Human immunodeficiency virus type 1 antigenemia in children

Human immunodeficiency virus type antigenemia in children William Borkowsky, MD, Keith Krasinski, MD, D e b o r a h Paul, PhD, R o b e r t Holzman, MD, Tiina M o o r e , MS, D o n n a B e b e n r o t h , BS, R o b e r t L a w r e n c e , MD, a n d Sulachni C h a n d w a n i , MD From the Departments of Pediatrics and Medicine, New York University Medical Center, New York City, and Abbott Laboratories, North Chicago, Illinois Human i m m u n o d e f i c i e n c y virus type 1 (HIV-1) core antigen was assayed in the plasma of children at risk for infection with HIV to determine its usefulness in the diagnosis of infection and to correlate it with the clinical stage of disease. Antigen was d e t e c t e d in the plasma of all children less than 15 months of a g e with a c q u i r e d immunodeficiency syndrome (AIDS). Two thirds of children with AIDS-related illnesses and half of children with a s y m p t o m a t i c infection had antigen. Although 53% of plasma specimens originating from HIV-infected children y o u n g e r than 6 months of a g e c o n t a i n e d antigen, only 25% of plasma specimens from children y o u n g e r than 6 months who had no symptoms and none of the 10 specimens from HIV-infected newborn infants c o n t a i n e d antigen. Half of the specimens c o n t a i n i n g c o r e antigen also c o n t a i n e d a n t i c o r e antibody. Quantitative mean antigen levels were more likely to be e l e v a t e d in children with AIDS (516 p g / m l ) than in children with AIDS-related illnesses (295 p g / m l ) or in those who had no symptoms (70 pg/ml). Antigen levels t e n d e d to increase over time in children with a d v a n c i n g clinical illness, but they t e n d e d to d e c r e a s e over time after a diagnosis of AIDS was made. Antigen was d e t e c t e d in the plasma of 4 of 14 children without symptoms w h o subsequently reverted to an HIV seronegative state. We c o n c l u d e that the d e t e c t i o n of c o r e a n t i g e n occurs with high f r e q u e n c y in children, even y o u n g infants, with symptomatic HIV infection. Plasma c o r e antigen was less frequent in children without symptoms and was not d e t e c t e d in 10 infected children when they were tested at birth. (J PEDIATR1989;114:940-5)

The measurement of antibody directed against the human immunodeficiency virus is helpful in the diagnosis of HIV infection in adults and older children, but this assay is not as useful in the examination of infants less than 15 months

Supported by grants from the Hasbro Foundation and the Schubert Foundation, and by a New York University-National Institutes of Health Biomedical Research Grant. Presented in part at the Society for Pediatric Research meetings, Anaheim, California, April 29, 1987, and Washington, D.C., May 15, 1988. Submitted for publication June 23, 1988; accepted Dec. 21, 1988. Reprint requests: William Borkowsky, MD, Director, Pediatric Infectious Diseases and Immunology,New York University Medical Center, 550 First Ave., New York, NY 10016. 940

of age because of the presence of maternal transplacentally derived antibody. The development of assays that detect the presence of H I y core protein (p24) has opened new avenues for evaluating the clinical status of HIV-infected individuals. When this technique was applied to adults AIDS ARI ELISA HIV-1

Acquired immunodeficiencysyndrome AIDS-related illness Enzyme-linked immunosorbent assay Human immunodeficiencyvirus type 1

with the acquired immunodeficiency syndrome, as many as 80% were found to have p24 in their serum. This frequency of antigenemia declined to 60% in those with AIDS-related complex and to 8% to 38% in adults without symptoms but with HIV seropositivity.1-3 In this study, we investigated the presence of HIV core antigen in the plasma of infants

