Macular Thickness and Associated Factors in African Immigrants: Insights from the Africans in America Study

Abstract

Purpose

Normative optical coherence tomography (OCT) databases may not be generalizable to Africans, one of the fastest growing populations in the United States. Given the paucity of normative OCT data in Africans, we evaluated macular thickness in African immigrants.

Patients and Methods

This cross-sectional study analyzed African-born participants enrolled in the Africans in America Study. They underwent history, physical, oral glucose tolerance tests, hemoglobin electrophoresis, and macular OCT (Cirrus HD-OCT). In the primary analysis, multivariable linear regression excluding eyes with retinopathy, glaucoma, or ocular hypertension was used to evaluate the association between the central subfield thickness (CST) and demographic, clinical, and metabolic factors. In the secondary analysis, multivariable analysis was used to evaluate the association between macular layer thicknesses and prediabetes status (prediabetes vs normal glucose tolerance). P values less than 0.05 and false discovery rate (FDR) within 10% were considered significant.

Results

Six hundred forty-three eyes of 330 participants (60% male; mean age: 40 ± 11 years) were analyzed. The mean CST thickness for the cohort was 238.53 ± 21.48 µm. On average, women had a lower CST than men [−10.51 µm, 95% confidence interval: (−15.75, −5.26), P < 0.001, FDR < 0.001]. No significant associations were found between prediabetes status and the thickness of any macular layer.

Conclusion

These findings provide descriptive reference data on Cirrus-derived macular thickness measurements among African-born adults. Women had a lower CST, on average, than men. In contrast to prior studies, there was no difference in macular layer thicknesses between participants with prediabetes and those with normal glucose tolerance.

Introduction

According to the Centers for Disease Control and Prevention, the prevalence of visual acuity loss or blindness in the United States (US) varies by age and race/ethnicity. Racial and ethnic minority groups, including Hispanic and African American individuals, have a greater ocular disease burden compared to Non-Hispanic White individuals.¹ For example, African-American individuals have a greater burden of diabetic retinopathy and glaucoma compared to White individuals.^2,3

Although ophthalmic disparities have been described across racial and ethnic groups, individuals of African ancestry are often analyzed as a single homogeneous category in research.^4,5 However, African-born immigrants represent a distinct and expanding population in the US,⁶ and this group may differ from US-born African-Americans. Africa is characterized by the highest genetic diversity of any continent.⁷ African populations demonstrate genetic contact across the continent.⁸ In contrast, African-American populations originate largely from Niger-Kordofanian with additional European and other African groups, consistent with ancestry from West and West-Central African populations with subsequent admixture.⁹ Additionally, African populations can differ from African-American populations in social determinants of health, such as insurance status, and these differences are relevant to health outcomes and disparities research.^5,10

Macular optical coherence tomography (OCT) measurements show variation by race/ethnicity and sex, supporting the need for population-based reference data when interpreting retinal thickness.¹¹ However, OCT normative databases have been criticized for limited diversity and reliance on race-based categorization, and recent work emphasizes that more representative datasets are needed to improve fairness and generalizability when applying these databases across populations.^12–14 Given the paucity of OCT data specifically in African-born immigrants, we focused on this group enrolled in the Africans in America study. Considering the prior literature demonstrating retinal thinning in prediabetes among other demographic groups,^15–17 we also evaluated whether African immigrants with prediabetes had retinal thinning compared to those with normal glucose tolerance.

Material and Methods

Africans in America Study

The Africans in America Study was designed as a cross-sectional study to identify the sociologic and biologic determinants of type 2 diabetes mellitus (T2DM) in African-born Black individuals living in the United States.¹⁸ Recruitment was achieved with the use of social media, newspaper advertisements, community events, presentations, and flyers. The National Institutes of Health (NIH) Institutional Review Board approved the protocol (99DK0002 / CR004159; Clinical Trials.gov Identifier: NCT00001853), which adheres to the tenets of the Declaration of Helsinki Accords. Prior to study commencement, all individuals gave written informed consent. It was not appropriate or possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research.

