Association of C-Reactive Protein Genetic Polymorphisms With Late Age-Related Macular Degeneration

Key Points

Question  Are variants in the C-reactive protein (CRP) gene that are known to influence circulating C-reactive protein concentrations associated with late age-related macular degeneration?

Findings  This pooled analysis examined 4 CRP variants among 1727 patients with late age-related macular degeneration and 1153 controls from 2 hospital-based case-control studies and 1 cross-sectional population-based study. No statistical association was found between these CRP variants and any type of late age-related macular degeneration.

Meaning  Circulating C-reactive protein concentrations are unlikely to be causally associated with age-related macular degeneration.

Importance  C-reactive protein (CRP) is a circulating inflammatory marker associated with late age-related macular degeneration (AMD). It remains uncertain whether the association between CRP concentrations and AMD is causal.

Objective  To assess whether CRP (OMIM 123260) single-nucleotide polymorphisms that influence circulating CRP concentrations are associated with late AMD.

Design, Setting, and Participants  Participants in 2 UK, hospital-based, case-control studies (Cambridge AMD study and Moorfields Eye Hospital AMD study) and 1 pan-European, cross-sectional, population-based study (the European Eye [EUREYE] Study) were recruited between November 6, 2000, and April 30, 2007. Participants underwent dilated stereo-digital fundus photography graded according to the International Classification of Age-related Maculopathy and Macular Degeneration. There were 1727 cases of late AMD (1151 neovascular, 384 geographic atrophy, and 192 mixed [neovascular AMD and geographic atrophy]) and 1153 controls. Early AMD cases (n = 574) were included only from the EUREYE Study. Data analysis was performed from August 1 to November 30, 2016. Four common single-nucleotide polymorphisms (rs1205, rs1130864, rs1800947, and rs3093077) were selected based on demonstrated influence on circulating CRP concentrations in the literature. In one study, genotyping of rs3093077 failed, and rs1800947 was typed in only 1 study.

Main Outcomes and Measures  A genetic multiplicative model was used for the association of single-nucleotide polymorphisms with late AMD adjusted for age and sex.

Results  Among the 1727 patients with late AMD, the mean (SD) age was 78.7 (7.4) years, and 668 (38.7%) were men. The mean (SD) age of the controls was 74.9 (7.0) years, and 510 (44.2%) were men. In the pooled results of all 3 studies, neither rs1205 (odds ratio [OR], 0.99; 95% CI, 0.86-1.14) nor rs1130864 (OR, 0.96; 95% CI, 0.83-1.11) was associated with late AMD. For geographic atrophy, rs1205 had an OR of 0.91 (95% CI, 0.74-1.13) and rs1130864 had an OR of 0.94 (95% CI, 0.76-1.16). For neovascular AMD, rs1205 had an OR of 1.01 (95% CI, 0.87-1.19) and rs1130864 had an OR of 0.99 (95% CI, 0.84-1.16). There was no association of rs3093077 and rs1800947 with late AMD or any late AMD phenotype. There were no significant findings for early AMD.

Conclusions and Relevance  Our results do not support a causal association between CRP concentrations and AMD.

Inflammation has been recognized to play a key role in age-related macular degeneration (AMD),^1,2 strengthened by the discovery of multiple genetic risk loci involved in the complement pathway.^3-7 The early and intermediate stages of AMD are characterized by drusen, which are accumulations of lipids and proteins that form discrete deposits under the retinal pigment epithelium, and these accumulations can both initiate and reflect inflammatory reactions within the complement pathway.¹ Several studies have investigated the association between circulating inflammatory markers and the onset and progression of AMD,^8-12 both to understand the pathogenesis of AMD and to assess their potential as prognostic biomarkers. One such inflammatory marker is C-reactive protein (CRP), an acute phase reactant¹³ that has been shown in 2 meta-analyses to be associated with late AMD.^14,15 C-reactive protein is involved in the initiation and modulation of the classical complement pathway, as well as the alternative complement pathway, through binding with complement factor H (CFH [OMIM 134370]). Therefore, it could be postulated that high concentrations of CRP play a causal role in the development of AMD by altering complement-related inflammatory responses to drusen.