Volume 114 Number 6

and children with symptomatic and asymptomatic HIV infection to determine its usefulness in diagnosis. METHODS Patient selection. Blood was collected in heparinized tubes from 77 pediatric patients with HIV infection after informed consent was secured. Sixty-eight children were born to HIV-infected mothers, eight were infected by HlV-contaminated blood during the first year of life, and one was infected sexually by an abusing adult. All patients had antibody to HIV when tested with a standard antiHIV enzyme-linked immunosorbent assay (Abbott Laboratories, North Chicago, I11.), an ELISA using recombinant envelope antigen (Envacor [Abbott]), and a Western blot assay (Epitope Co., Beaverton, Ore.). All methods were described in a previous publication.4 Blood from 14 children who were born to HIV-infected mothers was also tested for HIV core antigenemia; these children's HIV tests subsequently showed a lack of antibody, the children remained clinically well, and their HIV cultures were negative. In addition, blood from 14 adults and 11 children without risk factors for HIV infection was tested for HIV core antigenemia. Definitions. "Human immunodeficiency virus infection" was documented in children younger than 15 months of age by demonstrating (1) that HIV was cultured from their mononuclear cells, (2) that rising levels of antibody to HIV were observed in successive blood samples, as indicated by increasing absorbance values (>0.2 OD units per sample) measured by the standard ELISA mentioned above, or (3) that clinical criteria developed consistent with a case definition of AIDS. In older children, HIV infection was documented by the measurement of serum antibody to HIV by ELISA and confirmed by Western blot analysis. Levels of HIV infection were classified according to the Centers for Disease Control criteria revised in 1988. 5 All infections classified as P-1 were considered "asymptomatic" in this report. All HIV infections with a case definition of P-2D1 (i.e., for the most par t associated with opportunistic infections such as Pneumocystis carinii infection), P-2B (i.e., progressive neurologic disease as detected by the clinical loss of developmental milestones), and P-2E (i.e., one case of non-Hodgkin lymphoma) were characterized as AIDS in this report. All other infections with a P-2 classification were regarded as AIDS-related illness. We chose to restrict the category of AIDS to P-2D1 and P-2B because our experience indicates that children with infections in this category have the worst prognosis (data not shown). Thus HIV infections associated with lymphocytic interstitial proliferation and recurrent bacterial infections were categorized as ARI, and not AIDS. Assay methods. Antigen determinations were measured

H I V- I antigenemia

941

by one of us (D.P.) using coded specimens, by means of a commercial solid-phase, antigen-capture ELISA (Abbott) by methods previously described. 4 This assay reliably measures antigen levels >30 pg/ml and has a companydetermined false-positive rate of 1 in 5000. Although antigen determinations were performed on the plasma of 32 patients only once and on the plasma of an additional 16 patients twice, the plasma samples of 30 other persons were evaluated at least three times and as many as 10 times (a mean of five times). Two hundred seventeen plasma samples were evaluated in all. The HIV antibody was measured by a commercial ELISA (Abbott) and confirmed by Western blot (Epitope), as directed by the manufacturer. The HIV-1 was detected in cultures from phytohemagglutinin-stimulated peripheral blood mononuclear cells by measuring HIV core antigen in the supernatants twice weekly for 4 weeks. The cells were grown at a concentration of 2 million cells per milliliter in tissue culture medium containing calcium and supplemented w i ~ 10% fetal calf serum and 10% interleukin-2. The presence of increasing antigen levels in the supernatant on successive samples obtained after day 3 of culture was considered as evidence of HIV infection of the mononuclear cells. The measurement of core antigen in place of reverse transcriptase has been adopted as the technique of choice for measuring HIV by the virology committee of the National Institutes of Health AIDS Clinical Trials Groups. Although our cultures frequently did not employ the cocultivation of phytohemagglutinin-stimulated uninfected mononuclear cells, we have detected HIV antigen in the mononuclear cell cultures from >90% of the children with AIDS and from >80% of children beyond birth who either had no symptoms or had ARI when these cultures were grown on multiple occasions. Blood samples from all children less than 15 months of age without a clinical diagnosis of AIDS were cultured for virus. Data analysis. Quantitative plasma HIV antigen concentrations in the three clinical groups were analyzed by regression analysis with an analysis of variance or Wilcoxon scores by means of Prodas software (Conceptual Software, Inc., Houston, Texas) on an NCR AT computer (NCR Corp., Dayton, Ohio). The clinical status of 12 children changed during the course of this analysis. Our initial hypothesis was to evaluate antigenemia by age and clinical status, so each plasma antigen measurement was correlated with the patient's clinical status and age at the time the plasma sample was obtained. Consequently, the results in this study are expressed as the percentage of blood specimens derived from a patient at a given age, with a given clinical status, or both, rather than as the true prevalence in this population. The chi-square test and a Mantel-Haenszel test were used to analyze these data.