To be eligible for the study protocol, which was conducted at the NIH Clinical Center, individuals had to report that they were born in sub-Saharan Africa (with both parents also born in sub-Saharan Africa), self-identified as Black, were aged 18 to 70 years, and currently lived in metropolitan Washington, DC. As this protocol is designed to identify persons living with T2DM who are undiagnosed, no one with known T2DM was enrolled. Use of lipid-lowering and antihypertensive medications was permitted.

At Visit 1, 488 African immigrants provided their demographic and medical history, and physical examinations were performed, along with routine blood tests including hemoglobin, creatinine, liver function tests, and thyroid-stimulating hormone. Data was gathered on: age of immigration, years of United States (US) residence, weight gain in the US, exercise, income, education, marital status, and smoking and alcohol consumption. Thirty-eight individuals did not proceed to Visit 2.

Four-hundred fifty individuals were qualified to move to the baseline visit, which took place within 2 weeks of the screening visit. Weight, height, blood pressure, waist circumference, hemoglobin A1c (HbA1c), lipid profiles, and hemoglobin electrophoresis were obtained. In addition, an oral glucose tolerance test (OGTT) (Trutol 75, Custom Laboratories) was performed. Ninety-five participants did not have an eye examination because of scheduling issues related mostly to COVID, leaving 355 participants with eye examinations (Figure 1). Participants included in the current analysis underwent their first study visit from June 2013 to December 2024.

Figure 1.

Flowchart of participants and eyes included in the analysis. *four eyes were excluded for multiple reasons.

Abbreviation: OCT, optical coherence tomography.

Glucose Tolerance Status by OGTT

OGTT was used to diagnose participants with diabetes or prediabetes.¹⁹ A fasting glucose ≥126 mg/dL or 2h glucose ≥200 mg/dL was diagnosed as diabetes. A fasting glucose >100 mg/dL and <126 mg/dL or 2h glucose ≥ 140 mg/dL and <200 mg/dL were diagnosed as prediabetes.

Ophthalmic Imaging and Automatic Segmentation

Participants enrolled in the Africans in America Study underwent a comprehensive eye examination at the National Eye Institute (NEI), including visual acuity measurement, intraocular pressure measurement, fundus photography, and spectral-domain OCT of the macula (Zeiss CIRRUS HD-OCT 5000). A subset of participants underwent manifest refraction. Total retinal thickness measurements within the nine Early Treatment Diabetic Retinopathy Study (ETDRS) subfields, including the central subfield thickness, were obtained using the native Cirrus HD-OCT software. Separately, the OCT scans underwent automated segmentation of the retinal layers with the Iowa Reference Algorithms (v. 3.8) to quantify individual retinal cell layer thicknesses.^20–22 The Iowa Reference Algorithms have demonstrated better repeatability than commercial software (including the Zeiss software), and, unlike the Zeiss segmentation software, can segment 10 intra-retinal layers.²³ We recorded the thickness (in micrometers) of the total retina as well as the following retinal layers: retinal nerve fiber layer (RNFL), ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear layer (ONL), and retinal pigment epithelium (RPE). The thickness of the total retina and each of the retinal layers was collected within each of the ETDRS subfields²⁴ (Figure 2).²⁵

Figure 2.

Early Treatment diabetic retinopathy study subfields (top) of the macular retina (bottom) on Cirrus HD-OCT.

Statistical Analysis

We excluded eyes with an OCT signal strength worse than 7/10 or no quality score available, glaucoma or ocular hypertension, retinopathy, macular edema, retinal dystrophy, and eyes with scans unreadable by the segmentation software. Although the manufacturer defines signal strength ≥6 as acceptable,²⁶ we applied a more conservative threshold of ≥7 to reduce potential measurement variability and segmentation errors associated with lower-quality scans. Prior studies have also applied a threshold of ≥7.^27,28 Participants with any missing data among the covariates used in the multivariable regression models (discussed below) were also excluded from the analysis. The presence of pathology was determined by 4 retina specialists at the NEI. Participants at risk for glaucoma based on an intraocular pressure (IOP) greater than 21 mmHg were excluded. The temporal subfield was calculated as the average of the temporal inner macula (TIM) and temporal outer macula (TOM). We evaluated the temporal subfield separately because of prior findings of retinal thinning in sickle cell disease in this region.^29,30 The inner macular ring thickness was calculated as the average of the superior IM, nasal IM, inferior IM, and TIM. Similarly, the outer macular ring thickness was calculated as the average of the superior OM, nasal OM, inferior OM, and TOM. Statistical analysis was conducted with SAS v. 9.4 (SAS Institute, Cary, NC, USA). The unit of analysis was the eye and, to account for the correlation between both eyes of each participant, an unstructured covariance structure was specified.

