Racial/ethnic differences in eligibility for asthma biologics among pediatric populations

Eric M. Wohlford, MD, PhD,* Peter F. Huang, BA,* Jennifer R. Elhawary, MS,* Lauren A. Millette, PhD, Maria G. Contreras, BS, Jonathan Witonsky, MD, C ecile T. J. Holweg, PhD, Sam S. Oh, PhD, MPH, Christine Lee, PharmD, PhD, Christine Merenda, MPH, RN, Ronald L. Rabin, MD, Richardae Araojo, PharmD, MS, Angel C. Y. Mak, PhD, Celeste S. Eng, BS, Donglei Hu, PhD, Scott Huntsman, MS, Michael A. LeNoir, MD, Jose R. Rodr ıguez-Santana, MD, FAAP, FCCP, Luisa N. Borrell, DDS, PhD, and Esteban G. Burchard, MD, MPH San Francisco, South San Francisco, and Oakland, Calif; Silver Spring, Md; Caguas, Puerto Rico; and New York, NY


Abbreviations used
AEC: Absolute eosinophil count GALA II: Genes-environments & Admixture in Latino Americans study ICS: Inhaled corticosteroid OR: Odds ratio SAGE: Study of African Americans, Asthma, Genes, & Environments SES: Socioeconomic status WBC: White blood cell prescribed asthma therapies. [5][6][7][8][9] Although the etiology of asthma is complex, resulting from the confluence of genetic, environmental, and sociocultural risk factors, the causal basis for these racial/ethnic health disparities remains elusive.
Asthma is a heterogeneous disease composed of multiple, sometimes overlapping, pathobiological and clinical subtypes. 10 Delineating these clinical subtypes is a challenge for clinicians, who must choose between different biologic therapies for their patients. Currently approved asthma biologics in the United States primarily target allergic or eosinophilic asthma. Blood parameters are used to determine eligibility and dosing regimens of these biologics, yet evidence shows that clinical blood profiles differ by race/ethnicity. 11, 12 The lack of population-based information on asthma therapeutic-associated blood parameters in racially/ethnically diverse pediatric populations can leave clinicians uncertain about the choice of biologic therapy for their patients. This is especially true for non-White patients, who have been left out of pulmonary and asthma-related clinical and biomedical research. 13 Having accurate, population-based information for racial/ethnic minority children is of critical concern given that minorities comprise 50% of children in the United States. 14 Other asthma subtypes such as neutrophilic, late-onset, and obesity-related asthma 10 do not have subtype-directed asthma therapies at this time. Thus, the study of asthma blood profiles may provide insight into population-specific asthma biology and help inform therapeutic management.
We hypothesized that clinical blood profiles and asthma subtypes are differentially associated with asthma outcomes across racial/ethnic populations. We also hypothesized that eligibility for asthma biologic therapies differs across populations. To address these hypotheses, we examined the association of blood parameters and clinical subtypes with asthma outcomes using case-control and case-only analyses in 2 richly phenotyped pediatric African American and Latino cohorts with and without asthma. In addition, we assessed population-specific eligibility for blood biomarker-informed biologic therapies in those with moderate to severe asthma.

Recruitment of parent study population
The Genes-environments & Admixture in Latino Americans study (GALA II) and the Study of African Americans, Asthma, Genes, & Environments (SAGE) are parallel case-control studies using similar protocols and questionnaires, previously described. 15 Briefly, GALA II recruited Hispanics/Latinos from 5 urban study centers across the mainland United States (Chicago, Ill; Bronx, NY; Houston, Tex; and San Francisco Bay Area, Calif) and Puerto Rico between 2006 and 2014 using a combination of community and clinic-based recruitment. SAGE recruited African Americans from the San Francisco Bay Area only. In both studies, participants were 8 to 21 years old at recruitment. Cases had physician-diagnosed asthma and asthma symptoms and/or asthma medication use within the last 2 years, with no history of other lung or chronic nonallergic illnesses. Healthy controls had no history of asthma or allergies, use of allergy medications, or symptoms of wheezing or shortness of breath during their lifetime. Control subjects were 1:1 frequency matched within each recruitment center by age (within 1 year). Case subjects and control subjects were recruited from similar geographic regions. Those in the third trimester of pregnancy, current smokers, and those with at least a 10 pack year smoking history were ineligible. Parents and grandparents of study participants must have self-identified as Hispanic/Latino or African American. At the time of recruitment, detailed clinical measures and biological specimens were collected, along with questionnaire-based information regarding additional social, environmental, and historical risk factors. The study protocols for both GALA II and SAGE were approved by the University of California San Francisco Human Research Protection Program Institutional Review Board. Detailed consenting procedures can be found in this article's Online Repository at www.jacionline.org.