942

Borkowsky et al.

The Journal of Pediatrics June 1989

Table I. HIV antigenemia as a function of age and clinical status A g e at testing (mo)

15

Total

No.

%

No.

%

No.

%

No.

%

No.

%

7/7 8/8 10/12 10/17 3/11 5/20 20/30

100 100 83 59 27 25 67

11/11 18/18 9/12 12/17 4/6 6/9 24/29

100 100 75 71 67 67 83

5/5 5/6 3/7 3/8 3/4 7/8 11/16

100 83 43 38 75 88 69

9/14 13/25 18/28 13/69 0/6 0/12 27/48

64 48 64 64

32/37

86

40/59

68

10/27

37

Table II. Quantitative evaluation of HIV antigenemia by disease classification No. of samples

13-15

Antigen concentration" (pg/ml)

49

70.4 _+ 138.1a

111 57

295.2 + 520.2b 516.2 _+ 758.7c

Significance: a versus b: P = 0.003; b versus c: p = 0.03 (by analysisof variance); a versusc: p = 0.0001. *Valuesare expressed as mean _+SD.

Because some patients were reclassified clinically as the study progressed, regression lines were also generated from antigen determinations derived from patients in whom three or more serial measurements were available. The slopes are expressed as increments or decrements of core antigen per month; the first quartile, median, and third quartile values are reported. A positive slope indicates increasing antigen values over time, and a negative slope indicates decreasing antigenemia. A slope of zero indicates no change in quantitative antigen values over time. RESULTS

Prevalence of antigenemia. Seventy-seven HIV-infected patients were evaluated for the presence of plasma HIV antigen. Only 6 (40%) of 15 children without symptoms ever had antigenemia; 18 (37%) of 49 plasma specimens from this group were antigen-positive. Core antigen was detected in 33 (70%) of 47 of children with ARI; 63 of 111 plasma specimens assayed were antigen-positive, an incidence of 57%. Core antigen was detected in the plasma of all 27 children with AIDS. Of 57 antigen determinations made during the interval including and subsequent to clinical

56

staging, 44 (77%) contained measurable antigen. An analysis using the Mantel-Haenszel chi-square test revealed a linear trend in the prevalence of antigenemia according to disease status (AIDS > ARI > asymptomatic status; p < 0.0001). HIV core antigenemia in infants less than 15 months of age. Antibody determinations to HIV are difficult to interpret in infants younger than 15 months of age, so we evaluated the sensitivity of finding HIV antigen in the plasma of such children with HIV infection (Table I). Half of the plasma specimens from children less than 6 months of age contained HIV antigen. The likelihood of finding antigen appeared to increase in accordance with clinical severity. Antigenemia was not found in plasma samples taken from 10 babies in their first week of life; all HIV-infected neonates in our population were clinically well at birth. Eighty percent of specimens from infants between 7 and 12 months of age contained antigen; 67% of specimens from even children without symptoms were positive. Almost 70% of specimens from 16 children between the ages of 13 and 15 months of age were positive, including 88% of plasma specimens from patients without symptoms. Half of specimens from 42 children older than 15 months of age were positive. However, none of 12 specimens from children who remained without symptoms beyond this age group were positive. Relationship of HIV code antigenemia and anticore antibody. Antibody to core (p24) antigen was measured by both the Western blot method and ELISA (with a solidphase bound recombinant core antigen). Of the 66 children with antigenemia, 33 lacked antibody to p24; the other half of the children had the simultaneous presence of HIV antigen and anti-core antibody in their plasma. Quantitative evaluation of HIV antigenemia. Although the prevalence of antigenemia appeared to increase with

Volume 114 Number 6

H I V - I antigenemia

943

Table Ilk Results of serial antigen testing: Regression of antigen concentration in time Slopes" ( + pg/mo) expressed as percentiles Original clinical status

