Clinical features and treatment response to topical steroids in ethnic and racial minority patients with eosinophilic esophagitis

Abstract

Introduction:

Differences in EoE presentation and outcomes by ethnicity or race remain understudied. We aimed to determine whether EoE patients of Hispanic/Latinx ethnicity or non-White race have differences in presentation at diagnosis or response to topical corticosteroid (tCS) treatment.

Methods:

This retrospective cohort study included subjects of any age with a new diagnosis of EoE and documentation of ethnicity or race. For those who had treatment with tCS and follow-up endoscopy/biopsy, we assessed histologic response (<15 eosinophils/hpf), global symptom response, and endoscopic responses. Hispanic EoE patients were compared to non-Hispanics at baseline, and before and after treatment. The same analyses were repeated for White vs non-Whites.

Results:

Of 1026 EoE patients with ethnicity data, just 23 (2%) were Hispanic. Most clinical features at presentation were similar to non-Hispanic EoE patients but histologic response to tCS was numerically lower (38% vs 57%). Non-White EoE patients (13%) were younger at diagnosis, had less insurance, lower zip code-level income, shorter symptom duration, more vomiting, less dysphagia and food impaction, fewer typical endoscopic features, and less dilation. Of 475 patients with race data treated with tCS, non-Whites had a significantly lower histologic response rate (41% vs 59%; p=0.01), and odds of histologic response remained lower after controlling for potential confounders (aOR 0.40, 95%CI: 0.19-0.87).

Discussion:

Few EoE patients at our center were Hispanic, and they had similar clinical presentations as non-Hispanics. The non-White EoE group was larger, and presentation was less dysphagia-specific. Non-White patients were also less than half as likely to respond to tCS.

Introduction

Eosinophilic esophagitis (EoE) is a chronic allergic condition of the esophagus characterized by symptoms of esophageal dysfunction and eosinophilic infiltration of the esophageal epithelium.(1) While EoE is still considered a rare disease, with a prevalence of approximately 1/2000, the frequency has been increasing and outpacing any increased recognition.(2, 3) It is likely this evolving epidemiology is explained by changing environmental exposures that interact with an impaired esophageal barrier to instigate a Th2 allergic response.(4, 5) The presentation of EoE can be heterogenous, with multiple phenotypes and endotypes described.(6, 7)

EoE can affect patients of any age, sex, race, or ethnicity. However, most studies of EoE show that the majority of patients are White, with proportions of more than 90% in some studies.(2, 8–13) It remains unclear whether those high proportions reflect the specific population composition at single centers or disparities in care. Our prior work showed that at centers located in areas where there were more non-White patients, the proportion of non-White EoE patients were higher.(14) However, other reports find that follow-up after food impaction is lower among non-Whites,(15) and that initial EoE presentation in non-Whites may not involve the classic dysphagia-predominant symptoms in adolescents and adults.(9, 14, 16–22) In addition, these prior studies have tended to focus on Black patients when assessing non-White EoE patients, with very few investigations into Hispanic/Latinx ethnicities. There are similarly few studies that assess treatment response in non-White EoE patients.

Because differences in EoE presentation and treatment response by ethnic or racial minority status remains understudied, we aimed to determine whether EoE patients of Hispanic/Latinx ethnicity or non-White race have differences in presentation at diagnosis or response to topical corticosteroid (tCS) treatment.

Methods

We conducted a retrospective cohort study of the University of North Carolina (UNC) EoE Clinicopathologic database, a data source that has been previously described.(23–26) In brief, subjects of any age with a new diagnosis of EoE, as per guidelines at the time of diagnosis,(1, 27–29) were included. In addition to having a peak esophageal eosinophil count of at least 15 eosinophils per high-power field (eos/hpf), patients had to have symptoms attributable to the esophagus and no competing cause of eosinophilia. For the present study, we utilized the overall population (spanning cases diagnosed over the past two decades at UNC) for the main analysis for ethnicity, and also pre-specified two sub-populations for additional study. First was the EoE population diagnosed from 2009 forward, which was included in the racial (White vs non-White) analysis, so as to not overlap with our previously reported data on race in EoE.(14) Second was the sub-population who were treated with a tCS after their diagnosis and who had a follow-up endoscopy with biopsy to assess treatment response. At our center, tCS for EoE are prescribed per clinical protocol and can include either swallowed fluticasone (typically from an inhaler) or budesonide (mixed into a slurry from the aqueous formulation), with dosing dependent on patient age and size. The initial course is usually 2-3 months, and then response is assessed on follow-up endoscopy and biopsy.(30, 31)