Asthma outcomes
Asthma outcomes were defined as follows: asthma status (yes/no), asthma severity (moderate to severe/mild), asthma control (uncontrolled/controlled), and at least 1 asthma exacerbation in the year before recruitment (yes/no). Detailed definitions of asthma outcomes are provided in this article's Online Repository at www.jacionline.org.

Measurement of blood parameters
Serum total IgE was measured in our research lab from plasma in duplicate on a Phadia 100 detection system (ThermoFisher Scientific, Uppsala, Sweden). If both measurements were not within 10% concordance, a third measurement was assayed. White blood cell (WBC) counts were obtained from complete blood cell counts with differentials using commercially available and Clinical Laboratory Improvement Amendments-certified laboratories through Quest Diagnostics in the GALA II data set and through University of California, San Francisco Clinical Laboratories for SAGE. Serum total IgE and WBC counts were specified as continuous predictors. However, the distributions of serum total IgE, basophils, and eosinophils were highly right-skewed and were log-transformed for analysis purposes.

Defining asthma subtypes
Allergic asthma was defined as having asthma and any sensitization to aeroallergens by skin prick test. 16 Details on assessment of skin prick positivity can be found in this article's Online Repository at www.jacionline.org. Eosinophilic asthma was defined as having asthma and an absolute eosinophil count (AEC) greater than or equal to 150 c/mL (> _300 c/mL separately assessed in this article's Online Repository at www.jacionline.org). 17 These asthma subtypes were specified as yes or no. Overlap between subtypes was assessed only in African Americans because they were the only population in our study with an overlap between skin prick and WBC data.

Defining eligibility for blood biomarker-informed asthma biologic therapies
For allergic asthma therapy, dosing of anti-IgE therapy is based on pretreatment serum total IgE level. The US Food and Drug Administrationapproved dosing range for this therapy requires the level of pretreatment serum total IgE to be more than 30 kU/L and either less than 1300 kU/L for children 6 to 11 years old or less than 700 kU/L for those 12 years old. [18][19][20][21][22] For this eligibility analysis, allergic asthma was defined as having asthma and sensitization to at least 1 perennial aeroallergen (dust mite, dog, cat, cockroach, mouse, and rat) by skin prick test. Participants with moderate to severe, allergic asthma were considered eligible for anti-IgE therapy if they fell within their age-specific limits and ineligible otherwise. For eosinophilic asthma therapy, eligibility was assessed using a common clinical threshold for therapeutic use: AEC 150 c/mL. 17,[23][24][25][26] The age indication for most eosinophil-directed therapies is 12 years. [27][28][29] Therefore, participants must have been at least 12 years old, and were considered eligible if their AEC was equal to or above the threshold and ineligible otherwise.

Covariates
Covariates considered for this analysis were age, sex, obesity status, socioeconomic status (SES), and inhaled corticosteroid (ICS) use within 2 weeks of recruitment. Detailed descriptions are provided in this article's Online Repository at www.jacionline.org.

