated with nearby differentially expressed transcripts³⁶. These regions were not identified by GWAS and therefore implicate development or environmental factors regulating gene expression in contributing to UC pathogenesis. Finally, miRNAs are small, non-coding RNAs, 18-23 nucleotides in length which can broadly regulate gene expression³⁷. A number of studies have implicated specific miRNAs in regulating a broad array of IBD-associated genes, including NOD2, ATG16L1³⁸ and IL-23³⁹.

Population Differences in IBD

Initially thought to be a disease that affects primarily populations of European ancestry, it is increasingly evident that IBD involves almost all racial and ethnic groups⁴⁰. Our current understanding of IBD genetics is based on studies done primarily in populations of European ancestry, but increasingly, studies in non-European populations are being reported. Some studies have shown that disease markers identified in European ancestry populations may also be observed in admixed populations. For example, the frameshift NOD2 risk variant found in European populations are also associated with CD in African Americans, albeit at much lower allele frequencies, consistent with the recent European admixture present in African Americans^{41, 42}

IBD genetic studies in Asian populations

The first GWAS published in IBD was in a Japanese population, with an index cohort of only 94 cases, identifying TNFSF15 as a susceptibility gene for IBD². The ‘dominant’ CD TNFSF15 association in East Asians has been confirmed^{43, 44}, and associations, although more modest, confirmed in European ancestry populations suggesting that the genetic architecture of IBD would be similar in Asians and Europeans^{45, 46}. However, a number of studies have confirmed no role for NOD2 or ATG16L1 variants in East Asian CD suggesting that some susceptibility genes are shared across populations while others appear to have an effect in one population only^47-49. Subsequent studies have confirmed this phenomenon with a Japanese GWAS identifying 4 novel non- HLA associations with UC including variation at the FCGR2A⁵⁰, a finding subsequently confirmed in a European ancestry UC meta-analysis⁵¹. More recent Korean and Japanese CD GWAS identified a number of susceptibility regions not implicated in European populations including a CD association with ATG16L2 implicating a role for autophagy in CD in this population despite the lack of association from either NOD2 or ATG16L1^{52, 53}. Utilizing the Immunochip a study from Korea recently identified 6 additional loci CD loci shared with European ancestry individuals including NkX2-3, PTPN2, and ZNF365 bringing the total number of CD loci identified in Koreans to 15⁵⁴.

IBD genetic studies in Ashkenazi Jewish populations

A major goal of genetic studies in IBD is to account for the significantly higher disease prevalence observed in Ashkenazi Jewish compared to non-Jewish European ancestry cohorts. For common variants, the directionality of effects is almost exclusively in the same direction between these two cohorts, indicating a largely similar genetic architecture⁵⁵. Where Jewish populations may provide particularly important pathogenic insight may be in the interrogation of rare risk alleles that may be relatively population-specific.

Common and Rare IBD Genetic Variation: Sequencing Studies in IBD

Variants that affect protein function are more likely to have greater penetrance and may be more amenable to functional studies. Because natural selection acts against variants, such as Mendelian, or monogenic mutations, that confer large effects on disease susceptibility, such mutations are usually of low to rare frequency (Figure 1). However, rare mutations of higher effects (i.e., high odds ratios) are not only predicted to outnumber common variants in the human genome⁵⁶, but could also explain a substantial fraction of common complex diseases⁵⁷. Sequencing is the method of choice to comprehensively ascertain low frequency to rare variants that have effect sizes higher than those detected by GWAS.

Figure 1. Common and rare genetic variation.

It is estimated that about 85% of mutations with large effect size reside in the 1 to 2% portion of the human that represents the entire protein coding sequence known as the exome⁵⁸. Currently, sequencing of the whole exome has not only become a practical method but also a cost effective option to identify functionally relevant variants in the protein coding regions of the genome. Whole exome sequencing (WES) has emerged as a hypothesis free and unbiased way of surveying the entire genome for variants (known or novel) that may be at lower frequency and not well covered for detection by GWAS in IBD case-control studies. The first application of WES in a very early onset IBD case revealed a rare mutation affecting the regulatory function of the XIAP (X-linked inhibitor of apoptosis) gene in a boy who presented at 15 months with intractable IBD⁵⁹. Importantly, XIAP is a positive regulator of NOD2 function⁶⁰, and so therefore, loss-of-function mutations in XIAP would result in similar functional effects as those observed with NOD2 mutations. Subsequent successful application of WES has identified more rare functional variants in novel genes implicated in the pathogenesis of VEO (FOXP3⁶¹, IL10RB⁶²;⁶³ and adult onset (GSDMB⁶⁴ and NDP52 genes⁶⁵) IBD.