Primary Analysis

If associations were significant in the univariable regression analysis at P < 0.1, the factor was included as a covariate in the multivariable regression model. For categorical variables with more than two levels, inclusion was determined using the global Type III test P value. Factors examined in this analysis included: glucose tolerance status, age at enrollment, age at immigration, sex, smoking, body mass index (BMI), waist circumference, hip circumference, estimated glomerular filtration rate (EGFR), cholesterol, triglyceride, high-density lipoprotein, low-density lipoprotein, hemoglobin type, and HbA1c. We used a mixed-effects multivariable linear model (SAS Proc Mixed, version 9.4) with a continuous outcome (central subfield, temporal subfield, inner macular ring, or outer macular ring thickness), adjusting for clinical and demographic covariate with a P < 0.1 in the univariable analysis, with a repeated measures effect to account for correlation of eyes within an individual, where the correlation structure was specified as unstructured. We also conducted a sensitivity analysis adjusting for refractive error, in addition to the aforementioned covariates. Emmetropia was defined as a spherical equivalent of less than ±0.50 D. Myopia was defined as a spherical equivalent of less than −0.50 D, and hyperopia was defined as a spherical equivalent of greater than 0.50 D. The false discovery rate (FDR) was used to account for potential type I errors from multiple comparisons. P values < 0.05 and FDR within 10% were considered significant for the multivariable analysis.

Secondary Analysis

For the secondary analysis, we used multivariable regression analysis to compare retinal cell layer thickness in each ETDRS subfield, as well as the thickness of the inner and outer macular rings, between participants with prediabetes and those with normal glucose tolerance testing. All analyses were adjusted for age, sex, BMI, systolic blood pressure, high‑density lipoprotein, triglycerides, and smoking status. A P value < 0.05 and FDR within 10% was considered significant.

Results

At the time of this analysis, 355 participants were enrolled in the Africans in America Study and had completed OGTT and an eye examination. Of the 710 eyes, 15 were excluded due to a signal strength less than 7/10 or no macular OCT scan done, 38 due to glaucoma or IOP > 21 mmHg, 13 due to retinopathy, 1 due to macular edema, 2 due to retinal dystrophy, and 1 due to the OCT scans being unreadable by the segmentation software. Overall, 159 participants with OGTT data did not undergo a full eye examination or were excluded due to imaging criteria or confounding ocular pathology (Figure 1). These 159 individuals did not differ sociologically or geographically from those individuals who were included (Supplement Table 1). Six hundred forty-three eyes of 330 participants were included in the analysis. Table 1 displays the demographic and clinical characteristics of this cohort, and Table 2 shows the age and sex distribution. The average thickness of the central subfield, temporal subfield, inner macular ring, and outer macular ring were 238.53 ± 21.48 µm, 284.58 ± 15.21 µm, 312.42 ± 16.97 µm, and 274.73 ± 13.16 µm, respectively, and Figure 3 shows the average total retinal thickness at each ETDRS subfield among all eyes and eyes stratified by sex.

Table 1.