Study inclusion and exclusion criteria
Of the 5147 GALA II participants, those who identified themselves as Hispanic/Latino subgroups other than Mexican American or Puerto Rican or whose grandparents were not identified as the same race/ethnicity were excluded (N 5 957). Of the 1710 total SAGE participants, participants whose 4 grandparents did not all identify as African American were excluded (N 5 126). After combining both GALA II and SAGE participants, those missing both total IgE data and WBC count data (N 5 831), those who used oral steroids within 2 weeks of recruitment (N 5 40), and those missing obesity status (N 5 1165) were also excluded for an analytical sample of 3738 cases and controls. Analyses for asthma outcomes (severity, control, and exacerbations) included cases only (N 5 2743). Analyses investigating the proportion of each population eligible for various biologic therapies were restricted to participants with moderate to severe asthma (N 5 1917). A consort diagram is visualized in Fig E1 in

Statistical analysis
Descriptive statistics were calculated for both demographic and clinical characteristics for the total population and according to race/ethnicity. We used logistic regressions to test an interaction term of race/ethnicity with each exposure to examine the heterogeneity of their association with asthma outcomes in the total population. We then quantified these associations via odds ratios (ORs) and CI in each racial/ethnic group (African American, Mexican American, and Puerto Rican) adjusting for age, sex, obesity status, SES, and ICS use. Mexican American participants were excluded from WBC count analyses because of small sample size (N 5 33). Because of the bimodal distribution of basophils in Puerto Ricans (see Fig E2 in this article's Online Repository at www.jacionline.org), the association of basophils with asthma and asthma outcomes was analyzed only in African Americans. In addition, we examined the proportion of participants from each racial/ethnic group with moderate to severe asthma who would be ineligible for anti-IgE and eosinophilic asthma-directed therapies. We used chi-square tests of independence to examine whether the proportion of those ineligible differed between populations.
All statistical analyses were performed with R v.3.6.2 (R Foundation for Statistical Computing, Vienna, Austria).

Demographic characteristics
Baseline study population characteristics for the total population and stratified by race/ethnicity are presented in Table I. In our population (N 5 3738), 1275 self-identified as African American, 967 as Mexican American, and 1496 as Puerto Rican. Almost half of the study population were female (49.0%), and almost a third were classified as obese (31.9%). Mexican Americans were more likely to be obese compared with other racial/ ethnic groups (P < .001). In our study population, 2743 (73.4%) participants had asthma. Mexican Americans had the lowest proportion of poor asthma outcomes (P < .05). A higher proportion of Puerto Ricans had eosinophilic asthma compared with African Americans (P < .01), and proportions of those with allergic asthma differed significantly between racial/ethnic groups (P < .05; see Table E3 in this article's Online Repository at www.jacionline.org).

Distribution of blood parameters
Total IgE levels were significantly higher in Puerto Ricans compared with African Americans and Mexican Americans (P < .001). Basophil, eosinophil, monocyte, and neutrophil counts differed significantly between African Americans and Puerto Ricans (P < .001), as shown in Fig E2.
Higher eosinophil count was significantly associated with greater odds of asthma in African Americans and Puerto Ricans (OR, 1.55; 95% CI, 1.10-2.23, and OR, 1.66; 95% CI, 1.23-2.24, respectively), but associated with exacerbations (OR, 1.67; 95% CI, 1.18-2.38) in Puerto Ricans only. Higher lymphocyte counts were significantly associated with greater odds of asthma in African Americans only (OR, 1.05; 95% CI, 1.00-1.10), whereas higher monocyte counts were significantly associated with lower odds of asthma in Puerto Ricans only (OR, 0.84; 95% CI, 0.74-0.96). Neutrophils were not associated with asthma or asthma outcomes in African Americans or Puerto Ricans. All ORs and CIs for this analysis are provided in Table E4 in this article's Online Repository at www.jacionline.org.

