Clinical significance of eosinophilia in HIV-infected individuals

Clinical Significance of Eosinophilia Individuals

in HIV-infected

Daniel J. Skied, MD, Philip Keiser, MD, Dallas, Texas

PURPOSE:

peripheral individuals.

To assess the clinical significance eosinophilia in HIV-infected

E

of

osinophilia can be seen with several medical conditions including certain neoplasms, allergic diseases, various types of dermatitis, adrenal insufficiency, parasitic infections, drug react,ions, certain collagen vascular diseases, and the hypereosinophilic syndromes.‘~2 Patients with HIV/AIDS frequently have peripheral blood eosinophilia,3 ;and often have various concomitant conditions associated with eosinophilia. AIDS patients have a high incidence of allergic reactions,” and a high incidence of parasitic infections.5 In addition, studies have indicated that 40% to 92% of AIDS patients have evidence of adrenalitis at autopsy.6 Patients with AIDS have a high incidence of several malignancies, which can be associated with eosinophilia. Finally, Stuphylococcus aureus skin and nares colonization in HIV has been associated with eosinophilia.7 Despite the frequent finding of eosinophilia in AIDS patients, no studies have correlated the eosinophilia with specific etiologies, such as parasitic infections and drug reactions. We conducted a retrospective case-control study to determine what factors were associated with eosinophilia and to determine the clinical significance of eosinophilia in this population.

METHODS: In a retrospective case-control study we compared 42 HIV-infected patients (cases) with peripheral eosinophilia (absolute eosinophil count > 500 cells/mm3) with 84 HIV-infected controls without eosinophilia. Cases were matched to controls by date, and by CD4 cell count. Data on clinical parameters possibly associated with eosinophilia were collected and compared among cases and controls. RESULTS: Eosinophilia was seen in patients with late-stage HIV disease (median CD4 cell count of 26 cells/mm3). Cases were more likely to be black (52% versus 18%, P = O.OOOl), have pruritus (50% versus 20%, P = 0.002), and have a physician-documented rash (76% versus 52%, P = 0.02). Specific cutaneous diagnoses that were more prevalent in cases versus controls were eosinophilic folliculitis (24% versus l%, P = O.OOOl), atopic dermatitis (14% versus l%, P = O.Ol), and prurigo nodularis (7% versus 0, P = 0.07). Other parameters commonly associated with eosinophilia such as allergic reactions, parasitic infection, malignancy, and adrenal insufficiency were not found at higher incidence in cases. CONCLUSIONS: Eosinophilia in AIDS patients is associated with cutaneous disease, but not with other conditions commonly associated with eosinophilia including parasitic infections, allergic reactions, or malignancy. Extensive work up for asymptomatic eosinophilia in patients with AIDS and cutaneous disease is not warranted. Am J Med. 1997; 102:449-453, o 1997 by Excerpta Medica, Inc.

METHODS Study Population The study population was drawn from the population of all HIV-infected inpatients and outpatients seen at Parkland Memorial Hospital between March 1, 1994 and February 28, 1995. Parkland is a l,OOObed public hospital, which is the primary teaching hospital for The University of Texas Southwestern Medical School. The Parkland Hospital HIV clinic treats approximately 2,000 HIV-infected patients.

Study Design

From the Untverslty of Texas Southwestern Medical Center; the Dallas Veterans Affairs Medical Center, Dallas, Texas. Presented as abstract MOB.1279 at the XI International Conference on AIDS, Vancouver, July, 1996. Requests for reprints should be addressed to Daniel J. Skiest, MD, The University of Texas Southwestern Medical Center at Dallas, Division of Infectious Diseases, 5323 Harry Hines Boulevard, Dallas, Texas 75235-

9113. Manuscript submitted January 14, 1997.

August

9, 1996

In revised

form I

I

01997 by Excerpta All rights reserved.

and accepted

Medica,

Inc.