Demonstration of a causal role of CRP in AMD is problematic, as is illustrated by the abundant work performed on CRP and coronary heart disease (CHD). Despite the numerous studies (including many prospective studies) consistently showing an increased risk of CHD with increasing CRP concentrations,¹⁶ no studies have demonstrated an association between single-nucleotide polymorphisms (SNPs) in the CRP gene and the risk of CHD.^17,18 This approach, described as Mendelian randomization,¹⁹ showed that, although certain CRP SNPs have been shown to influence the concentrations of circulating CRP,²⁰ the lack of an association between those CRP SNPs and CHD suggests that CRP is unlikely to have a causal influence on CHD.¹⁷ A key point in the Mendelian randomization approach is that the causal role of a risk exposure for a specific disease is assessed by testing the association between genetic variants (that have been reliably shown to influence the exposure) and the disease of interest. The Mendelian randomization approach avoids bias due to reverse causation because genotypes are fixed at conception and, therefore, are not subject to confounding or modification by the onset of disease.²¹

Three previous studies have investigated the association between CRP genetic variants and AMD, but none have found an association.^22-24 The total number of late AMD cases in the 3 studies was 394, and each study may have lacked power. We investigated the association between AMD and CRP polymorphisms in 3 studies: the European Eye (EUREYE) Study, a cross-sectional, population-based epidemiologic study of AMD prevalence and risk factors,²⁵ and 2 hospital-based AMD case-control studies, the Cambridge AMD study⁷ and the Moorfields Eye Hospital (MEH) AMD study.²⁶

For all 3 studies, ethical approval was obtained from either national or local ethics committees (for the Cambridge AMD study, Anglia and Oxford Multicentre Research Ethics Committee; for the MEH AMD study, Moorfields and Whittington Local Research Ethics Committee; and for the EUREYE study, the local ethics committees of all 7 participating centers in 7 different countries). The study adhered to the tenets of the Declaration of Helsinki.²⁷ All participants provided written informed consent to undergo the clinical examination and epidemiologic data collection and to provide a blood sample for biochemical and genetic analyses.

Participants were recruited between November 6, 2000, and November 15, 2002, by random sampling of the population 65 years of age or older in the following 7 centers: Bergen (Norway), Tallinn (Estonia), Belfast (United Kingdom), Paris-Creteil (France), Verona (Italy), Thessaloniki (Greece), and Alicante (Spain). The methods have been published in full elsewhere.²⁵ Trained field workers administered structured questionnaires including smoking status, diet, sunlight exposure, and self-reported medical history. The ophthalmic examination included dilated stereo-digital fundus photography. Stereo fundus images were graded at a single reading center (Erasmus University Rotterdam) according to the International Classification of Age-related Maculopathy and Macular Degeneration²⁸ and then categorized into 5 mutually exclusive grades. Grade 0 was defined as a macula free of drusen or pigmentary irregularities or with hard drusen (<63 µm) only. Early AMD was subdivided into grade 1, defined as soft distinct drusen (≥63 µm) or pigmentary abnormalities; grade 2, defined as soft indistinct drusen (≥125 µm), reticular drusen only, or soft distinct drusen (≥63 µm) with pigmentary abnormalities; and grade 3, defined as soft indistinct drusen (≥125 µm) or reticular drusen with pigmentary abnormalities. Grade 4 was defined as the presence of either neovascular AMD (nvAMD; the presence of any of the following: serous or hemorrhagic retinal or retinal pigment epithelial detachment, subretinal neovascular membrane, or periretinal fibrous scar) or geographic atrophy (GA; well-demarcated area of retinal pigment atrophy with visible choroidal vessels).

In the present analysis, for reason of costs, a random sample of controls (AMD stage 0) was frequency matched on age (within 1 year), sex, and center with patients with late AMD (nvAMD or GA) and patients with early AMD (stages 2 and 3). Nonfasting venous blood samples were collected at the ophthalmologic examination and separated within 4 hours of collection. Serum was stored at –20°C for up to 4 weeks before frozen samples were transferred to a single laboratory (Queens University Belfast) for storage at –80°C.