Demographic and Clinical Characteristics of Participants.*

	Total (N=330)
Age at enrollment, y, mean ± standard deviation	39.6 ± 10.7
Age at immigration, y, mean ± standard deviation	28.0 ± 11.7
Years in the United States, mean ± standard deviation	11.8 ± 10.3
Female, no. (%)	133 (40.3)
Has Health Insurance, no. (%)	235 (71.2)
Smoking Status, no. (%)
Current	15 (4.6)
Former	6 (1.8)
Never	309 (93.6)
Education, no. (%)
≤ High School Diploma	22 (6.7)
Entered College	53 (16.1)
Completed College	108 (32.7)
Entered Graduate School	28 (8.5)
Completed Graduate School	118 (35.8)
Other	1 (0.3)
Waist Circumference Superior to Iliac Crest, cm, mean ± standard deviation	91.2 ± 12.0
Hip Circumference, cm, mean ± standard deviation	104.3 ± 10.0
BMI, kg/m2, mean ± standard deviation	27.7 ± 4.6
High-density lipoprotein, mg/dL, mean ± standard deviation	52.5 ± 13.3
Low-density lipoprotein, mg/dL, mean ± standard deviation	102.8 ± 32.0
Cholesterol, mg/dL, mean ± standard deviation	170.1 ± 35.5
Triglycerides, mg/dL, mean ± standard deviation	75.0 ± 38.8
EGFR, mL/min/1.73 m2, mean ± standard deviation	106.5 ± 20.4
Hemoglobin Type, no. (%)
AA	270 (81.8)
AC	7 (2.1)
AS	50 (15.2)
AF	3 (0.9)
HbA1c, %, mean ± standard deviation	5.4 ± 0.7
2-hour glucose, mg/dL, mean ± standard deviation	136.5 ± 45.8
Glucose Tolerance Status by OGTT, no. (%)
Diabetes Mellitus	29 (8.8)
Prediabetes	103 (31.2)
Normal OGTT	198 (60.0)

Notes: *Missing data on 1 participant for waist circumference, 1 participant for cholesterol, and 3 participants for HbA1c.

Abbreviations: BMI, body mass index; EGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; OGTT, oral glucose tolerance test.

Table 2.

Distribution of Participants Stratified by Age and Sex

Age Category (y)	Male, no. (%)	Female, no. (%)
20–29	42 (21.3)	24 (18.1)
30–39	72 (36.6)	38 (28.6)
40–49	50 (25.4)	35 (26.3)
50–59	23 (11.7)	29 (21.8)
60–69	10 (5.1)	7 (5.3)

Figure 3.

Total retinal thickness (µm) at each ETDRS subfield of eyes included in the analysis (A), eyes from male participants (B), and eyes from female participants (C). Darker shades represent higher thicknesses.

Supplement Table 2 shows the results of the univariable regression analysis for the association between retinal thickness and demographic, clinical, and metabolic factors. Central subfield thickness was associated with sex, hip circumference, BMI, high-density lipoprotein, and hemoglobin AC (P < 0.1). Temporal subfield thickness was associated with age at enrollment, sex, hip circumference, BMI, and high-density lipoprotein. The inner macular ring thickness was associated with age at enrollment, age at immigration, sex, hip circumference, BMI, and high-density lipoprotein (P < 0.1). The outer macular ring thickness was associated with age at enrollment, years in the US, sex, high-density lipoprotein, and EGFR (P < 0.1).

In all multivariable models (Table 3), the variance inflation factors (VIF) were below 5, indicating no substantial multicollinearity (maximum VIF for central subfield thickness: 3.89; inner macular ring: 3.83; outer macular ring: 1.25; temporal subfield: 3.84). Female sex was statistically significantly associated with decreased central subfield (−10.51 µm [−15.75, −5.26], P < 0.001, FDR < 0.001), temporal subfield (−7.21 µm [−10.64, −3.79], P < 0.001, FDR = 0.0005), and inner macular thickness (−8.37 µm [−12.47, −4.27], P < 0.001, FDR = 0.0006) (Table 3). Hemoglobin AC was significantly associated with decreased central subfield relative to hemoglobin AA (−18.24 µm [−33.48, −3.01], P = 0.019, FDR = 0.09).

Table 3.