Genetics of Early onset Cases of IBD

While childhood onset IBD represents only 10-25% of all IBD cases, genetic research of pediatric IBD has contributed new knowledge and revealed unsuspected pathways. A substantial proportion of patients with monogenic diseases present with very early onset intestinal inflammation (at less than 10 years of age), that is reminiscent of very early onset IBD. There is also considerable overlap with primary immunodeficiencies and very early onset IBD (Table 2), a topic which has been reviewed recently⁶⁶.

Table 2. Genes associated with disorders where early onset IBD is well recognized. Broader function of the genes / pathways are indicated in the first column.

Function	Gene	Associated conditions
Epthelial barrierand epithelial response defects	COL7A1	Dystrophic epidermolysis bullosa
	FERMT1	Kindler syndrome
	IKBKG	X linked ectodermal dysplasia and immunodeficiency
	ADAM17	ADAM-17 deficiency
	GUCY2C	Familial diarrhoea
Neutropenia and defects in phagocyte bacterial killing	CYBB, CYBA, NCF1, NCF2, NCF4	Chronic granulomatous disease
	SLC37A4	Glycogen storage disease type 1b
	G6PC3	Congenital neutropenia
	ITGB2	Leucocyte adhesion deficiency 1
Hyper – and autoinflammation	MVK	Mevalonate kinase deficiency
	PLCG2	Phospholipase Cγ2 defects
	MEFV	Familial Mediterranean fever
	STXBP2	Familial haemophagocytic lymphohistiocytosis type 5
	XIAP	X linked lymphoproliferative syndrome 2
	SH2D1A	X linked lymphoproliferative syndrome 1
	HPS1, HPS4, HPS 6	Hermansky–Pudlak syndrome
	ICOS	Common variable immunodeficiency type 1
	LRBA	Common variable immunodeficiency type 8
	BTK, PIK3R1	Agammaglobulinaemia
	CD40LG, AICDA	Hyper-IgM syndrome
	WAS	Wiskott–Aldrich syndrome
	DCLRE 1C	Omenn syndrome
	DOCK8	Hyper IgE syndrome
	SKIV2L, TTC37	Trichohepatoenteric syndrome
	PTEN	PTEN hamartoma tumour syndrome
Regulatory T cells and immune regulation	FOXP3, IL2RA	X linked immune dysregulation, polyendocrinopathy, enteropathy
	IL10RA, IL10RB, IL10	IL-10 signalling defects
Defects in intestinal innervation	RET	Hirschsprung's disease

Modified from Uhlig⁶⁶.

The definition of childhood or early onset IBD can be arbitrary because the age at which childhood ends and adulthood begins represents a continuum. The exact determinants of the age of onset and disease course remain largely unexplained. Except for NOD2, common alleles of large effect size have not been reported for IBD, and additional larger GWAS are highly unlikely to reveal more common variants with larger effect. Interestingly, increasing genetic burden was associated with earlier age of onset for CD, but not for UC⁶⁷. Importantly, most GWAS studies involved mainly adult or adolescent onset IBD cases, and very early onset (VEO) cases were relatively uncommon. The VEO group experiences a more severe disease course and more frequently shows a positive family history for IBD, in support of higher genetic load ⁶⁸;⁶⁹;⁷⁰. This also suggests that VEO cases are likely to display Mendelian-like forms of IBD characterized by highly penetrant variants of higher effect size as shown by mutations identified in the IL10R gene⁶². It is necessary to study VEO populations to ascertain low frequency variations (major allele frequency of 0.5 to 1%) to expand our understanding of the contribution of genetics to IBD. Children constitute a population with distinct physiology and disease risks, such that studying VEO-IBD offer the opportunity to dissect the initial immune response and the early gut microbiome as well as changes over the time, the natural history of the disease, and the impact of early environmental modifiers. In short, the pediatric population represents a relatively unexplored and fertile source of novel and helpful information that promise to elucidate the triggers and pathogenic mechanisms of IBD.