Effect of asthma subtypes on asthma outcomes
We observed some associations between asthma subtypes and asthma outcomes in Puerto Ricans and Mexican Americans but not in African Americans (Fig 2). After controlling for age, sex, obesity status, SES, and ICS use, eosinophilic asthma was associated with worse asthma severity (OR, 1.96; 95% CI, 1.15-3.37), worse asthma control (OR, 1.74; 95% CI, 1.01-3.00), and exacerbations (OR, 1.81; 95% CI, 1.10-2.98) in Puerto Ricans only. Allergic asthma was associated with worse asthma control (OR, 1.49; 95% CI, 1.02-2.19) and exacerbations (OR, 1.82; 95% CI, 1.19-2.83) only in Mexican Americans. Heterogeneity was not observed for the associations between asthma subtypes and asthma-related outcomes by race/ethnicity (P > .05). All ORs P values for between-population comparisons are provided in Table E2. Q 1 , First quartile; Q 3 , third quartile. *All values shown are N values and percentages of the population with the exception of age, AEC, and IgE, which represent medians and first and third quartiles. Expressed as percentage of population with asthma. àNot analyzed because of small sample size. §Expressed as percentage of asthma cases with measured eosinophil count. Defined as eosinophil count > _150 c/mL. kExpressed as percentage of asthma cases with skin prick measurements. Defined as having positive aeroallergen skin prick test results.
FIG 1. Population-specific adjusted ORs for blood parameters on asthma outcomes. ORs adjusted for age, sex, obesity status, SES, and ICS use within 2 weeks of recruitment are plotted with their 95% CI on a log scale. Values shown are associated with a 1-log increase for eosinophils (c/mL) and IgE (kU/L). ORs indicate that the exposure is associated with a poor outcome (asthma/moderate to severe asthma/uncontrolled asthma/exacerbations within the past year). ORs for WBCs in Mexican American children not plotted because of small sample size. J ALLERGY CLIN IMMUNOL VOLUME 148, NUMBER 5 and CIs for this analysis are provided in Table E5 in this article's Online Repository at www.jacionline.org. Overlap of allergic and eosinophilic asthma in 122 African Americans is presented in Table E6 in this article's Online Repository at www.jacionline. org.

Eligibility for allergic and eosinophilic asthma biologics across populations
In our study, Puerto Ricans were significantly less likely than other groups to be eligible for anti-IgE therapy, but significantly more likely to qualify for eosinophilic asthma-directed therapies. Specifically, 17.2% of African Americans, 21.2% of Mexican Americans, and 31.4% of Puerto Ricans with moderate to severe, allergic asthma had pretreatment IgE either too low to qualify (<30 kU/L) or too high to recommend a dose for anti-IgE therapy (>700 kU/L for ages 121 years, >1300 kU/L for ages 6-11 years ; Fig 3, A). 20 Proportion plots for each pretreatment IgE threshold can be found in Fig E3 in this article's Online Repository at www.jacionline.org. We found that 51.3% of African Americans and 26.8% of Puerto Ricans age 12 years and older with moderate to severe asthma had AEC less than 150 c/mL (Fig 3, B) and would not qualify for eosinophilic asthmadirected therapies. A proportion plot for an AEC less than 300 c/mL cutoff can be found in Fig E4 in this article's Online Repository at www.jacionline.org. We additionally examined these proportions for those age 6 years and older, as is indicated for the eosinophilic asthma-directed therapy dupilumab, 29 and found that the proportions were nearly identical. Significant differences in population eligibility for anti-IgE therapy and eosinophilic asthma-directed therapy for age 12 years and older are presented in Table E7 in this article's Online Repository at www.jacionline.org.