The Cambridge AMD study is a case-control study of AMD with participants recruited from ophthalmic clinics in London, the southeast of England, and the northwest of England between November 1, 2001, and April 30, 2007.⁷ All patients had at least 1 eye affected by nvAMD and/or GA. Patients were excluded if they had greater than 6 diopters of myopic refractive error or evidence of other inflammatory or retinovascular disease (such as retinal vessel occlusion, diabetic retinopathy, or chorioretinitis) that could contribute to the development of or confound the diagnosis of maculopathy. Almost all of the controls were spouses or partners of index patients, and the remainder were friends of patients. All participants described their race/ethnicity as white rather than other on a recruitment questionnaire. Participants were examined by an ophthalmologist and underwent color stereoscopic fundus photography of the macular region. Health, lifestyle, and smoking data were also collected from the participants. Images were graded at the Reading Centre, MEH, London, using the International Classification of Age-related Maculopathy and Macular Degeneration.²⁸ Blood samples were obtained at the time of interview; EDTA samples were obtained for DNA extraction, and lithium-heparin plasma samples stored at –80°C were later used for CRP measurements.

The MEH AMD study is a case-control study of AMD undertaken at MEH, London, between February 1, 2001, and January 31, 2004.²⁶ Only patients and controls of European origin were included in the study. Patients attending routine follow-up appointments and all new referrals from primary care clinics, casualty, and other centers with any type of AMD were enrolled to try to limit selection bias. Patients were excluded if they had retinochoroidal inflammatory disease, diabetic retinopathy, branch retinal vein or artery occlusion, or any other cause of visual loss other than amblyopia. Controls were recruited from spouses or friends of patients or from local residential homes for elderly persons within 8 km (5 miles) of the hospital. Each participant was interviewed specifically for the study, and a family history, smoking history, and other medical history were obtained. The ophthalmic examination included Snellen acuity, slitlamp examination, and biomicroscopic ophthalmoscopy. All participants underwent color stereoscopic fundus photography of the macular region, and the images were graded at the Reading Centre, MEH, London, using the International Classification of Age-related Maculopathy and Macular Degeneration.²⁸ Controls were excluded if drusen larger than 63 µm were evident. A sample of peripheral blood was obtained from each participant and stored at –20°C for DNA extraction and –80°C for analyte measurement.

Circulating CRP concentrations were measured in the EUREYE and Cambridge AMD studies. In the EUREYE Study, the concentrations of serum CRP were assessed by a high-sensitivity, latex-enhanced turbidometric immunoassay (Wako Chemicals GmbH) using a Cobas Fara analyzer (Roche Diagnostics). In the Cambridge AMD study, the concentrations of plasma CRP were measured by a high-sensitivity, particle-enhanced immunonephelometric assay (Siemens CardioPhase hsCRP; Siemens Medical Solutions) using a Siemens Dade Behring BNII Nephelometer. Participants with CRP levels higher than 10 mg/L (to convert CRP to nanomoles per liter, multiply by 9.524), indicative of acute infection,²⁹ were not included in the analysis (90 from the EUREYE Study and 51 from the Cambridge AMD study). C-reactive protein concentrations were natural logarithmically transformed to ensure normality of the distribution and further Z transformed for investigation of the association of a 1-SD increase in log CRP with AMD.

We chose 4 common SNPs (rs1205, rs1130864, and rs1800947 in the 3′ UTR [untranslated region] and synonymous rs3093077 in exon 2) in the CRP gene previously reported in a meta-analysis on CRP and CHD using a Mendelian randomization approach.¹⁷ These variants were selected as the most parsimonious set of tagging SNPs capturing maximum haplotype diversity and of demonstrated association with CRP concentrations among individuals of European descent.^21,30 The EUREYE Study samples were genotyped at LGC Genomics (Hoddesdon, England) using KASPar chemistry, a competitive allele-specific polymerase chain reaction SNP genotyping system using fluorescence resonance energy transfer quencher cassette oligos. Genotyping in the Cambridge AMD study was carried out using the ABI PRISM SNaPshot ddNTP Primer Extension Kit and a 3100 Genetic Analyzer (Applied Biosystems). Genotyping in the MEH AMD study was carried out using a Taqman (ABI) assay (Applied Biosystems). Departure from Hardy-Weinberg equilibrium was assessed for all SNPs in the control groups using a χ² goodness-of-fit test.