Multivariable Regression Results for the Association Between Central Subfield Thickness and Demographic, Clinical, and Metabolic Factors

	Central Subfield			Temporal Subfield			Inner Macular Ring			Outer Macular Ring
	Adjusted Estimate (95% CI)	P	FDR	Adjusted Estimate (95% CI)	P	FDR	Adjusted Estimate (95% CI)	P	FDR	Adjusted Estimate (95% CI)	P	FDR
Age at Enrollment	*	*	*	−0.09 (−0.23, −0.05)	0.206	0.38	−0.06 (−0.27, 0.15)	0.555	0.56	−0.13 (−0.27, 0.01)	0.077	0.30
Age at Immigration	*	*	*	*	*	*	−0.14 (−0.32, 0.05)	0.141	0.42	*	*	*
Years in US	*	*	*	*	*	*	*	*	*	−0.06 (−0.21, 0.09)	0.416	0.42
Sex
Female	−10.51 (−15.75, −5.26)	<0.001	<0.001	−7.21 (−10.64, −3.79)	<0.001	0.0005	−8.37 (−12.47, −4.27)	<0.001	0.0006	−2.36 (−5.35, 0.62)	0.120	0.30
Male	Reference			Reference			Reference			Reference
BMI	−0.53(−1.43, 0.38)	0.251	0.31	−0.15 (−0.74, 0.48)	0.631	0.79	−0.34 (−1.05, 0.37)	0.347	0.52	*	*	*
High Density Lipoprotein	−0.18 (−0.36, −0.002)	0.047	0.09	−0.07 (−0.19, 0.04)	0.224	0.38	−0.08 (−0.23, 0.06)	0.247	0.49	−0.08 (−0.18, 0.03)	0.178	0.30
Hip Circumference	0.11 (−0.32, 0.54)	0.618	0.62	0.02 (−0.26, 0.30)	0.880	0.88	0.12 (−0.22, 0.45)	0.498	0.55	*	*	*
EGFR	*	*	*	*	*	*	*	*	*	0.04 (−0.03, 0.11)	0.309	0.39
Hemoglobin Type			0.09	*	*	*	*	*	*	*	*	*
AS	−5.00 (−11.18, 1.18)	0.112		*	*	*	*	*	*	*	*	*
AC	−18.24 (−33.48, −3.01)	0.019		*	*	*	*	*	*	*	*	*
AF	3.08 (−20.13, 26.28)	0.794		*	*	*	*	*	*	*	*	*
AA	Reference			*	*	*	*	*	*	*	*	*

Notes: Boxes shaded in gray represent factors that were not included as covariates in a particular multivariable model due to P > 0.1 in the univariable screening for the outcome of interest (central subfield thickness, temporal subfield thickness, inner macular ring thickness, or outer macular ring thickness).

Abbreviations: BMI, body mass index; EGFR, estimated glomerular filtration rate; HbA1c, hemoglobin A1c; OGTT, oral glucose tolerance test.

In the sensitivity analysis additionally adjusting for refractive error (n=170 eyes [69.4% myopic] of 87 participants), female sex was associated with a decreased temporal subfield (−12.91 µm [−19.34, −6.48], P < 0.001, FDR < 0.001) and inner macular ring (−14.86 µm [−22.51, −7.20], P < 0.001, FDR = 0.001) thickness. Myopia was significantly associated with a decreased outer macular ring thickness (−4.44 µm [−7.45, −1.13], P = 0.009, FDR = 0.08) compared to emmetropia. Higher age at enrollment (−0.43 µm/year [−0.71, −0.15], P = 0.003, FDR = 0.02) and high-density lipoprotein (−0.20 µm/mg/dL [−0.40, −0.0006], P = 0.049, FDR = 0.098) were also associated with a decreased outer macular ring thickness.

For the secondary analysis, there were no significant associations between prediabetes status and the thickness of any retinal cell layers in any ETDRS subfields or inner or outer macular rings, after accounting for type I errors with FDR.

Discussion

The objectives of this study were to evaluate factors associated with macular thickness among African immigrants in the US and determine whether those with prediabetes may undergo retinal thinning compared to those with a normal glucose tolerance test. To our knowledge, this is the first study characterizing healthy retinal structure among African immigrants, one of the fastest-growing populations in the US.⁶ Similar to studies among other demographic groups, we found that female sex was associated with thinner macular thickness compared to male sex. In contrast to prior reports evaluating other racial groups, we found that African immigrants with prediabetes do not tend to have a thinner macula.