DISCUSSION
Clinical blood profiles affect asthma outcomes and determine many clinical and therapeutic options. These profiles also differ by race/ethnicity. 11,12 We found that increased levels of serum total IgE were highly associated with asthma status across African American, Mexican American, and Puerto Rican children. Serum total IgE is a biomarker and biologic therapeutic target associated with allergic asthma and with decreased lung function in patients with asthma. 30 Previous research in children has associated IgE with asthma severity. 31,32 Within our Puerto Rican population, increased IgE levels were significantly associated with severe asthma, worse asthma control, and a history of asthma exacerbations. We did not observe the same pattern among African Americans, in whom increased IgE was only associated with severe asthma and poor asthma control, or Mexican Americans, in whom increased IgE was not associated with any of these outcomes.
Our findings of racial/ethnic-specific blood profiles and associations with asthma outcomes may suggest different asthma pathobiology or environmental effects in the populations studied. It has been previously shown that peripheral blood eosinophil counts are associated with eosinophilic asthma, as measured by airway eosinophils and asthma outcomes. 33,34 Although eosinophil count was significantly associated with asthma in both African American and Puerto Rican groups, only in our Puerto Rican population was higher eosinophil count associated with a history of asthma exacerbations. It is possible that Puerto Ricans and Population-specific adjusted OR for asthma subtypes on asthma outcomes. ORs adjusted for age, sex, obesity status, SES, and ICS use within 2 weeks of recruitment are plotted with their 95% CI on a log scale. Eosinophilic asthma defined as eosinophil count greater than or equal to 150 c/mL. Allergic asthma was defined as having any sensitization to aeroallergens by skin prick test. Values represent odds of the outcome for a given asthma subtype. ORs indicate that the exposure is associated with a poor outcome (moderate to severe asthma/uncontrolled asthma/exacerbations within the past year).
African Americans may have a different pathobiological basis for poor asthma outcomes, and that Puerto Ricans may benefit more than other racial/ethnic groups from eosinophil-directed biologic therapies. We note that baseline AEC was elevated in Puerto Ricans relative to African Americans regardless of asthma status. It has been reported that elevated baseline AEC is a predictor of treatment response to mepolizumab therapy for asthma. 35 Whether or not elevated baseline AEC predisposes the Puerto Rican population to eosinophilic disease is yet unknown.
The associations between clinically significant predictors and disease outcomes produced surprising results. Specifically, Mexican Americans were the only population studied with any significant associations between allergic asthma and worse asthma outcomes. In addition, eosinophilic asthma was significantly associated with worse asthma severity, control, and exacerbations in Puerto Ricans only, which suggests that eosinophil-directed biologic therapies may benefit Puerto Ricans more than other populations. Although these findings may be due to population-specific differences in social and environmental exposures or genetic ancestry proportions, further studies are needed to fully understand the effect that race/ethnicity has on allergic asthma outcomes.
Our study of children with moderate to severe asthma from different racial/ethnic groups provide novel insights into the therapeutic options available for these 3 populations of children. Overall, we found that a smaller proportion of Puerto Ricans would meet US Food and Drug Administration-approved thresholds for anti-IgE dosing using pretreatment IgE thresholds of less than 700 kU/L or less than 1300 kU/L, compared with African Americans or Mexican Americans. In contrast, for eosinophilic asthma, a smaller proportion of African Americans qualified for eosinophilic asthma-directed therapies based on pretreatment AEC greater than or equal to 150, when compared with Puerto Ricans. These differences highlight the need for future study of asthma biologic therapies in racially/ethnically diverse populations, which may identify clinical subgroups that would benefit most from targeted biologic therapies.
The underlying basis for the differences in peripheral blood parameters and racial/ethnic groups is likely multifactorial. 11 WBC counts are known to be affected by genetic ancestry, with African ancestry being associated with lower neutrophil counts and some elevations in eosinophil and other WBC counts, due to variation at the Duffy locus. 36,37 Herein, Puerto Ricans had higher average absolute neutrophil counts and elevated eosinophil counts relative to African Americans. On average, Puerto Ricans have lower proportions than African Americans of African ancestry. 38 Relative to Mexican Americans and other Latino populations, the Puerto Rican population has more African admixture and a lower proportion of Native American ancestry. 39 More broadly, the expressed blood parameters of various global populations may be influenced by their particular genetic background, related to history, geography, and other factors. An expanding list of genetic variants is known to play a role in WBC count differentials, which varies by race/ethnicity across studies. 37,40,41 Although environmental conditions 42 may affect eosinophil counts, and could relate to the observed differences between Puerto Ricans and African Americans, participants were enrolled in urban settings for both GALA II and SAGE, respectively. Although parasitic infections can increase AEC, these infections are now rare in Puerto Rico, consistent with the island's high human development index and Centers for Disease Control and Prevention traveler guidelines on the safety of drinking water 43,44 in urban areas, where all patients were enrolled. None of the populations in our study had testing for parasitic infections, and we cannot evaluate whether this could have affected AEC values. Environmental and hereditary factors have been previously associated with serum total IgE levels. 45,46 There are likely yet unknown genetic and environmental factors affecting the peripheral blood counts and IgE levels we observed, which may be clarified by future studies in racially/ethnically diverse populations.
One limitation of our study was that none of the participants were on biologic therapy for asthma, limiting our ability to make conclusions about blood profiles and associations with biologic J ALLERGY CLIN IMMUNOL VOLUME 148, NUMBER 5 therapeutic response. In this study, few participants had both skin prick and WBC data, which prevented us from assessing how overlapping subtypes affect asthma. In the real world, asthma subgroups are not mutually exclusive; patients can have both allergic and eosinophilic asthma. Further research is needed to examine whether having 1 or multiple subtypes influence an individual's risk for poor asthma outcomes or eligibility for asthma biologics. Furthermore, because of small samples sizes in some analyses as the result of unmeasured WBC count data, we cannot make definitive conclusions about racial/ethnic grouplevel differences in WBC counts and eligibility for eosinophilic asthma-directed therapy. We were also not able to determine which of the 40 patients taking oral corticosteroids in the previous 2 weeks had been on chronic versus acute oral corticosteroid therapy, and thus, could not assess the small proportion of patients who could qualify for dupilumab for oral corticosteroid-dependent asthma. Although research has suggested that differences in gut microbiomes may affect asthma outcomes, we were not able to assess the impact of microbiome differences in this study. Our study was also limited by the fact that we did not have a White population for comparison purposes. In addition, the blood profiles in our populations were measured from a single point in time, which meant we could not analyze how longitudinal changes in blood profiles may impact disease outcomes or eligibility for biologics. However, our WBC data are reliable, given that these measurements were obtained from Clinical Laboratory Improvement Amendmentscertified labs. In addition, there remains a lack of research in minority populations with asthma or lung disease. 13 This precluded us from being able to replicate our findings in an independent data set for these 3 minority populations. Regardless, our analyses included the largest gene-environment study of asthma in minority children to date.