Data for our cohort were extracted from electronic medical records. This included patient demographics, health insurance, comorbidities, symptoms, concomitant atopic conditions, and procedural data for both the baseline and post-treatment endoscopies. Ethnicity (Hispanic/non-Hispanic) and race (White, Black, Asian, Native American/Alaskan, and other) were recorded as documented in the chart. Give that this was a retrospective cohort study, we could not use patient self-report with the most preferred updated terms,(32) so relied on the terminology recorded in the record. In addition, to assess for social determinants of health, we examined the median household income and highest education level achieved by zipcode in North Carolina from 2021 census data (data.census.gov); we did not have access to patient level income or education level data.

For the subset who had treatment with a tCS and a follow-up endoscopy/biopsy, we extracted data on the specific tCS type and dosing. We defined histologic response as <15 eosinophils/hpf, but also assessed additional thresholds of ≤6 eos/hpf and <1 eos/hpf.(33–35) For endoscopic findings, we collected data on the EoE Endoscopic Reference Score (EREFS) when available (from 2013 forward) to quantify the severity of exudates, rings, edema, furrows, and strictures (score range 0-9, with higher scores indicating increased severity).(36, 37) However, to provide a of endoscopic activity for all patients in the study, we also calculated an Endoscopic Severity Score (ESS) as data for this were available for all patients. This was the sum of the presence (1 point) or absence (0 points) of edema, rings, exudates, furrows, or stricture (range: 0-5, with higher scores indicating worsening severity), and such a scoring option has recently been assessed.(38) Based on our study design, validated patient-reported outcomes were not available in our database. We therefore recorded a global symptom response, as documented by the provider in the chart, an outcome we have previously used successfully.(25, 30, 34) We similarly recorded a global endoscopic response as reported by the performing endoscopist.

Statistical analysis was done using STATA (version 12; College Station, TX). Baseline and post-treatment characteristics of the study population were summarized with descriptive statistics. Hispanic EoE patients were compared to non-Hispanics at baseline, and before and after treatment. For this set of analyses, the two different ethnic groups were compared with 2 sample t-tests for means and chi-square for proportions for baseline data. For the pre-post treatment analysis within study groups, paired t-tests were used for means and McNemar’s test was used for proportions. The same set of analyses were then repeated for White vs non-White patients. Finally, we performed multivariate logistic regression to test for independent association of baseline characteristics between White vs non-White EoE patients; there were too few EoE patients of Hispanic ethnicity for modelling. Covariates included in the model were those of clinical interest, as well as those that differed between groups on bivariate analysis, including median income by zip code as proxies for social determinants of health; given that the addition of highest education level attained by zip code did not change the results of the model, this covariate was not included in the final model. This study was approved by the UNC Instituational Review Board.

Results

EoE and ethnicity – baseline characteristics and treatment responses

Of 1026 EoE patients with ethnicity data in our records, just 23 (2%) were Hispanic. When comparing the baseline features of EoE patients by ethnicity, Hispanic EoE patients tended to be younger (22.6 vs 30.0 years; p=0.08) with a higher proportion of children <18 years (57% vs 34%; p=0.03), were less likely to be White (37% vs 88%; p<0.001) or have health insurance (65% vs 84%; p=0.02), more likely to speak a non-English language (23% vs <1%; p<0.001), and had a numerically lower zip code level income ($65,724 vs $75,598; p=0.34) (Table 1). However, zip code level education attained, atopic conditions, baseline symptoms, baseline endoscopic features, and baseline eosinophil counts were largely similar by ethnicity (Table 1).

Table 1.

Comparison of baseline characteristics at diagnosis between non-Hispanic and Hispanic EoE patients