Final clinical status

Asymptomatic infection (n = 11) Asymptomatic infection (n = 8) ARI (n = 2) AIDS (n = 1) ARI (n = 24) ARI (n = 18) AIDS (n = 6) AIDS (n = 11)

50thl-

25th

75th

0 0 -1.82 +91.07 0 0 +13.10 - 18.04

--0.43 -0.32 -3.64 --35.13 -22.56 -55.87 -329.00

+0.64 +0.48 0 -+21.54 +9.36 +132.15 +2.74

*A separate response was calculated for each subject. The slope represents the linear component in the char~gein antigen concentration over the course of time. r clinical severity of illness, we also attempted to determine whether there were quantitative differences. Thus HIV antigen determinations in plasma were analyzed and classified by clinical grouping (Table II). These quantitative measurements revealed that a significantly higher antigen level was seen as the clinical severity of illness increased. A striking example of the quantitative differences of antigenemia m different clinical states can be seen in data derived from two sets of twins, with each set having one child with AIDS. In one instance the other twin was clinically without symptoms, and in the other instance the remaining twin had lymphadenopathy, in each twin set the child with AIDS had an antigen level of at [east one log higher than the less ill sibling (3026 vs 144. and 1014 vs 132 pg/ml, respectively). Serial antigen determinations. We performed linear regression analyses on serial core antigen determinations from 46 children to estimate the "natural history" of HIV antigenemia (Table III). In 11 children, infections were originally classified either as asymptomatic or as AIDS; 24 children started with a diagnosis of ARI. The slope of regression lines derived from serial samples of children who originally had no symptoms tended to increase slightly over time. Eight children in this group continued to be without symptoms, and their regression line slopes did not differ from those of two children whose HIV infection progressed to clinical ARI. One child whose infection advanced tO AIDS during a 9-month period had a slope of 91 pg/mo. Children who began in the ARI classification demonstrated no obvious trend in antigen levels. The sera of six children whose infection progressed to AIDS demonstrated a tendency tQward increasing core antigen levels. Eighteen children whose clinical status remained stable had a

median regression slope of zero. Serial antigen levels of children with an original diagnosis of AIDS tended to decrease over time. Specificity of HIV antigenemia. To determine the specificity of antigenemia in children without clear evidence of HIV infection, we tested plasma from 14 adults and 11 children without risk factors for HIV infection for the presence of HIV antigen. None of these specimens contained measurable HIV antigen. We also tested 14 children, born t o HIV-infected mothers, whose serum ultimately showed a lack of antibody during the first year of life and who remained clinically well. Plasma was obtained at t h e time of their birth and at subsequent occasions during their first 2 years of life. Of the 14 infants, four were found to have plasma HIV antigen at some point during this period. Of 52 samp!es derived from these children, seven were positive, with five of the samples derived from children less than 6 months of age and two derived from children 6 to 9 months of age. The antigen levels detected in the positive specimens ranged from 92 to 208 pg/ml of plasma. None of the parents of the children with positive test results had antigenemia at the time of delivery, although one of the four developed antigenemia 6 months after the birth of the child. None of the 19 samples derived from these children who were older than 1 year of age contained antigen. DISCUSSION The development of assays to measure the presence of HIV core protein (p24) has facilitated the measurement of virus derived from stimulated T cells and macrophages in vitro, largely supplanting the measurement of viral reverse transcriptase. It has also allowed for the detection of HIV infection in persons in whom seroconversion has not yet occurred, 6 as well as ifl children who cannot make enough

944

Borkowsky et al.