Patients were identified from The University of Texas Southwestern Medical Center AIDS Case Registry, which is a computerized database containing demographic, clinical, and laboratory data on ail HIV-infected patients seen at Parkland Memorial Hospital. Case-patients (cases) were selected by searching the database for all absolute eosinophil counts (AEC) of > 500 cells/mm3. Charts of these patients were then reviewed to determine the data of the first AEC > 500. This date was designated as the baseline. Each case was paired with 2 controls, Qow-9343/97/.$17.m PII SO!302-9343(97)00048-X

449

EOSINOPHILIA IN HIV-INFECTED PATIENTS/SKIEST

TABLE

I Baseline Characteristics Study Cohort n = 42 35.9 38 (90.5)

Age Male no. 1%) Race no. (%I Black White Hispanic Risk Category n (%I Male-male sex Heterosexual sex IDU IDU and male-male sex Transfusion-related IDU and heterosexual CD4 cell count (cells/mm3)* WBC (cells/mm3)* AEC (cells/mm3)’ * median

ET AL

(range),

IDU = Intravenous

22 (52.4) 13 (30.9) 7 (16.7)

0.48

drug use; WBC

= white blood

Statistical Analysis Differences between the two groups were determined by the chi square test or the Fisher’s exact test (two-tailed). Odds ratio (OR) analyses were performed to determine the magnitude of parameters associated with eosinophilia; 95% confidence intervals (CIs) were calculated. A P value 5 0.05 was considered to be significant. The True EPISTAT software package (True EPISTAT Services, Richardson, Texas) was used for all analyses. May

1997

The American

Journal

P 0.87 0.74 0.0001

15 (17.8) 58 (69.0) 11 (13.1)

28 (66.7) 7 (16.7) 3 (7.1) 1 (2.4) 2 (4.8) 1 (2.4) 29 (l-334) 3,900 (1,400-9,200) 716 (504-3,318)

which were randomly selected from the AIDS Case Registry, and matched to cases by CD4 cell counts, in intervals of 50 cells/mm” and by baseline date in 6-month intervals. The baseline date for each control was selected to correspond with the matched case patient’s first AEC > 500. The following information was extracted from each medical record: demographic information including gender, age, race, and HIV risk factor; previous opportunistic infections; diarrhea1 illness (diagnosed 4 months prior to or less than 1 month after baseline); pathogens isolated from stool studies; history of atopic disease (defined as sinusitis, allergic rhinitis, or asthma); previous non-HIV related medical history; medications at baseline; history of drug allergy; new medical diagnoses (2 months before or after baseline); baseline CD4 cell and white blood cell (WBC) counts; and rash (anytime before baseline and/or less than 6 months after baseline). If rash was present, the specific diagnosis was recorded, as well as whether the rash was most likely drug related (as recorded in the chart by the patient’s physician). Most of the specific dermatologic diagnoses were made by a dermatologist.

450

Control Cohort n = 84 36.1 76 (90.5)

of Medicine8

Volume

102

55 (67.9) 9 (11.1) 7 (8.6) 8 (9.9) 1 (1.2) l(1.2) 25 (2-345) 3,200 (900-30,600) 53 (O-490) cells; AEC = absolute

eosinophil

count.

RESULTS Forty-two cases and 84 controls were included in the study. Baseline characteristics of the patients are depicted in Table I. There were no significant differences between cases and controls with respect to median age, gender, or HIV risk category. Patients were predominately male and most had iacquired HIV via male-male sex. The cases had a significantly higher proportion of black patients (52.4% versus 17.8%, P = 0.0001). The percentage of HIV-infected black patients who were seen at our institution during the study period was 33%.The median CD4 count for all patients studied was 26 cells/mm3, and the mean was 50 cells/mm3. Most of the subjects with eosinophilia (35 of 42) had a CD4 cell lcount < 100 and almost all had a diagnosis of AIDS (40 of 42). Table II lists the various patient conditions that were examined for a correlation with absolute eosinophilia. The only significant difference between the two groups was the higher prevalence of dermatological conditions in the cases. The caseshad a higher proportion of patients with physician-diagnosed rash (76% versus 53%, P = 0.02) with an OR of 2.84 (95% CI, 1.15to 7.l), and patient-reported pruritus (50%versus 20%,P = 0.002) with an OR of 3.89, (95% CI, 1.61 to 9.44). Specific dermatologic conditions occurring in cases and controls are shown in Table III. Conditions that occurred at a greater rate among study patients were eosinophilic pustular folli~culitis (EPF) (24% versus l%, P = O.OOOl),atopic dermatitis (14% versus l%, P = O.Ol), and prurigo nodular-is (7% versus 0%, P = 0.07). Folliculitis from all lcauses(EPF and non-EPF) also occurred at a higher rate in the cases (38% versus 12%, P = 0.002). Other dermato-