	Non-Hispanic (n = 1003)	Hispanic (n = 23)	p
Age at diagnosis (mean years ± SD)	30.0 ± 20.0	22.6 ± 20.6	0.08
Children <18 year (n, %)	344 (34)	13 (57)	0.03
Male (n, %)	671 (67)	19 (83)	0.11
Race (n, %)			< 0.001
White	880 (88)	8 (37)
Black	85 (9)	1 (5)
Asian	10 (1)	1 (5)
Native American/Alaskan	10 (1)	0 (0)
Other	14 (1)	12 (55)
BMI (mean kg/m2 ± SD)	24.8 ± 7.1	22.1 ± 6.3	0.08
Non-English language (n, %)	1 (<1)	5 (23)	< 0.001
Any insurance (n, %)†	837 (84)	15 (65)	0.02
Insurance type (n, %)			0.001
Private	620 (62)	8 (35)
Medicare	28 (3)	0 (0)
Medicaid	109 (11)	4 (4)
Military	27 (3)	0 (0)
Private + Medicare	27 (3)	0 (0)
Mediare + Medicaid	3 (<1)	0 (0)
Military + any other	8 (1)	2 (9)
Private + Medicaid	15 (2)	1 (4)
Income by zipcode (median dollars, IQR)	75,598 (57066-86692)	65724 (55090-85714)	0.34
Highest education level achieved (percent by zip code)
Less than high school (%)	9	10	0.36
High school degree (%)	21	22	0.68
Some college or associate’s Degree (%)	30	31	0.58
Bachelor’s degree (%)	23	22	0.41
Graduate degree (%)	17	16	0.65
Any atopic condition (n, %)	557 (60)	14 (61)	0.91
Allergic rhinitis	427 (44)	10 (43)	0.99
Asthma	246 (25)	8 (35)	0.29
Eczema	147 (15)	6 (26)	0.13
Food allergy	272 (29)	7 (30)	0.86
Symptom length prior to diagnosis (mean years ± SD)	7.5 ± 8.4	8.1 ± 10.9	0.75
Symptoms (n, %)
Dysphagia	757 (76)	15 (65)	0.23
Food impaction	335 (34)	5 (22)	0.22
Heartburn	362 (37)	7 (30)	0.53
Chest pain	107 (11)	2 (9)	0.74
Abdominal pain	169 (17)	3 (3)	0.61
Nausea	95 (10)	1 (4)	0.39
Vomiting	225 (23)	10 (43)	0.02
Endoscopic findings (n, %)
Exudates	408 (41)	11 (48)	0.50
Rings	519 (52)	7 (30)	0.04
Edema	399 (40)	10 (43)	0.73
Furrows	674 (67)	14 (61)	0.51
Stricture	293 (29)	6 (26)	0.74
Narrowing	162 (16)	3 (13)	0.68
Crepe-paper mucosa	40 (4)	1 (4)	0.93
Dilation	308 (31)	4 (17)	0.17
Total EREFS (mean ± SD)*	3.7 ± 2.0	3.4 ± 1.6	0.61
Total ESS (mean ± SD)*	2.3 ± 1.6	2.1 ± 1.6	0.52
Peak eosinophil count (mean eos/hpf ± SD)	64.6 ± 44.4	57.5 ± 41.1	0.44

^†Insurance data unavailable for n=4 patients

^*EREFS data available for n=480; ESS = endoscopic severity score, data available for all subjects

There were 466 patients with ethnicity data who received tCS therapy and had follow-up endoscopy and biopsy in our study population, but only 8 were Hispanic. Hispanic EoE patients had numerically higher eosinophil counts (47.0 vs 24.5; p=0.09) and numerically lower histologic response at <15 eos/hpf (38% vs 57%; p=0.27) post-treatment, but other post-treatment features were similar (Supplemental Table 1).

EoE and race – baseline characteristics and treatment responses

There were 885 patients in our database from 2009 forward who had race data available. When comparing EoE patients in terms of race, non-White patients (n=114; 13%) had many differences in presentation compared to White patients. Non-White EoE patients were younger (21.3 vs 31.9 years; p<0.001), less likely to have health insurance (75% vs 88%; p<0.001), had a lower zip code level income ($61,304 vs $74,737; p<0.001), had a lower zip code level of education attained (for example, 21% vs 24% having a Bachelor’s degree; p<0.001), had a shorter symptom duration prior to diagnosis (4.5 vs 8.1 years; p<0.001) and a lower BMI (22.1 vs 22.4; p=0.003), were less likely to have dysphagia (60% vs 80%; p<0.001) and food impaction (19% vs 37%; p<0.001), and more likely to have symptoms of vomiting (41% vs 20%; p<0.001 (Table 2). With the exception of exudates, non-White EoE patients had fewer typical endoscopic features, including less frequent strictures (19% vs 32%; p=0.003), narrowing (7% vs 19%; p=0.002), and dilation (19% vs 34%; p=0.002) (Table 2). After multivariate analyses, younger age (adjusted odds ratio [aOR] 0.98 for each 1 year decrease in age, 95% CI: 0.96-0.99), symptoms of vomiting (aOR 1.94, 95% CI: 1.23-3.05), and less frequent furrows on endoscopy (aOR 0.41, 95% CI: 0.27-0.62) remained independently associated with non-White race.