The mean thickness of each subfield in our study was within the Cirrus HD-OCT normative range.³¹ This supports that Cirrus-derived total macular thickness measurements among African-born adults were not systematically outside the existing reference ranges. However, larger population-based studies are needed to determine whether current normative databases perform comparably across diverse African-born populations. Clinicians should continue to interpret measurements near normative cutoffs in the context of demographic factors associated with retinal thickness, including sex. OCT normative comparisons are used in clinical decision-making and can contribute to over- or under-diagnosis when reference datasets are not representative. This is relevant given prior studies demonstrating differences in macular thickness based on race and sex, including a thinner central macula in African-American individuals compared to White individuals.¹¹ Furthermore, Addis et al reported that the Cirrus HD-OCT often coded the temporal RNFL thickness of African-Americans (without any nerve fiber layer defects) as abnormal, because it did not account for the variation in RNFL between races and ethnicities,³² which may lead to overdiagnosis of glaucoma. Additionally, there were significant differences in optic nerve head parameters of African-Americans and the cohort comprising the normative database of the Cirrus HD-OCT machine.³² This emphasizes that normative-range findings should be interpreted within the appropriate demographic context. Given that African-descent populations are frequently combined in research and remain underrepresented in OCT normative datasets, further efforts to improve the diversity and generalizability of OCT reference standards are warranted.

As expected, female sex was associated with a thinner macula among African immigrants. This has been demonstrated in other study populations of other geographical areas, including Germany,³³ the Netherlands,³⁴ Spain,³⁵ Norway,³⁶ China,^37,38 and Western Australia.³⁹ Age-related macular thinning involves degeneration of photoreceptors, ganglion cells, retinal pigment epithelial cells, and optic nerve fibers over time.^38,40,41 In contrast to prior literature, we found that age was not significantly associated with macular thickness, except at the outer macular ring. This may be due to a non-linear relationship between age and retinal thickness in other ETDRS regions. For example, Wu et al reported that central macular thickness increased among Chinese participants less than 60 years of age, and decreased among those greater than 60 years of age.³⁸ They hypothesized that the age-related thickening of the central macula in the former group could be due to the accumulation of debris.³⁸

To our knowledge, there are limited cases of retinopathy with hemoglobin type AC, which presents similarly to sickle cell retinopathy.^42,43 Prior OCT reports of sickle cell disease (hemoglobin type SS and SC) have described retinal thinning in the temporal and superior subfields with central foveal sparing (protected from vaso-occlusive injury due to its avascularity) compared to controls.^29,30 In contrast, our cohort consisted of participants with hemoglobin AA, AS, AC, or AF rather than SS or SC. We found that African immigrants with hemoglobin type AC had a significantly lower central subfield thickness compared to those with type AA. Furthermore, there were no significant differences in temporal subfield retinal thickness between hemoglobin types. These differences in results are likely due to a small sample size of participants with hemoglobin type AC (n=7) and differences between hemoglobin traits and sickle cell disease. Overall, the structural retinal phenotype of hemoglobin AC remains incompletely characterized, and our results are exploratory and hypothesis-generating. Future studies with larger cohorts are necessary to establish whether hemoglobin AC has an associated OCT phenotype.

Retinal pathology in diabetes, including neurodegeneration, is driven by hyperglycemia and impaired insulin signaling.⁴⁴ In contrast to prior literature,^15–17 we did not find retinal thinning among African immigrants with prediabetes compared to those with a normal glucose tolerance test. This could be due to differences in the underlying etiology of abnormal glucose tolerance in this population. Prior work from our group demonstrated that β-cell failure, rather than insulin resistance, was the cause of abnormal glucose tolerance among the majority of participants in the Africans in America cohort.¹⁸ Ishimwe et al also found that African immigrants with β-cell failure had a lower prevalence of obesity, milder hyperglycemia, and a more favorable lipid profile than those with insulin resistance.¹⁸ Given that insulin resistance and adipose tissue–driven inflammation can promote retinal thinning independent of hyperglycemia,^44,45 these findings suggest that African immigrants with prediabetes may be relatively protected from the retinal neurodegeneration commonly associated with insulin resistance, due to their distinct metabolic phenotype. Our results are supported by Liu et al, who reported that retinal neurodegeneration is associated with severe insulin resistance diabetes, whereas choroidal microvascular damage is associated with severe insulin deficiency (from β-cell failure) diabetes.⁴⁶ Nonetheless, our results could be due to a small sample size (103 participants with prediabetes and 198 participants with normal glucose tolerance), although our sample size is larger than some of the reports demonstrating retinal thinning in prediabetes.^16,17