antibody for detection in HIV antibody E L I S A assays. 4 However, the value of these assays in the diagnosis of HIV infection in adults is limited to the sickest individuals. In one study the prevalence of antigenemia was 81% in a group of adults with opportunistic infections but only 8% among adults without symptoms? Another study, using an identical commercial assay for antigen but assaying a larger group of individuals, revealed a prevalence of 40% in patients witt/AIDS-related complex and less than 30% in less severely affected adults with HIV infection? A third study, Using a noncommercial antigen detection system, detected core antigen in the plasma of 41% of adults with symptoms but in the plasma of only 13% of subjects without symptoms but with seropositivity for HIV. 3 Although the first assay measures predominantly core (p24) antigen, it is also able to detect other HIV antigens. which may explain the enhanced sensitivity of antigen detection in the first two studies in comparison with t h e latter. Our study indicates that HIV-infected children are much more likely than adults to have detectable HIV antigen in their plasma. We used the commercial assay identical to that used in the first two studies of adults, 1'2 so these results cannot be attributed to differences in assay sensitivity. The prevalence of antigenemia in any given plasma specimen was 77%. 57%. and 37%, in children with AIDS, children with ARI, and those without symptoms, respectively. The high rate of antigenemia m this last group, two fifths of whom had measurable plasma antigen on at least one occasion, was surprising. Why is antigenemia more prevalent in children than in adults? Previous studies have noted the inverse relationship between levels of core antigen and levels of antibody to core antigen in a given patient?. 7 This observation suggests either that excess core antigen forms a complex with specific antibody, reducing its measurable quantity, or that a loss of the ability to produce antibody reduces the clearance of antigen. A recent study 8 addressed these issues by attempting to measure both "free" and "compleXed" HIV core antigen prospectively in men with HIV seropositivity; antigenemia was detected in 37% of men with detectable HIV core antibody, and in most blood samples, the antigen was m a "complexed" state. Core antigen, in both "freer' and "complexed" forms, was detected in 86% of HIV core antibody-negative specimens. Only 3 (8%) of 38 specimens contained "free" core antigen and antibody simultaneously. In our patient population, half of the children had the simultaneous presence of core amlgen and antibody, indicating that children are more likely than adults to have viral replication, or that the

The Journal of Pediatrics June 1989

avidity and clonality of their antibody response to the core antigen are limited, or both. The increased number of mononuclear cells in children less than 5 years of age could represent an enriched environment for viral replication. Alternatively, most of the children exposed to the virus perinatally also receive maternally derived antibody to the core antigen, and their subsequent antibody response might be restricted in its antiepitopic repertoire (i.e., they recognize only a fraction of the potential immunogenic sites), allowing for enhanced recognition of viral antigens by test antisera with a more diverse repertoire. Studies in which infant mice received maternal idiotypic antibody reveal long-lasting effects on their subsequent antibody repertoire. 9-1~ Most of the children are also infected at a time when their response to many antigens is impaired by physiologic immaturity or their immune system, and the resultant antibody response to the core antigen may represent a low-avidity immunoglobulin, resulting in enhanced recognition of viral antigen in a test system using a probe antibody with higher affinity or avidity. Regardless of the explanation for the increased prevalence of HIV core antigenemia in HIV-infected children, the presence of antigen may help to identify HIV-infected children at a time when their antibody status might be ambiguous because of residual transplacentally derived antibody. As a caveat, chiidren without symptoms who were tested in the first month or so of life tended not to have detectable antigen, perhaps because of a low viral burden. The antigen level in this group did increase in the ensuing months, sometimes accompanied by the development of symptoms such as lymphadenopathy. The finding that mean antigen levels correlated with increasing Clinical Severity suggests that changing antigen concentrations in a given patient might predict future clinical status. Serial antigen measurements in a subset of our patients revealed a considerable degree of fluctuation. However, the quantitative levels tended to remain stable in children whose infection remained asymptomatic but tended to increase in the majority of patients with progression to more clinically advanced disease. Although HIV antigen levels varied unpredictably in children with ARI, antigenemia in children with AIDS tended to decrease over time. This finding should warn all investigators evaluating anti-HIV chemotherapy in children with AIDS that falling antigen levels could represent the "natural history" of antigenemia and not necessarily a favorable treatment response. In spite of the tendency for antigen levels to decrease in patients with AIDS, the mean antigen level in this group remains significantly higher than the mean level in patients with other ARIs. The mean antigen level in