EOSINOPHILIA IN HIV-INFECTED PATIENTS/SKIEST

TABLE

II

Patient Diagnoses/Conditions Study Control Cohort Cohort n = 42 n = a4 Physician-diagnosed rash 32 (76%) 44 (52%) 21 (50%) 17 (20%) Pruritis Diarrhea1 illness’ 14 (33%) 26 (31%) Confirmed parasitic 3 (7%) 5 (6%) infection History of atopic conditions+ 10 (24%) 25 (30%) Drug rash 1 (2%) 9 (11%) History of drug allergy 16 (38%) 36 (43%) Newly diagnosed Ok++ 7 (17%) 14(17%) Neoplastic disease 0 (0%) 4 (5%) * Recent diarrhea1 illness defined as diarrhea1 Illness 54 months basellne or 51 month after baseline; + Includes sinusitis, allergic and asthma but does not include atopic dermatitis; ++ Diagnosed months of first AEC > 500; 01 = opportunistic infections.

P 0.02 0.0015 0.92 0.77 0.59 0.09 0.74 0.84 0.36 prior to rhinitis, wtthln 2

ET AL

pneumonia and disseminated Mycobacthum anium infection, which were diagnosed in 37% and 12%, respectively. Atopic conditions were c’ommon in both cases and controls; however, with the exception of atopic dermatitis, they did not differ in indigence among the two cohorts. There were also no differences in the incidence or type of non-HIV related illnesses or new diagnoses made within 2 months of baseline. We found no differences in individual medications used at baseline or the total number of medications per patient. The 2 most common medications were trimethoprim-sulfamethoxazole, which was used by 52% of cases versus 36% controls (P = 0.08) and fluconazole, which was used by 55% versus 61%, respectively (P = 0.55). Finally, a comparison between the cases and controls revealed a similar number who reported a previous drug allergy.

DISCUSSION TABLE

III Dermatologic Conditions Study Control Cohort Cohort n = 42 n = a4

EPF Folliculitis (non-EPF) Atopic dermatitis Prurigo nodularis Lichen simplex chronicus Xerosis Psoriasis Molluscum contagiosum Scabies Seborrheic dermatitis EPF = eosinophilic

pustular

10 6 6 3 2 7 5 7 3 10

1 9 1 0 1 6 3 6 5 15

P 0.0001 0.79 0.01 0.065 0.54 0.185 0.16 0.185 0.88 0.50

folliculitis.

logic conditions including seborrheic dermatitis, psoriasis, xerosis, molluscum contagiosum, and drug-related rash occurred at similar rates in the cases and controls. Diarrheal illness was common in both cohorts, occurring in 33% of cases and 31% of controls. However, only a few patients with diarrhea had any parasites found on stool exam (3 of 14 cases versus 5 of 26 controls), and none of these was a parasite commonly associated with peripheral eosinophilia. One case patient was diagnosed with both Isosporu belli and Strongyloides stexoralis; however, this diagnosis was made 6 months after the baseline elevation in AEC. No differences were found between cases and controls with respect to specific opportunistic infections or to the number of previous opportunistic infections/illnesses per patient: 1.5 versus 1.6 (P = 0.72), respectively. The most common opportunistic infections diagnosed were Pneumocystis car&ii