Statistical analysis was performed from August 1 to November 30, 2016. We investigated the association of CRP genotype with late AMD and by late AMD phenotype (GA or nvAMD). We also report results for early AMD (available in the EUREYE Study only). All association analyses were conducted as 2-stage, fixed-effects meta-analysis of the available individual participant data from the 3 studies. Heterogeneity across study centers was assessed using the I² statistic. We calculated odds ratios (ORs) and their corresponding 95% CIs using logistic regression models adjusted for sex and age. Odds ratios were expressed as per 1 minor allele (multiplicative genetic model).

To confirm that, in our study population, the selected SNP genotypes influence circulating CRP concentrations, we examined the association of CRP genotype with CRP concentrations in controls via a Wald test on the genotype coded as 0, 1, and 2 according to the number of minor alleles using linear regression models adjusted for age and sex. We report geometric mean (SD) values for concentrations by genotype.

For comparison with previous studies, we analyzed the association of CRP concentrations with late AMD and late AMD phenotype using logistic regression models. These analyses were adjusted for age and sex and, additionally, for a priori confounders including history of myocardial infarction or stroke, diabetes, and smoking. Odds ratios were expressed as per 1-SD change of log CRP concentration.

All statistical analyses were conducted using Stata software, version 13.1 (StataCorp), and ipdmetan and mvmeta commands were used for conducting meta-analyses of individual participant data. P < .05 was considered significant.

There were 1727 cases of late AMD (1151 neovascular, 384 geographic atrophy, and 192 mixed [neovascular AMD and geographic atrophy]) and 1153 controls. Among the 1727 patients with late AMD, the mean (SD) age was 78.7 (7.4) years, and 668 (38.7%) were men. The mean (SD) age of the controls was 74.9 (7.0) years, and 510 (44.2%) were men. The characteristics of the participants and the genotype frequencies by study are shown in Table 1. Genotyping of rs3093077 failed in the EUREYE Study, and rs1800947 was not typed in the MEH AMD study and the Cambridge AMD study. In each study, no deviation from Hardy-Weinberg equilibrium was observed for any SNP, and control minimum allele frequencies (MAF) were similar both between studies and to those reported for the 1000 Genomes Project^31,32 Phase 3 European (EUR) samples: rs1205 MAF (T allele), 0.34; rs1130864 MAF (T allele), 0.33; rs3093077 MAF (G allele), 0.07; and rs1800947 MAF (C allele), 0.05. There were significant associations in the pooled analysis for the rs1205 T allele with lower CRP concentrations in controls and for the rs1130864 T allele with higher CRP concentrations in controls (Table 2). For the 2 rarer SNPs rs3093077 and rs1800947, there was no association with CRP concentrations in controls.

There was no evidence of any association of CRP SNPs rs1205, rs1130864, or rs3093077 with late AMD, or with type of late AMD, either from the individual studies or in the meta-analysis, or for SNP rs1800947 in the EUREYE Study (Table 3).

A significant association was observed in the pooled age- and sex-adjusted analysis of CRP concentrations with late AMD (Table 4). Additional adjustment for smoking, myocardial infarction and stroke, and diabetes slightly reduced the ORs (from 1.19 to 1.16), and the P value changed from 0.03 to 0.07. Analyses of CRP concentrations by type of late AMD showed that GA had an OR of 1.35 (95% CI, 1.03-1.76; P = .03) and that nvAMD had an OR of 1.09 (95% CI, 0.91-1.29; P = .36).

Only the EUREYE Study had population-based data available on early AMD. There was no association of CRP concentrations with early AMD in age- and sex-adjusted or fully adjusted analyses (OR, 1.01; 95% CI, 0.82-1.25; P = .90) or of any CRP SNPs with early AMD (rs1205: OR, 1.08; 95% CI, 0.98-1.20; P = .13; rs1130864: OR, 0.89; 95% CI, 0.75-1.06; P = .20; and rs1800947: OR, 1.20; 95% CI, 0.89-1.63; P = .23).