Conclusions
Overall, our findings suggest that peripheral blood parameters are associated with asthma outcomes in African American, Mexican American, and Puerto Rican children. In the United States, available asthma biologic therapies target eosinophilic or allergic asthma. We found that a greater proportion of Puerto Rican participants had eosinophilic asthma compared with African Americans. Unlike Puerto Ricans, African Americans did not have associations between increased eosinophil counts and poor asthma outcomes including more severe asthma or asthma exacerbations. Further research is needed to determine whether these observed differences relate to differential treatment responses between racial/ethnic groups. Mexican Americans had the highest frequency of allergic asthma, which was uniquely associated with worse asthma control and exacerbations. Of participants with moderate to severe asthma, a higher proportion of African American and Mexican American participants than Puerto Rican participants met dosing criteria for treatment with anti-IgE therapy for allergic asthma, whereas a higher proportion of Puerto Rican than African American participants met criteria for eosinophilic asthma therapies. Selecting the correct biologic asthma therapy for a given patient remains a struggle for clinicians, which may be compounded in racial/ethnic minorities given the lack of clinical and biomedical research in these populations. Although biologic therapies represent a new dawn in asthma care, that dawn has not yet risen for all patients of different racial/ethnic backgrounds. It is critical that current and future biomarker-driven asthma therapeutics are studied in patients of diverse backgrounds to bring maximal benefits to all patients.