In our case-control study peripheral eosinophilia was mainly observed in patients with advanced HIV. Eighty-three percent of patients had a CD4 celU count of < 100 cells/mm3 and the mean number of previous opportunistic infections was 1.5 per patient. We found a positive correlation of eosinophilia with cutaneous disease and pruritus, but no clinically apparent association with other conditions commonly associated with eosinophilia. During the study period none of the study patients had a documented parasitic illness known to be associated with eosinophilia. One patient with eosinophilia did have Strongyloides isolated from stool 6 months after eosinophilia was first noted. Thus, eosinophilia in this patient may have been due to the Strongyloides. In addition, the group of patients with eosinophilia did not have an increased incidence of drug allergy or drug-related rash, atopic illness (excluding atopic dermatitis), malignancy, adrenal insufficiency, or other conditions often associated with eosinophilia. The high rate of cutaneous abnormalities found in our cohort is not unexpected. Others have reported high rates of skin disease in patients infected with HIV, with most but not all series showing an increased incidence of skin disease with advancing stage of HIV infection.4,8-10 These studies have documented a 79% to 92% rate of cutaneous matiestations.4z’oThe specific cutaneous conditions found at a higher rate among our patients with eosinophilia included pruritus, EPF, and atopic dermatitis. There was also a trend towards an increased incidence of prurigo nodularis. Pruritus in patients with AIDS has previously been associated with peripheral leosinophilia and elevated immunoglobulin E (IgE) l~evels.” EPF, which occurred at a significantly higher rate in the cases, is characterized clinically by pruritic May

1997

The American

Journal

of Medicine@

Volume

102

451

EOSINOPHILIA IN HN-INFECTED PATIENTS/SKIEST

ET AL

papules and pustules on the face, trunk, and proximal extremities, and histologically by a perivascular and perifollicular infiltrate with variable numbers of eosinophils.‘“-‘4 Many of the patients previously reported have had peripheral eosinophilia and elevated IgE levels.‘2~13’15 Thus, in our study cohort a substantial number of the cases of peripheral eosinophilia were likely related to the diagnosis of EPF. Although the etiology of EPF is unknown, it has been suggested that it may be related to the immune dysregulation that occurs with advanced stages of HIV.16 Another explanation is that an organism(s) colonizes the follicle resulting in tissue and peripheral eosinophilia from activation of an overstimulated immune system~“.l”>‘”

count in our cases was 29 cells/mm3.)~Thus as the switch from Thl to Th2 subsets occurs, in later stages of disease, IL5 and IL4 production may increase resulting in eosinophilia and elevation of serum IgE levels, respectively. Based on previous studies and the current study there are two possibilities for the association of cutaneous abnormalities with peripheral eosinophilia in HIV-infected patients. (1) The eosinophilia is the primary abnormality (presumably occurring because of immune dysregulation), which results in the cutaneous manifestations, similar to what is thought to occur in the idiopathic hypereosinophilic syndrome; (2) The eosinophilia represents an allergic response to an unidentified stimulus, which is p.rimarily conThe immunological abnormalities that occur in fined to the skin. Either of these explanations would later stages of HIV include nonspecific immunoglob- account for the fact that our study cohort did not ulin production as well as thymic dysregulation. The have an increased incidence of most of the condilatter is associated with a switch from predominately tions commonly associated with eosinophilia. Thl to Th2 subsets of T-helper cells. The Thl reThis study has certain limitations. Most of the dersponses are thought to be protective while the Th2 matologic diagnoses were not biopsy proven, thus responses enhance progression to AIDS.7,17,‘sInter- the specific diagnoses may be inaccurate. For exleukin-5 (IL@, which is a product of Th2 helper ample, patients diagnosed clinically with atopic dercells, has been shown to be an important cytokine matitis may have had an atopiclike skji condition, involved in eosinophil development and differentiawhich is different from the typical atopic dermatitis tion.lg Increased levels of IL5 have been detected in seen in non-HIV infected patients, in which there is the serum of patients with eosinophilia of various a genetic component. Most of the diagnoses of EPF etiologies.‘” In HIV-infected individuals a correlation were made clinically and thus it is possible that some between stage of HIV and degree of eosinophilia has of the diagnoses recorded as nonspecific folliculitis been observed.3 could represent EPF. Patients were not specifically Th2 cells also produce K-4, which is responsible evaluated for all conditions potentially associated for the regulation of IgE.17xz1Several studies have with eosinophilia such as parasitic infections, maligfound elevated serum IgE levels in HIV-infected in- nancy, and adrenal insufficiency. However, at our individuals, with some demonstrating a correlation be- stitution, HIV-infected patients with CD4 cell counts tween high serum IgE levels and low CD4 cell of < 100 cells/mm3 are followed very closely (every counts.22-‘4 In one series higher IgE levels correlated 1 to 2 months), and follow-up data was collected for with a more advanced clinical stage of HIV disease at least 6 months after the baseline eosinophil count. and were predictive of progression to AIDS.“3 Thus it is unlikely that there were a signilicant numSmith et al reported eight HIV-infected individuals ber of patients in whom the above conditions went with peripheral eosinophilia and no apparent cause.3 undiagnosed. Finally, the study population was deMost of these patients had extensive dermatitis and rived from a single county hospital and thus the reseveral had features of the idiopathic hypereosino- sults may not be applicable to other .populations, philia syndrome. Our patients did not demonstrate specifically populations from underdeveloped features of the idiopathic hypereosinophilia syn- regions where the incidence of parasitic diseases is higher. drome, ie, multiorgan system dysfunction involving the cutaneous, cardiac, neurologic, lymphoreticular, and pulmonary system.“5 Our cohort was more sim- CONCLUSION ilar to the patients described by Drabick et al and Peripheral eosinophilia in HIV-infected individuals May et al, who described HIV-infected patients with is associated with low CD4 cell counts and a high peripheral eosinophilia and predominately cutane- incidence of cutaneous disease, especially EPF and ous manifestations of the hypereosinophilic syn- atopic dermatitis. However, other conditions typically associated with eosinophilia including parasitic drome.26”7 These studies suggests that eosinophilia (and ele- infections and atopic conditions are nor more comvation in serum IgE) in HIV-infected patients may mon in this subgroup of patients. Thus searching for result from the immune dysregulation, which occurs conditions such as parasitic infections, drug reacwith advanced stages of HIV. (The median CD4 tions, collagen vascular disorders, and neoplasms, 452