Volume 114 Number 6

these children, in turn, is significantly higher than that in children who are well. Finally, the specificity of H I V antigenemia needs to be addressed. Tests of specificity performed by Abbott Laboratories reveal a very low incidence of false-positive test results, approximately 1 in 5000 (unpublished results). When false-positive results are obtained, they tend to be from tests of pregnant women. O f our 14 children who were born to HIV-infected mothers and who subsequently showed a lack of antibody, four had detectable antigen despite remaining clinically and immunologically well. All attempts to culture H I u from these children have failed. It is not known whether H I V antigen can. cross theplacenta, but this possibility needs to be explored. This suggestion seems an unlikely explanation for our finding antigen in these children at several months of age. It is also conceivable that we are not measuring actual H I V antigen in these children but, rather, a substance mimicking antigen in its specificity, such as an anti-idiotypic antibody. Our observation that quantitative antigen levels correlated with disease severity in the children argues that, for the most part, we are actually measuring antigen. In addition, the recent descriptions (1) of HIVAnfeCted adults who lost specific antibody and the recent demonstration (2) of H I V D N A in persons without detectable antibody who are sex partners of HIV-infected patients, by means of gene amplificatio n with the polymeras e chain,reaction technique, force us to consider that these children may possibly have been infected with HIV?:, 13 This possibility has been suggested by another study TM that detected H I V proviral D N A Sequences in a group of children who were born tO HIV-infected mothers but who did not have detectable antibody. We thank Dr. Saul Krugman and Dr. Joseph Dancis for their editorial helpl The expert technical assistance of David Mack is acknowledged and much al~preciated. EDITOR'S NOTE: Interested readers may wish to review the related article by Epstein et al. (Pediatrics 1988;82:9!9-24), which appeared just after final acceptance of tl/is article.-J.M.G.

H I V - I antigenemia

945

REFERENCES 1. Goudsmit J. deWolf F. Paul D, et al. The expression of human immunodeficiency virus antigen in serum and cerebrospinal fluid during acute and chronic infection. Lancet 1986:2:17780. 2. Cao Y, Valentine F, Hojvat S, et al. Detection of HIV antigen and specific antibodies to HIV core and envelope proteins in sera of patients with HIV infection. Blood 1987:70:575-8. 3. Wittek AE. Phelan MA. Wells MA. et al. Detection of HIV core protein in plasma by enzyme immunoassay. Ann Intern Med 1987:107:286-97. 4. Borkowsky W. Krasinski K. Paul D. Moore T. Bebenroth D, Chandwani S. Human immunodeficiency virus infections in infants negative for anti-HIV by enzyme-linked immun0assay. Lancet 1987;1:1168-71; 5. Centers for Disease Control: Classification system for human immunodefieieney virus (HIV) infection in children under 13 years of age. MMWR 1988:36:225-30. 6. Allain JP, Laurian Y. Paul D, et al. Serologic markers in early stages of human immunodeficieney virus infection in haemophiliacs. Lancet 1986;2:1233-6. 7. Lange J M A , Paul DA, Huisman HG, et al. Persistent HIV antigenemia and decline of HIV core antibodies associated with:transition to AIDS. Br Med J 1986:293:1459-62. 8. Lange JM.A, Paul DA, deWolf F. Coutinho RA. Goudsmit J. Viral gene expression, antibody production, and immune complex formation in human immunodefieiency virus infection. AIDS 1987:1:15-20. 9. Wikler M, Demeur C. Dewasme G, Urbain J. Immunodefielency role of maternal idiotypes. J Exp Med 1980; 152:102435. 10 Olson JC. Leslie GA. Inheritance patterns of idiotype expression: maternal-fetal immune regulatory networks. Immunogenetics 1981:13:39-56. 11 Rothstein TL, Vastola AP. Homologous monoelonal antibodies induce idiotype-specific suppression m neonates through maternal influence and in adults exposed during fetal and neonatal life. J Immunol 1984:133:1151-4. 12. Farzadegan H. Polis MA. Wolinsky S, et al. Loss of human immunodeficiency virus type l (HIV-1) antibodies with evidence of viral infection in asymptomatic homosexual men. Ann-Intern Med 1988;108:785-90. 13. Loche M, Mach B. Identification of HIV-infected seronegatire individuals by a direct diagnostic test based on hybridisation to amplified viral DNA. Lancet 1988:2:418-21. 14. Laure F. Courgnaud V. Rouzioux C, et al Detection of HIV-1 DNA in infants and children by means of the polymerase chain reaction. Lancet 1988;2:538-41.