Association between Monogenetic Disorders and Very Early Onset IBD

Inflammatory bowel disease-undetermined (IBD-U) is a class of IBD seen at higher rate in young children (34% in children under 2 years and 21% in children under 7 years) as opposed to adults (6%)⁷¹. The distinct and more severe disease phenotypes seen in VEO cases and often classified as IBD-U⁷¹ include manifestations of known monogenic diseases. Recently, a Mendelian form of VEO IBD was confirmed through a mutation in the IL10R gene⁶² that underlined an association of IBD with primary immunodeficiency. Many monogenic disease genes have been shown to confer overlapping pathology with IBD (known as IBD-like pathogenesis) and are seen more frequently in VEO cases ⁶⁶. These diseases represent potential targets for identifying additional VEO heritability using exome sequencing. A list of genes underlying known monogenic conditions associated with IBD is detailed in Table 1. This confirms previous reports that single gene disorders also predispose to complex disorders⁷²;⁷³ and suggest that many Mendelian and complex disorders could share a similar genetic architecture, where Mendelian loci may contain common variants with low effect size (detectable by GWAS) characterized by incomplete penetrance. Such common variants are capable of modifier functions and likely contribute to complex diseases alongside Mendelian, high effect size variants, detectable mainly by sequencing. Importantly, the utility of targeted gene panel sequencing in therapy refractory VEOIBD in influencing clinical decision-making has been recently advocated⁷⁴.

Clinical Applications and Genetic Testing

Of particular relevance to IBD genetics has been the launching of major initiatives in precision medicine involving prevention and treatment strategies that take individual variability into account ⁷⁵. Inflammatory bowel diseases, because of their heterogeneity with respect to natural history and response to therapy, are, arguably, the prototype immune-mediated disease for incorporating genetic parameters into everyday clinical practice. This, coupled with, the very significant advances in understanding the genetic architecture of IBD, present wonderful opportunities for improved care. However, significant challenges still remain. The history of discovery in genetics has taught researchers the value of a collegial approach to research allowing the generation of large cohorts. These lessons should be heeded as we look to additional discovery with respect to genetic influence on clinical phenotypes and pharmacogenetics.

Genetic testing for diagnosis in common forms of IBD

In retrospect, a naïve, hope for advances in IBD genetics was the potential role of variants to contribute to diagnosing IBD. Early studies suggested however that, as a consequence of the low pre-test probability (i.e. background prevalence of IBD) together with moderate genotype-relative risks of IBD-associated loci, the utility of genetic testing for a diagnosis of IBD is poor ⁷⁶. This remained the case for a composite genetic and smoking status score ⁷⁶. Another study suggested that a combination of genetic, serological and inflammatory markers was superior to a serological panel alone for discriminating non-IBD (a combination of healthy, IBS, and inflammatory ‘controls’) from IBD ⁷⁷. The AUC for IBD increased to 0.87 from 0.80 for the combination score versus the serological only panel although its unclear how much the addition of genetic data from 4 IBD-associated loci (ATG16L1, NKX2-3, ECM1, STAT3) contributed (ECM1 and STAT3 where not independently associated with IBD in this cohort). The study omitted any mention of study cohort ethnicity and any correction for ethnicity or population structure in the analyses. Despite the limitations suggested by such studies, consumers now have the ability, through direct to consumer (DTC) genetic services, to obtain their own data from genetic markers associated with more than ‘254 specific diseases and conditions’ ⁷⁸. The difficulty of using individual SNP data to ‘predict’ risk of disease was emphasized by one study investigating disease risk assessment by two different DTC genetic testing products ⁷⁹ in which the direction of risk for Crohn's disease was discordant for 3 of 5 individuals. In fact, for seven of the diseases tested, ‘50% or less of the predictions agreed across the 5 individuals.’ The authors of the study recommended communicating better about risks distributed with the results, observed that markers should be assessed in all ethnicities, and noted that formal studies should be performed to see if receiving genetic information alters behavior. In fact, in a well-chronicled dispute, the Food and Drug Administration (FDA), suggesting that 23andMe™ had failed to show that it had ‘analytically or clinically validated their “Personal Genome Services” (PGS) for its intended uses,’ ordered the company to cease providing information on its disease and condition-associated markers to new consumers. This is likely the first skirmish in an ongoing debate about consent, access, education and other issues related to DTC genetic tests. Nevertheless, the consensus is that consumers will have increased access to their genotype/sequence data and that healthcare providers will need to develop skills and strategies for explaining the limitations and possible benefits of these data when confronted in the clinic.