Table 2:

Comparison of baseline characteristics at diagnosis between White and non-White EoE patients, using data from 2009 to present

	White (n = 771)	Non-White† (n = 114)	p
Age at diagnosis (mean years ± SD)	31.9 ± 19.4	21.3 ± 21.4	< 0.001
Children <18 year (n, %)	220 (29)	69 (61)	< 0.001
Male (n, %)	501 (65)	82 (72)	0.14
Non-Hispanic (n, %)	753 (99)	99 (90)	< 0.001
BMI (mean kg/m2 ± SD)	25.1 ± 6.7	22.4 ± 8.3	0.003
Non-English language (n, %)	0 (0)	5 (4)	< 0.001
Any insurance (n, %)	673 (88)	85 (75)	< 0.001
Insurance type (n, %)#			< 0.001
Private	526 (68)	39 (35)
Medicare	19 (2)	3 (3)
Medicaid	65 (8)	34 (30)
Military	25 (3)	2 (2)
Private + Medicare	19 (2)	1 (1)
Mediare + Medicaid	3 (<1)	0 (0)
Military + any other	8 (1)	2 (2)
Private + Medicaid	8 (1)	4 (4)
Income by zipcode (median dollars, IQR)	74737 (58253-87598)	61304 (51431-82896)	< 0.001
Highest education level achieved (percent by zip code)
Less than high school (%)	8	10	<0.001
High school degree (%)	20	24	< 0.001
Some college or associate’s Degree (%)	30	31	0.01
Bachelor’s degree (%)	24	21	< 0.001
Graduate degree (%)	18	14	< 0.001
Any atopic condition (n, %)	454 (59)	67 (59)	0.91
Allergic rhinitis	344 (45)	47 (41)	0.45
Asthma	190 (25)	33 (29)	0.35
Eczema	116 (15)	24 (21)	0.10
Food allergy	221 (29)	39 (34)	0.25
Symptom length prior to diagnosis (mean years ± SD)	8.1 ± 8.8	4.5 ± 5.8	< 0.001
Symptoms (n, %)
Dysphagia	615 (80)	68 (60)	< 0.001
Food impaction	283 (37)	22 (19)	< 0.001
Heartburn	268 (35)	45 (40)	0.31
Chest pain	87 (11)	9 (8)	0.28
Abdominal pain	124 (16)	22 (19)	0.38
Nausea	68 (9)	9 (8)	0.86
Vomiting	155 (20)	46 (41)	< 0.001
Endoscopic findings (n, %)
Exudates	357 (46)	44 (39)	0.14
Rings	436 (57)	37 (33)	< 0.001
Edema	362 (47)	37 (33)	0.004
Furrows	588 (76)	63 (56)	< 0.001
Stricture	250 (32)	21 (19)	0.003
Narrowing	148 (19)	8 (7)	0.002
Crepe-paper mucosa	28 (4)	3 (3)	0.60
Dilation	260 (34)	22 (19)	0.002
Total EREFS (mean ± SD)*	3.7 ± 1.9	3.4 ± 2.0	0.32
Total ESS (mean ± SD)*	2.6 ± 1.5	1.8 ± 1.5	< 0.001
Peak eosinophil count (mean eos/hpf ± SD)	64.2 ± 44.4	64.8 ± 43.1	0.90

^#Insurance data unavailable for n=3 patients

^*EREFS data available for n=483; ESS = endoscopic severity score, data available for all subjects

^†Includes people of Black, Asian, Native American/Alaskan, and other races

There were 475 patients with race data who were treated with tCS, including 49 non-Whites. Compared with the White EoE patients, non-Whites had a significantly lower histologic response rates at the 15 eos/hpf (41% vs 59%; p=0.01) and 6 eos/hpf (27% vs 52%; p=0.04) thresholds (Table 3). Accordingly, the post-treatment peak eosinophil count was higher for non-Whites (39.0 vs 22.5 eos/hpf; p=0.003) and there was less endoscopic response (54% vs 73%; p=0.006). The unadjusted odds ratio for histologic response for non-Whites at the <15 eos/hpf level was 0.48 (95% CI: 0.26-0.86). After controlling for age, BMI, insurance, median income at the zip code level, symptom length prior to diagnosis, total steroid dose, and whether dilation was performed, non-Whites remained less than half as likely to have histologic response (aOR 0.40, 95% CI: 0.19-0.87); zip code level of education attained did not impact this model and was not included in the final version (Supplemental Table 2).