In this meta-analysis of 3 studies of AMD including 1727 patients with late AMD and 1153 controls, we found no evidence from analysis of CRP variants to support a causal association of CRP concentrations with late AMD, either nvAMD or GA. In agreement with many other studies,^17,18,33 we found that the T allele of rs1205 was associated with lower CRP concentrations in controls. However, there was no protective association between the T allele and either nvAMD or GA in any of the 3 studies. Similarly, the T allele of rs1130864 was associated with higher CRP concentrations in controls, but the T allele was not associated with higher odds of either late AMD phenotype.¹⁷

None of the 3 previous studies that investigated CRP polymorphisms and AMD found an association.^22-24 These studies had smaller sample sizes than our investigation and included two²³ or three^22,24 of the SNPs we investigated (rs1205, rs1130864, and rs1800947). In a case-control study (111 mixed early and late cases and 401 controls) nested within the prospective follow-up of the Physicians Health Study, AMD risk was not influenced by any of 7 CRP SNPs²⁴ or by any of 9 CRP SNPs in a sib-pair study of 112 cases of nvAMD.²² The Rotterdam study (5861 participants, including 171 with late AMD) suggested that CRP haplotypes may modify the association of CFH Y402H genotype with late AMD, but the CRP haplotypes did not influence AMD risk directly; results were not presented for the 3 CRP SNPs individually.²³ In addition, a meta-analysis of 15 genome-wide association studies (GWASs)³⁴ and a recent large GWAS (16 144 patients and 17 832 controls)³⁵ of AMD did not show any genome-wide significant association signals at the CRP locus.

In our fully adjusted association analyses between CRP concentrations and AMD, we did not find a significant association with all cases of late AMD (P = .07; 1324 patients, including 489 with late AMD) and nvAMD (P = .36; 318 patients) or with early AMD (P = .90; 574 patients), which has been reported in other studies.^9,15,36 The association with AMD is not consistent, particularly in studies investigating early AMD.^11,37-40 In a large meta-analysis comparing low (<1 mg/L) vs high (>3 mg/L) CRP concentrations, Hong et al¹⁴ reported an OR of 1.30 (95% CI, 1.03-1.65) in 3 studies that combined early and late AMD (4522 patients, including 1629 with AMD) and an OR of 2.19 (95% CI, 1.38-3.47) in 8 studies of late AMD (35 168 patients, including 1603 with AMD). Estimates for late AMD varied considerably according to method of AMD ascertainment, with much higher ORs reported for clinical diagnosis (3.8) compared with fundus photography (1.4), adjustment for diabetes (1.6) compared with no adjustment (2.5), and study design. Population-based cross-sectional or case-control studies had lower ORs than did clinic case-control studies or longitudinal studies. In a pooled analysis of prospective, nested case-control studies from 5 population-based US cohorts (647 patients, including 190 with nvAMD), CRP concentrations above 3 mg/L were associated with risk of all late AMD and nvAMD compared with concentrations less than 1 mg/L.¹⁵

Although our study was larger than the 3 previous studies that investigated the association between CRP variants and AMD, it does have several limitations, especially with respect to the association between CRP concentrations and AMD. In the EUREYE and Cambridge studies, the blood samples used for measuring CRP concentrations were collected at the same time as ascertainment of AMD, and therefore we cannot exclude reverse causation (eg, an inflammatory response) due to risk factors for AMD, such as atherosclerosis and cardiovascular disease. We only had 1 measure of CRP, which may not represent longer-term usual concentrations. In addition, circulating CRP concentrations may not adequately reflect the CRP concentration in the relevant tissues. The monomeric form of CRP has been shown to be responsible for the upregulation of proinflammatory cytokines interleukin 8 and chemokine ligand 2 in retinal pigment epithelial cells, and patients with AMD carrying the risk variants of CFH Y204H showed impaired binding of monomeric CRP.⁴¹

Genetic variants in CRP are not subject to reverse causation^19,21 and, therefore, provide stronger evidence on the possible association of CRP concentrations with AMD. We investigated only a small number of CRP SNPs. Nevertheless, we selected SNPs expected to show a causal association with AMD because these SNPs influence CRP concentrations.^17,20,21,30 Variants associated with CRP concentrations in other loci have been identified in large GWASs and replication studies.^18,42 These variants are found in genes primarily involved in lipid, metabolic, or immune pathways and include IL6R (OMIM 147880), HNF1A (OMIM 142410), and APOCI (OMIM 107710). In a GWAS of CHD, the APOC1 variant (rs4420638) was associated with increased risk of CHD but in an opposite direction to that anticipated from its association with CRP and consistent with its association with low-density lipoprotein and high-density lipoprotein cholesterol and triglycerides.¹⁸ This variant has also been identified at genome-wide significance to increase the risk of AMD.³⁴APOC1 is part of the apolipoprotein gene cluster that plays a major role in lipid transport and metabolism and includes the well-established APOE (OMIM 107741) AMD risk variant (rs429358).⁴³ Linkage disequilibrium (r² = 0.7)⁴⁴ is observed between rs4420638 and rs429358 in the EUR populations from Phase 3 of the 1000 Genomes Project.^31,32