Genetic testing in very early onset IBD

The genetic testing of common variants as a screening or diagnostic too has little clinical value in the general population. For example, common variants in NOD2 have the highest effect size in IBD, but are found in 10-20% of the general population and in less than 5% of positive cases who have or will develop CD. On the other hand, with monogenetic disorders, the finding a gene variants is almost always associated with a disease. To date, as many as 50 or more single genes causing IBD (mono-genic forms of IBD) are implicated in very early onset IBD, and number of genes is expected to increase. Traditional and individual single gene testing can be cumbersome and expensive for clinical use. Clinicians are often confused about what gene to test first, when they encounter a clinical problem such as very early onset IBD that is polygenic. Clinicians have been waiting for a reliable, fast, accurate and affordable gene testing solution for their young patients with very early onset IBD or IBD-like intestinal inflammation. Although gene panel testing is widely available for many diseases/disorders, commercially available IBD gene panel is offered by few organizations. Any commercial testing that is covered by insurance for patients living in the USA and certify by CLIA (Clinical Laboratory Improvement Amendments) and/or CAP (College of American Pathologists) is necessary. Such testing provides targeted full gene sequencing for the 50+ genes implicated with early onset IBD and provide comprehensive interpretation of the results. Some laboratories offer testing that is scalable to full exome genome sequencing in patients with initial initial results, however, full exome sequencing needs to be requested separately after the first round of results. The genetic testing for IBD not only establishes the molecular diagnosis with the basis of pathogenesis, but it also allows rationale for patient-specific early intervention with emerging or experimental therapeutics and cell based approaches and the opportunity to screen family members for carrier detection and genetic counseling. For a substantial proportion of IBD or IBD-like diseases, these monogenic disorders also overlap with immunodeficiency affecting granulocyte and phagocyte activity, hyper- and autoinflammatory disorders, defects with disturbed T and B lymphocyte selection and activation, and defects in immune regulation affecting regulatory T cell activity and interleukin (IL)-10 signaling. An area of IBD genetic research with significant progress prompting possible clinical use has been the genetics of very early onset IBD (VEOIBD). In a seminal study researchers from Europe and North America identified a homozygous interleukin 10 receptor (IL10R) variant encoding a premature stop codon in a young child with severe IBD resistant to conventional therapy ⁶². Functional studies suggested a hematopoietic source for the functional defect and a bone marrow transplant resolved the IBD. Further studies have confirmed the role of IL10R variants in VEOIBD, and such VEOIBD findings have prompted the use of whole exome sequencing as a clinical standard in VEOIBD ^{62, 63, 80-82}.

Pharmacogenetics

Predicting response to adverse events from IBD therapies has been a longtime goal of research efforts. The ability to identify individuals likely to respond to, or have adverse effects from a therapy may become increasingly important as the number of IBD treatment agents expand. The limited success of studying pharmacogenetics of anti-TNF therapies has been partially due to significant inter-individual variation in the pharmacokinetics of these agents suggesting that many ‘non-responders’ were likely under-dosed, confounding previous studies. Nevertheless, a number of studies have been reported, the largest suggesting that variants in apoptotic genes can be combined to form an ‘apoptotic pharmacogenetic index.’ The index, combined with clinical factors, helped construct an algorithm for predicting response to infliximab in CD⁸³. Other studies have utilized mucosal gene expression for predicting response to anti-TNF; one study used a combination of gene expression profiles from 5 innate immunity genes to generate a ‘score’ with 95% sensitivity and 85% specificity for response to infliximab in UC⁸⁴. A second study identified that a renal tissue gene signature associated with graft rejection after renal transplant was similar to that seen in the mucosa of UC patients unresponsive to anti-TNF therapy⁸⁵. These findings clearly require replication, but ready access to disease tissue may be a unique and not fully explored opportunity for IBD.