Table 3.

Treatment and response data compared between White and non-White EoE patients

	White (n = 426)	Non-White (n = 49)	p
Type of steroid used (n, %)			0.06
Fluticasone	140 (33)	8 (16)
Budesonide	285 (67)	41 (84)
Mean steroid dose (mcg ± SD)	1755 ± 711	1415 ± 635	0.002
Symptom response (n, %)*	130 (78)	9 (64)	0.27
Post-treatment peak eosinophil count (mean eos/hpf ± SD)	22.5 ± 34.3	39.0 ± 49.8	0.003
p value vs baseline	< 0.001	0.02
Histologic response (n, %)
<15 eos/hpf	253 (59)	20 (41)	0.01
≤6 eos/hpf	223 (52)	18 (27)	0.04
<1 eos/hpf	131 (31)	12 (24)	0.37
Post-treatment endoscopic findings (n, %)
Normal	84 (20)	15 (31)	0.08
Exudates	96 (23)	16 (33)	0.11
Rings	206 (49)	7 (15)	< 0.001
Edema	114 (27)	17 (35)	0.24
Furrows	192 (46)	20 (42)	0.58
Stricture	145 (35)	8 (17)	0.01
Narrowing	80 (19)	1 (2)	0.003
Crepe-paper mucosa	3 (1)	1 (2)	0.33
Dilation	139 (33)	8 (19)	0.04
Candida	32 (8)	2 (4)	0.38
Endoscopic response (n, %)	311 (73)	26 (54)	0.006
Post-treatment endoscopic severity (mean scores ± SD)
ERFES**	2.3 ± 1.9	2.2 ± 2.0	0.83
p value vs baseline	< 0.001	0.008
ESS	1.8 ± 1.5	1.4 ± 1.5	0.09
p value vs baseline	< 0.001	0.13

^*available for 168 and 14;

^**available for 238 and 16

Discussion

Though the typical presentation of EoE is often considered to be solid food dysphagia or food impaction in younger males who are White, EoE is frequently heterogeneous and can present with a variety of phenotypes and endotypes.(2, 6, 7) There are relatively few investigations into presentation of EoE by race and ethnicity, and fewer regarding treatment outcomes in these subgroups. In this retrospective cohort study of a large number of EoE patients, we found that only a small proportion (2%) were of Hispanic/Latinx ethnicity, and while this population was younger, less likely to speak English, and less likely to have health insurance than non-Hispanic EoE patients, most other clinical and endoscopic features were similar at baseline. In contrast, we had a larger non-White population (13%) and this sub-group was younger and more likely to have symptoms of vomiting at diagnosis, less likely to have the dysphagia/food impaction presentation and typical endoscopic findings of EoE. Moreover, we report for the first time that non-Whites with EoE are less likely to be histologic responders to topical steroids, even after controlling for other factors that might explain non-response including health insurance, median income by zip code (as a proxy for a social determinant of health), steroid doses, age, length of symptoms prior to diagnosis, BMI, and need for dilation. Overall, our findings suggest that when non-White patients are assessed for GI symptoms, EoE should be on the differential diagnosis even in the absence of dysphagia-predominant symptoms, and that a tCS treatment option should be discussed in detail. The lower response rate raises the question of whether unmeasured aspects of health disparities could adversely impact treatment outcomes, though it is interesting that that the lower response rate remained even after controlling for zip code level income data, and addition of zip code levels of education did not impact the model. However, the limitation of these metrics is that they may not represent true income and education at the patient level.