May

1997

The American

Journal

of Medicine@

Volume

102

EOSINOPHILIA IN HIV-INFECTED PATIENTS/SKIEST

which is generally advocated in nonAIDS patients,’ is not likely to be cost effective and is probably unwarranted unless there are other reasons to suggest these disorders. We recommend that in an AIDS patient with peripheral eosinophilia and a cutaneous disorder, in the absence of clinical signs or symptoms of other causes of peripheral eosinophilia, an extensive work-up should not be undertaken.

REFERENCES 1. Dale DC. Leukocytosis,

leukopenia, and eosinophilia. wald E, lsselbacher KJ, et al., eds. Harrison’s Principles 12th edition, New York: McGraw-Hill; 1991:359-362.

In: Wilson of internal

JD, BraunMedicine.

2. Spry CJF. Eosinoph~ls: A Comprehensive Revrew, and Guide to the Scientific and Medical Literature. Oxford, UK: Oxford University Press: 1988. 3. Smith KJ, Skelton HG, Drablck JJ, et al. Hypereosinophilia secondary to immunodysregulation

in

patients

with

HIV-1

disease.

Arch

Dermatol.

1994;130:119-121. 4. Coopman SA, Johnson reactions In HIV infection.

RA, Platt R, Stern RS. Cutaneous NEJM. 1993;328:1670-1674.

disease

and drug

5. Smith PD, Lane HC, Gill VJ, et al. Intestinal infections in patients with the acquired immunodeficiency syndrome (AIDS): etiology and response to therapy. Ann intern Med. 1988;108:328-333. 6. Schambelan M, Grunfeld C. Endocrine abnormalities associated with HIV Infection and AIDS. In: Broder S, Merigan TC, Bolognesi D, eds. Textbook of AIDS Medione. Baltimore: Willlams and Wilkins; 1994;629-636. 7. Scully M, Berger TG. Pruritis, Staphylococcus aureus, and human immunodeficiency virus Infection. Arch Dermatol. 1990; 126684-685. 8. Smith KJ, Skelton HG, Yeager J, et al. Cutaneous findings patients: a 42 month prospective study. J Am Acad Dermatol. 754.

in HIV-1 poslbve

1994;31:746-

9. Liautaud B, Pape JW, DeHovitz JA, et al. Pruritic skin lesions. A common initial presentation of acquired immunodeficiency syndrome. Arch Dermatol.

in patients

infected

1989;125:357-361. Il. Hevia 0, Jimenez-Acosta the acquired Dermatol.

immunodeficlency

1991:24:231-235.