The ultimate pharmacogenetic approach may be the genetic identification of pathways and processes involved in disease pathogenesis and therefore pathways for therapeutic intervention either through novel development or for re-purposing of existing drugs. Determining whether this approach will be effective in IBD is premature, but a large study in rheumatoid arthritis (RA) suggested, as a proof of principle, that RA genetic associations tagged pathways targeted by existing RA drugs⁸⁶. Furthermore, this study identified novel pathways for potential therapeutic intervention and drugs, already in use in other areas, that may be beneficial if ‘re-purposed’ for RA. Astra-Zeneca published a retrospective review of their small-molecule drug projects from 2005 to 2010 and identified that 73% of phase II projects with ‘human genetic linkage of the target to the disease indication’ were successful compared to only 43% with no genetic ‘linkage.’ These results suggest that an understanding of the genetic architecture of a trait may help streamline the drug discovery process⁸⁷.

In addition, testing for thiopurine methyltransferase (TPMT) variation, prior to thiopurine initiation in order to reduce the risk of bone marrow toxicity, has influenced clinical practice and therapeutic choices in IBD. Two recent studies have significantly increased our understanding of genetic factors associated with thiopurine-related adverse events. Recognizing a higher prevalence of thiopurine induced leucopenia in Koreans despite lower TPMT variant frequency, Yang et al. performed an unbiased genome-wide study identifying a non-synonymous (basic arginine to polar cysteine at codon 139) NUDT15 polymorphism that is notably associated with both early and late leucopenia after thiopurine therapy. The effect size of this variant for leucopenia was greater than that for TPMT variants in Koreans and while this variant was rare in caucasians (< 1%), it was still associated with leucopenia in this population⁸⁸. In addition, a similarly designed study from Heap et al. identified HLA-DQA1-HLA-DRB1 variants associated with developing pancreatitis after thiopurine exposure in Europeans⁸⁹. These variants being added to TPMT status as constituents of a ‘thiopurine panel,’ enabling clinicians to reduce the risk of serious adverse events related to these useful compounds, is highly likely in the near future.

Prognostics

The heterogeneity seen in the natural history of the IBDs has prompted a number of studies investigating genetic associations with disease severity. These studies have largely been retrospective, underpowered, and lacking replication. Nevertheless, the studies have presented some interesting findings and a methodology that may influence future studies. An unbiased genome-wide approach identified a composite score of 46 SNPs that identified subjects with medically-refractory ulcerative colitis (MRUC)⁹⁰. These findings clearly require replication but show the potential utility of including a number of SNPs, each of moderate effect size, to develop a composite risk score. Similarly, a study looking at associations with first surgery in Crohn's disease combined genetic variants, serological markers, and clinical characteristics in a multi-modal scoring system⁹¹. A variant that remained significant in all models tagged IL12B, a locus, although at a different SNP, that was previously identified with disease severity in CD⁹². An intriguing recent study suggested that a noncoding FOXO3A SNP was associated with a milder clinical natural history in both Crohn's disease and rheumatoid arthritis but also an increased risk of severe malaria ⁹³. However, this finding was not replicated in a subsequent, independent study⁹⁴. Functional studies demonstrated that altered inflammatory responses in monocytes with decreased pro-inflammatory and increased anti-inflammatory cytokine production are associated with this variant. The same group also confirmed that a gene expression signature in CD8 positive T cells, previously associated with more severe disease in lupus, identified both UC and CD subjects with more severe disease ⁹⁵. These studies, also in need of replication, extend the pleiotropic observations of the effect of genetic variation from disease susceptibility to association with disease behavior. Prospective inception cohorts currently being followed⁹⁶ will be invaluable in confirming these and other findings.

Extra-intestinal manifestations

The study of genetic associations with extra-intestinal manifestations (EIMs) has, on the whole, been hampered by underpowered studies, with the possible exception of the study of primary sclerosing cholangitis (PSC). A number of PSC GWAS and large-scale genotyping studies have been performed with the latest immunochip-based study of nearly 30,000 subjects extending the number of susceptibility loci to 16^97-99. A further analysis of these data incorporating associations from immune-mediated diseases associated with PSC suggested additional loci totaling 30 non-HLA associations. Approximately half of these PSC loci are known IBD ‘genes’ and PSC ‘genetically’ is more similar to UC than CD with an AUC of 0.624 and 0.559 for PSC from UC and CD risk alleles respectively⁹⁷. A functional network analysis of where PSC genes fall in an IBD network disappointingly revealed that the PSC genes were scattered around the network and not localized to one part, which suggests that PSC is not associated with a molecular subset of IBD ⁹⁷. This may simply reflect limitations of our understanding of the full ‘molecular’ PSC story or limitations in current analytic approaches. Similar to IBD, only a minority of variance is explained by these genetic variants, prompting investigators to examine the role of the environment, including the microbiome in PSC pathogenesis¹⁰⁰. An intriguing study identified an association between fucosyltransferase 2 (FUT2) genotype and episodes of cholangitis, fungobilia (biliary Candida infections), and dominant stenosis, suggesting that FUT2 may be a marker of ‘host microbial diversity’ and ‘disease progression’ in PSC. These results support other studies suggesting that FUT2 variants ‘garden’ the microbiome in both healthy and IBD populations^101-103.