Consenting procedures
Questionnaires were administered to adult participants and parents of minors. All participants provided written consent to be in the study. Consent was obtained from all adult participants and parent/legal guardians of minor participants. The study protocols for both GALA II and SAGE were approved by the University of California, San Francisco Human Research Protection Program Institutional Review Board, and all institutions participating in recruitment obtained the appropriate approvals from their institutional review boards for recruitment-related activity.

Definition of asthma outcomes
Asthma status was recorded on the basis of a history of physician diagnosis and asthma symptoms, asthma medication use within the last 2 years, with no history of other lung or chronic nonallergic illnesses.
Report 3 of the National Asthma Education and Prevention Program (NAEPP) Guidelines E1 assesses asthma control and severity by both impairment and risk. Impairment is determined by the individuals' recall from the previous 2 to 4 weeks of symptoms, use of rescue medication, and in the case of asthma severity, also spirometry. Risk is broadly determined as exacerbations requiring oral steroids. We leveraged the validated Childhood Asthma Control Test E2 and the Asthma Control Questionnaire E3 to assess symptoms of impairment for both asthma control severity scores as outlined in the NAEPP guidelines. Asthma control was scored as controlled, not well controlled, or very poorly controlled on the basis of impairment or risk feature that displayed the worse control. For analysis, asthma control was dichotomized into controlled and uncontrolled (not well controlled or very poorly controlled).
Scoring severity was not as straightforward as asthma control because the NAEPP guidelines for assessing severity are outlined for individuals not currently taking control medication. Because most of our participants were on control medication, severity was instead assessed across impairment and treatment. Specifically, impairment features were categorized as mild persistent, moderate persistent, and severe persistent while treatment was categorized as mild intermittent, mild persistent, moderate persistent, and severe persistent on the basis of (1) the use of a short beta-agonist, (2) one of an inhaled steroid, leukotriene inhibitor, or theophylline tablet, (3) more than or a combination of an inhaled steroid, leukotriene inhibitor, or theophylline tablet, or (4) oral steroid use in the last week, respectively. The final severity score was determined to be the treatment score. However, if the participant had not well controlled or poorly controlled asthma but was scored as mild intermittent for treatment, their highest severity impairment feature score was used. In addition, if treatment was scored as mild or moderate persistent, yet the participant had not well controlled or poorly controlled asthma, their severity score was bumped to moderate and severe persistent, respectively. This bump-up of severity from asthma control is consistent with the NAEPP stepwise approach for managing asthma. For analysis, severity was dichotomized into mild (mild intermittent or mild persistent) and severe (moderate or severe persistent).
Asthma exacerbations were scored on the basis of definition set by the American Thoracic Society and European Respiratory Society. E4 Briefly, we assigned points for each reported hospitalization (1), emergency department visit (1), and oral steroid use (1 point was given if the participant used oral steroids in the 12 months before recruitment). Points were summed to derive a composite score. To avoid double counting for oral steroid prescriptions, a point was given to those who used any oral steroid in the last 12 months only among those participants who did not report a history of emergency visits or hospitalizations for asthma. For analysis, exacerbations were dichotomized into none or at least 1 exacerbation in the last 12 months.

Detailed description of covariate data
Age at time of recruitment was calculated using the difference between recruitment date and the participants' birth date.
Sex was self-reported. Body mass index (BMI) values were quantified using standardized methodologies with a clinical scale and a stadiometer. For participants younger than 20 years, BMI percentiles were calculated using the recommended Centers for Disease Control and Prevention sex-age curves. Conversely, for participants older than 20 years, BMI was calculated using the standard formula, height (kg)/ weight (m) 2 . Raw BMI and BMI percentiles were used to specify the following categories: underweight, normal, overweight, and obese using the Centers for Disease Control and Prevention guidelines. E5 BMI categories were then collapsed into obese and nonobese for analysis.
SES was derived from a combination of mother's education level, insurance status, and household income weighted by region of recruitment. Each component of the SES was scored on a 3-point scale, with 1 being lowest and 3 being the highest. Mother's education was broken down by less than high school (1), high school diploma or equivalent (2), and greater than high school (3). Insurance status was broken down by no insurance (1), insurance through the government (2), and insurance through self, a family member, or an employer (3). Household income was split into tertiles, with recruitment region acting as a weight for income level. The 3 components were then averaged and partitioned into a 3-level categorical variable (high, medium, low) using tertile cutoff points. SES was adjusted for in our models as a 3-level ordinal variable.