PR. Prevalence and clinical spectrum of skin with human immunodeficlency virus. Arch Dermatol. F, Ceballos syndrome:

PI, et al. Prurittc A clinicopathologic

papular eruption of study. JAm Acad

RA. Eosinophllic pustular folliculitis syndrome. J Am Acad Dermatol.

in patients

with

1986; 14:1020-

1022. 13. Rosenthal virus-associated advanced

D, LeBolt PE. Klumpp L, Berger TG. Human immunodeficiency eosinophilic follicullbs. A unique dermatosts associated with

human

127:206-209. 14. Buchness

immunodefictency

virus

MR, Lim HW, Hatcher

in the acquired lmmunodeficiency 15. Blauvelt A, Plott RT, Spooner

infectlon.

Arch

1991;

Dermatol.

VA, et al. Eosinophllic

pustular

folkculitis

syndrome. NEJM. 1988;318:1183-1186. K, et al. Eosinophilic follicuktis associated

with the acquired immunodeficlency Arch Dermatol. 1995; 131:360-361. 16. Magro C, Crowson A. Eosinophilic

syndrome pustular

responds

well to permethnn.

folllcular

reaction:

of immune dysregulation. Int J Dermatol. 1994;33:172-178, 17. Clerici M, Shearer GM. A THl-TH2 switch is a critical of HIV infection. lmmunol Today. 1993; 14:107-l 11.

A paradigm

step in the etiology

18. Karzon DT. Preventive vaccines. In: Broder S, Merigan TC Jr, Bolognesl D, eds. Textbook ofAIDS Medicine. Baltimore: Willlams & Wilkins: 1994:667-692. 19. Wierenga EA, Backx B, Snoek M, et al. Relative contributions of human types 1 and 2 T-helper of eosinophilia. Blood.

cell-derived

eosinophilotrophic

cytokines

to development

1993;82:1471-1479.

20. Weller PF. Cytokine regulation nopathol. 1992;62(Suppl):55-59.

of eosinophll

function.

21. Vercelli D, Geha RS. Regulation of IgE synthesis signals. J Allergy C/in Immunol. 1991;88:285-297. 22. Lucey DR, Zajac RA, Melcher GP, et al. Serum

Clrn Immunol

in humans:

Immu-

a tale of MO

IgE levels

in 622

persons

with human immunodeficiency virus Infection: IgE elevation with marked depletion of CD4+ T-cells. AIDS Res Hum Retroviruses. 1990;6:427-429, 23. Israel-Blet D, Labrousse F, Tourani J, et al. Elevation of IgE in HIV-infected subjects: A marker of poor prognosis. 24. Sample S, Chernoff DN, Lenahan

JAllergy Clin Immunol. 1992;S9:68-75. GA, et al. Elevated serum concentrations

of IgE antibodies to environmental antigens rn HIV-seroposltive male homosexuals. J Allergy Clin Immonol. 1990;86:876-880. 25. Chusid MJ, Dale DC, West BC, Wolf SM. The hypereosinophrlic syndrome: analysis

1989;125:629-663. 10. Coldiron BM, Bergstresser disease

12. Soeprono FF, Schinella acquired immunodeficiency

ET AL

of fourteen

cases

with

review

of the

literature.

1975;

Medicine.

54:1-27. 26.

Drablck

JJ, MaGill

AJ, Smith KJ, et al. Hypereosinophllic

syndrome

crated with HIV infection. South Med J. 1994;87:525-529. 27. May LP, Kelly J, Sanchez M. Hypereosinophilic syndrome taneous

manifestations

in two men with HIV infecbon.

asso-

with unusual

J Am Acad

cu-

Dermatol.

1990;23:202-204.

May

1997

The American

Journal

of Medicines

Volume

10s

453