From the analogy with CHD¹⁷ and the identification of alternative genetic loci for CRP concentrations already discussed, along with the lack of association of variants in the CRP locus with AMD shown in our meta-analysis corroborating results from previous studies, we conclude that CRP concentrations are unlikely to be causally associated with AMD.

Corresponding Author: Valentina Cipriani, PhD, UCL Institute of Ophthalmology, Department of Ocular Biology and Therapeutics, University College London, 11-43 Bath St, Wolfson Bldg, 3rd Floor, Room 3/5, London EC1V 9EL, England (v.cipriani@ucl.ac.uk).

Accepted for Publication: May 19, 2017.

Published Online: July 27, 2017. doi:10.1001/jamaophthalmol.2017.2191

Author Contributions: Drs Cipriani and Fletcher had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Drs Cipriani and Hogg contributed equally to this work.

Study concept and design: Cipriani, Sofat, Fletcher.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Cipriani, Hogg, Fletcher.

Critical revision of the manuscript for important intellectual content: Hogg, Sofat, Moore, Webster, Yates, Fletcher.

Statistical analysis: Cipriani.

Obtained funding: Sofat, Yates, Fletcher.

Administrative, technical, or material support: Sofat.

Study supervision: Moore, Webster, Yates, Fletcher.

Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest and none were reported.

Funding/Support: The European Eye (EUREYE) Study was supported by grant QLK6-CT-1999-02094 from the European Commission Vth Framework, with additional funding for cameras provided by the Macular Disease Society UK for DNA extraction and genotyping (MRC Biomarkers Award G0601354). The Cambridge Age-related Macular Degeneration (AMD) study was funded by grant G0000067 from the Medical Research Council. The Moorfields Eye Hospital AMD study was partly supported by Fight for Sight, including the Mercer Fund. Dr Cipriani is funded by the National Institute for Health Research Biomedical Research Centre at Moorfields Eye Hospital National Health Service Foundation Trust and the University College London Institute of Ophthalmology. Dr Sofat was funded by the British Heart Foundation (Schillingford) Clinical Research Training Fellowship FS/07/011.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Group Information: The EUREYE Study group members include Astrid E. Fletcher, PhD (London School of Hygiene and Tropical Medicine, London, England); Graham Bentham, MA (University of East Anglia, Norwich, England); Usha Chakravarthy, MD (Queen’s University Belfast, Belfast, United Kingdom); Paulus TVM de Jong, MD (Netherlands Institute for Ophthalmic Research, Amsterdam, Netherlands); Mati Rahu, PhD (National Institute for Health Development, Tallinn, Estonia); John Seland, MD (University of Bergen, Bergen, Norway); Gisele Soubrane, MD (Universitaire de Créteil, Paris, France); Laura Tomazzoli, MD (Università degli studi di Verona, Verona, Italy); Fotis Topouzis, MD (Aristotle University of Thessaloniki, Thessaloniki, Greece); Jesus Vioque, MD (Universidad Miguel Hernandez, Alicante, Spain); and Ian Young, MD (Queen’s University Belfast, Belfast, United Kingdom).

Disclaimer: The views expressed in this publication are those of the authors and not necessarily those of the funding agencies.

Additional Contributions: We acknowledge help with patient recruitment for the Cambridge AMD study from members of the Genetic Factors in AMD Study Group. Jane Khan, MBBS, FRCOphth, MD, Centre for Ophthalmology and Visual Science, University of Western Australia, and Humma Shahid, BM, BCh, MA, DM, FRCOphth, Cambridge Institute for Medical Research, Cambridge, England, and Addenbrooke’s Hospital, Cambridge University Hospitals NHS Foundation Trust, provided expert clinical evaluation and sampling of the study participants. Dr Khan was compensated for her contribution. Dr Shahid was not compensated.