Assessment of aeroallergen sensitization
Skin testing was performed on the volar aspect of the forearm, using the Multi-Test II Device (Lincoln Diagnostics, Decatur, Ill) for a panel of 14 aeroallergens including dust mites (Dermatophygoides pteronyssinus and Dermatophygoides farinae), pets (dog and cat), pests (mix of German and American cockroach, mouse epithelium, and rat epithelium), trees (oak and olive/elm), grasses (perennial ryegrass/timothy and short ragweed), and molds (Alternaria tenuis, Hormodendrum cladosporioides, and Aspergillus mix). Any allergen with a mean wheal diameter of at least 3 mm greater than the mean wheal diameter of the saline control and that had a flare greater than the wheal was considered positive. In addition, subjects with tests considered invalid (having the mean wheal diameter of the histamine <3 mm greater than the mean wheal diameter of the saline control) were removed from the study. A score of 1 or 0 was given depending on whether the individual experienced at least 1 positive allergen test result or none.      Total IgE  1  1267  966  1321  3554  WBCs  2  251  -471  722  Asthma outcomes à §  Total IgE  1, 3  891  636  1092  2619  WBCs  2, 3  147  -339  486  Eosinophilic asthma  2, 3  147  -339  486  Allergic asthma  3, 4  736  580  443  1759 Sample sizes for logistic regression models stratified by race/ethnicity. Inclusion criteria key: (1) measured IgE, (2) measured WBCs, (3) asthma, (4) a valid aeroallergen skin prick test, (5) sensitization to at least 1 perennial aeroallergen by skin prick test, (6) moderate to severe asthma, (7) age > _ 12 y, and (8) age 6-11 y. *Case-control analysis. Case-only analysis. àOutcomes: asthma severity, asthma control, and history of exacerbations.
§A total of 12 participants with asthma were missing asthma severity data.    Adjusted ORs (95% CI) of the association between commonly used asthma subtypes and asthma outcomes in pediatric African Americans, Mexican Americans, and Puerto Ricans. Eosinophilic asthma untested in Mexican Americans because of small sample size (n 5 11). Eosinophilic asthma defined as eosinophil count > _150 c/mL and separately as eosinophil count > _300 c/mL. Allergic asthma was defined as having any sensitization to aeroallergens by skin prick test. ORs indicate that the exposure is associated with a poor outcome (asthma/moderate to severe asthma/uncontrolled asthma/exacerbations within the past year). Significant associations (P < .05) are denoted in boldface. Associations approaching significance (.05 < P < .01) are denoted in italics. Overlap of allergic and eosinophilic asthma in 122 African Americans. Eosinophilic asthma defined as eosinophil count > _150 c/mL and separately as eosinophil count > _300 c/mL. Allergic asthma was defined as having any sensitization aeroallergens by skin prick test. Overlap not assessed in Mexican Americans or Puerto Ricans due to lack of skin prick and WBC count data overlap. .09 12 y and older Eosinophil count <300 c/mL -<.001 -12 y and older Eosinophil count <150 c/mL -<.001 -12 y and older x 2 comparisons for biologic eligibility across populations. Participants had moderate to severe asthma. Sensitization to at least 1 perennial aeroallergen (dust mite, dog, cat, cockroach, mouse, and rat) by skin prick test was required for inclusion in anti-IgE analyses. Significant differences (P < .05) are denoted in boldface. AA, African American; MX, Mexican American; PR, Puerto Rican. *Combined anti-IgE compared ineligibility across all age groups considering both the lower bound (30 kU/L) and the upper bound (1300 kU/L for ages 6-11 y, 700 kU/L for ages 12 y and older). Not analyzed because of small sample size.