There has been considerable interest in genetic pleiotropy among the immune-mediated diseases, and IBD ‘shares’ more susceptibility loci with spondyloarthropathy (SpA) than any other trait¹⁰⁴. Whether SpA is an EIM of IBD or IBD is an extra-articular manifestation of SpA, or whether they are in fact conditions with similar molecular architecture with different end-organ manifestations remains to be seen^{105, 106}. However, the vast majority of shared susceptibility loci are concordant between IBD and SpA, which is in contrast to other traits, including psoriasis in which approximately one quarter of shared loci show discordance. The relationship between psoriasis and IBD is complex. Recent observations that anti-TNF agents can cause psoriasiform lesions in IBD patients, that this phenomenon may be associated with IL23R variants and respond to anti-IL12/23 antibody therapy^{107, 108} and that anti-IL17a therapy, so successful in psoriasis, appears to worsen Crohn's disease, only adds to the complexity¹⁰⁹. The finding that a subset of CD patients tagged by a TNFSF15 variant improved with anti-IL17a therapy (secukinumab) adds to the complexity and surely makes the case for inclusion of genotyping in clinical trials of novel therapeutic agents¹⁰⁹. These observations require further study as they surely reveal insights to underlying mechanisms leading to both IBD and psoriasis.

There have been few ‘within-IBD’ studies searching for genetic variants associated with EIMs, and these studies have largely been underpowered and lacked the depth of genotyping that is now standard in genetic studies. Historical studies identified association with HLA variants and the development of eye and skin manifestations as well as both peripheral and axial arthropathy in IBD^{110, 111}. More recently, an unbiased approach identified interesting genetic associations with pyoderma gangrenosum (including IL8RA, TIMP3). However, no associations achieved genome-wide significance, unsurprising given the small sample size¹¹². This study, in keeping with previous observations, confirmed the co-segregation of the EIMs in IBD, suggesting shared genetic mechanisms. Studies delineating the molecular causes of the EIMs will clearly require collaboration across multiple-sites in order to generate appropriately sized study cohorts. Studying these under-researched traits is a challenge that the international community should embrace, as identification of novel areas for therapeutic targeting for EIMs, which can often be as disabling as the underlying IBD, would potentially have utility beyond IBD.

Future Directions

The advances in our understanding of the genetic variation underlying IBD, particularly in European ancestry individuals, has been spectacular and has provided an outstanding framework for basic and translational research. These advances have also generated significant challenges and posed many additional questions. From an ethical and scientific perspective it is imperative that the resources applied to generating these datasets be applied to non-European ancestry populations. Large-scale genotyping studies are designed to identify loci and particular those tagged by common variants. A natural extension of these studies is large-scale fine-mapping approaches to ‘narrow’ associated regions and to identify ‘causative’ variants thereby greatly facilitating functional studies. Many of these variants are likely to be in non-coding regions and understanding how these alter gene expression is an important challenge the research community will need to address. There have been some interesting clues from recent studies but it is increasingly apparent that these eQTL studies will also need to be performed in single-cell compartments relevant to disease pathogenesis. While this will help overcome some of the heterogeneity of cell types in standard tissue sampling there is an evolving story of heterogeneity even within cell types suggesting that this approach may need to be extended to the single-cell level. Ultimately, the purpose of this research is to improve the lives of people with IBD and therefore increasing migration of these advances to the clinic is imperative. This will likely come, over the next few years, in the arenas of biomarker and drug development for the wider population but more immediately for ‘diagnostics’ in the rare but very serious VEOIBD subjects where these data have the ability to transform clinical approaches and outcomes. There will likely be significant ‘push-back’ to the current attitudes to sharing genetic data with research subjects and also evolution of the regulation of the direct to consumer companies and it is likely that practicing clinicians will be faced with patients armed with their genotype data in the near future.