There have been only a few prior studies of ethnicity in EoE. In one retrospective study conducted by Yu and colleagues at institutions in Los Angeles and El Paso chosen specifically because they served a large Hispanic population, patients undergoing endoscopy were assessed for EoE.(10) Similar to our results, they found that EoE was less common in Hispanic patients (3.2% vs 9.4% in Los Angeles and 0.96% vs 7.26% in El Paso), but because of the small number of patients identified, detailed comparisons between the groups was not possible. In a similarly designed study, García-Compeán and colleagues assessed rates of EoE detected during food impaction at their center in Mexico, but could only confirm the diagnosis in 2 of 36 patients (though only 17 had biopsies).(17) In several other studies, the numbers of patients identified of Hispanic ethnicity was also low.(8, 20, 39) Other previous studies examining racial differences in EoE presentation most often compared White to non-White EoE patients.(9, 14, 16, 19, 20, 22, 39) While many studies found a low prevalence of non-White patients with EoE, this seemed to depend on where the study was conducted.(14, 22) In the largest of these studies spanning 5 centers, 17% of EoE patients were non-White, and similar to our study, they tended to have less dysphagia or food impaction and fewer typical endoscopic findings of EoE.(9) These findings have been further corroborated in other studies,(14, 16, 19, 21, 22, 40) strongly suggesting that some of the lower EoE prevalence in non-White or Hispanic/Latinx patients may be due to detection bias from “atypical” EoE presentations, though recent investigations have delved into potential genetic or microbial factors.(18, 41)

Our findings of decreased treatment response to tCS in non-White EoE patients is somewhat more difficult to contextualize, as we are not aware of other studies specifically investigating this question. Prediction of treatment response related to tCS is an area of active research. Non-response has been associated with esophageal dilation,(42–44) severe esophageal narrowing,(45) increased BMI,(30) younger age,(31, 44) and molecular features.(46, 47) On our analysis, the lower response rate in non-Whites persisted despite adjusting for these other known factors related to tCS response. Our study design does not allow us to fully elucidate the reason that non-White and possibly Hispanic patients with EoE would have lower histologic response rates to tCS. Future studies would need to examine if this could be due to social determinants, other health-related discrepancies, disease-specific factors, or issues with the obtaining or using the medications themselves.

There are several limitation of this study to acknowledge. First, this is a single center study, so results may not be generalizable to all practice settings, especially given the small numbers in some of the race and ethnicity groups. Nevertheless, the patients included in this database were newly diagnosed with EoE at our center, not prevalent cases referred in for difficult-to-treat disease. Second, because this was a retrospective study, we had to rely on reports of ethnicity and race from the chart, so this could contain inaccuracies compared to patient self-report, and we are not able to report the most up to date racial categorizations or assess patients of multiple races.(32) It is a limitation to conduct the analysis as “White vs non-White”, but we opted for this approach to include as many patients in the analysis, as there were too few in some race subgroups to be analyzed independently. Future work should aim to overrepresent these other racial groups. We also could not assess tCS adherence (this was not routinely recorded in the chart) or assess prescription refills as a proxy for adherence, and if non-White patients were less likely to be adherent to treatment, this could be one explanation for lower response rates. In addition, symptom data were collected with non-validated instruments so the overall symptom response rates should be interpreted with caution. Strengths of the study included the large and well-characterized patient cohort that allowed comparisons based on ethnicity as well as race, systematic collection of granular patient-level data, consistent clinical protocols for tCS treatment and follow-up, and objective histologic response data.

In conclusion, in this retrospective cohort study assessing a large population of newly diagnosed EoE patients, only a small proportion (2%) were Hispanic. Though younger, they had generally similar clinical presentations as non-Hispanic EoE patients. Notably, they were less likely to have health insurance and speak English, which could impact on their care. While treatment response in Hispanic EoE patients tended to be numerically lower than in non-Hispanics, this assessment was limited by a small sample size. In contrast, the non-White EoE group was larger (13%), and presentation was less dysphagia-specific, with younger age, symptoms of vomiting, and less frequent esophageal furrows independently associated with non-White race. In addition, non-White patients had a lower histologic response to tCS, with less than half the odds of response compared to Whites, and this persisted after accounting for differences in presentation and other features previously associated with non-response. Therefore, the diagnosis of EoE should be actively considered in all patients who present with upper GI symptoms, regardless of ethnicity or race. Future research should focus on potential disparities in care and other potential causes that could lead to the lower histologic response rate to tCS in non-White EoE patients.

Study highlights.

WHAT IS KNOWN

• Some differences in EoE presentation and outcomes by race have been identified but few studies have examined ethnicity.

• Treatment outcomes by ethnicity or race in EoE have not been extensively described.

WHAT IS NEW HERE

• Only 2% of our large EoE cohort were Hispanic, though presentation at baseline was generally similar between Hispanic and non-Hispanic EoE patients.

• In the 13% of EoE patients who were non-White, presentation was less dysphagia-specific and there were fewer typical endoscopic features.

• Non-White EoE patients had less than half the histologic treatment response to topical corticosteroids compared to White EoE patients even after controlling for clinical factors and proxies for social determinants of health.