Furthermore, it is possible that retinal neurodegeneration in particular layers or subfields was not detected in our analysis using a categorical glucose tolerance predictor variable. A recent population-based study from the LIFE-Adult cohort demonstrated that HOMA2-IR, a marker of insulin resistance, was negatively associated with the thickness of the ONL and photoreceptor myoid zone, with findings confirmed in all ETDRS subfield-specific analyses.⁴⁷ Future studies could evaluate the association between insulin resistance and sensitivity markers and retinal layer and subfield-specific thickness to validate evidence that metabolic phenotypes, rather than glucose tolerance categories alone, may drive early retinal structural differences.

Overall, this study was strengthened by its observational cohort study design that allowed us to apply strict inclusion and exclusion criteria when enrolling participants, as well as the exclusion criteria based on image quality and pathology to ensure that only retinas without macular pathology, including retinopathy or ocular hypertension, were analyzed in this study. A challenge in this investigation is the use of a convenience sample. However, this is one of the largest studies of OCT images in Africans and the cohort appears to be representative of the African immigrant community, indicating that the study results could be generalized to African immigrants living outside of Washington D.C. Consistent with US Census Data, the cohort was majority male, mostly from West and East Africa, and arrived in the United States after 2000.⁴⁸ Furthermore, the cohort was large and heterogenous enough that we could detect that sickle cell trait was more common in immigrants from West and East Africa and that HbC trait occurred only in West Africa (Supplement Table 3).⁴⁹ Finally, the detection of glucose tolerance status was highly accurate, as both OGTT and HbA1c were done in all participants.

There are also limitations inherent to the study design. Although axial length is known to correlate with macular thickness, we could not adjust for this factor in our analysis as it was not measured in our participants. While we conducted a sensitivity analysis adjusting for refractive error, residual confounding by ocular biometry remains possible. Additionally, refractive error data were only present for approximately 26% of participants, which limits the generalizability of the dataset. Furthermore, the results for hemoglobin type are limited by a small sample size of type AC participants. Although we excluded eyes with glaucoma or ocular hypertension, retinopathy, macular edema, and retinal dystrophy, the study did not comprehensively ascertain all systemic neurologic or optic nerve disorders that may influence retinal cell layer thickness. Therefore, confounding from other causes of retinal neurodegeneration remains possible. While eyes with a clinical diagnosis of glaucoma or with an intraocular pressure greater than 21 mmHg were excluded, we did not systematically evaluate optic nerve head parameters such as cup-to-disc ratio, neuroretinal rim appearance, or automated visual fields. Therefore, undiagnosed early-stage or normal tension glaucoma may have been included in the analysis and influenced retinal cell layer thickness measurements. Finally, individual retinal layer thicknesses were derived using the Iowa Reference Algorithms and may not be directly interchangeable with measurements generated by native commercial segmentation software.

Conclusion

In conclusion, this study provides descriptive reference data on Cirrus-derived total macular thickness and factors associated with macular thickness in a cohort of African-born immigrants. We found that sex is associated with macular thickness in African immigrants, similar to other demographic groups. Macular OCT of African-born immigrants should be analyzed in the context of these associations, which may not be accounted for in existing normative databases. African immigrants with prediabetes do not show retinal thinning compared to those with normal glucose tolerance; this may be due to β-cell failure rather than insulin resistance. Additional population-based studies incorporating axial length measurements are needed to establish normative reference standards. Future studies could compare retinal thickness in various ETDRS subfields between African immigrants and other demographic groups, as well as explore any potential differential effects of β-cell failure and insulin resistance on the retina.

Funding Statement

This research was supported within National Institutes of Health by the intramural programs of NIDDK, NIMHD, and NEI.

Author Contributions

All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The author(s) report no conflicts of interest in this work.