Prognostic Markers for AIDS James J. Goedert,
MD
Human immunodeficiency virus (HIV ) mf ec tton causes a number of clinical syndromes and many laboratory abnormalities, often heralding the development of the life-threatening opportunistic infections or malignancies thaat are known as the acquired immunodeficiency syndrome (AIDS ). Drawing heavily on the results ofprospectivecohort studies, particularly those that my colleagues at the National Cancer Institute and I have conducted, this paper reviews the relationship ofAIDS to clinical signs and symptoms, immunologic measures, and viral assays. The risk ofAIDS in the next 3 years is at least 25 to 50% for HIV-infected subjects who hawe oral candidiasis, unexplained fewer, unexplained weight loss, a CD4’ lymphocyte count below 200 cells/p.l, or combinations of these. Elevated serum levels of beta* microglobulin and neopterin also appear to be strong predictive markers of AIDS, but further work is needed in diverse HIV-infected populations, such as intravenous drug users and persons in pattern 11 countries, such as Haiti and central Africa. Elevated levels of interferon or HIV-p24 antigen in the serum are insensitive but highly specific AZDS markers that may have pedictiue value independent of CD4 lymphocyte levels. Several potentially oraluabk immunologic (immunoglobulin kvels, tumor necrosis factor, soluble interkukin 2) and virologic (HIV uiremia) assays remain to be thoroughly evaluated or technically
simplified. Datu from prospectiwecohort studies have provided clinical and laboratory markers of AIDS risk that have proved essential for therapeutic trials and other clinical drcisions. As effective treatments for HIV infection and its complications begin to emerge, these marker data will also prove inv&abk for mathematic modeling of the scolle, course, and public health response epidemic. Ann Epidemiol f990; J : 129- 139. KEY WORIX:
to the
HIV, AIDS, disease markers.
INTRODUCTION
Acquired immunodeficiency syndrome (AIDS) is the final disease manifestation of the destruction of the human cellular immune system caused by infection with human immunodeficiency virus (HIV). Th ere is a prolonged and highly variable latency period between HIV infection and AIDS d’lagnosis. Interval disorders range from asymptomatic lymphadenopathy with immunologic aberrations, to acute and chronic neurologic and muscular disorders, to hematologic and clinical syndromes including immune thrombocytopenia, herpes zoster, and oral candidiasis. Finally, AIDS, in its conventional serveillance definition, is one or more specific opportunistic infections, most often Pneumocystis cur&ii pneumonia, Kaposi’s sarcoma, or a high-grade B-cell non-Hodgkin’s lymphoma. These features have been previously reviewed in detail, as have the special features of AIDS in infants (1). Early in the epidemic, we recognized that the best way to establish associations and quantify risks was through the prospective cohort study design. This paper reviews our approaches to quantifying the risk of AIDS and to establishing and assessing the predictive value of markers of AIDS risk.
From
the Viral
Epidemiology
Section,
Natwnal
Cancer Instttute. Bethesda, MD, Viral Epidemiology Institute, 434 Executive Plaza North. Rm. 434, Bethesda, MD 20892. Received Aprd 11, 1990; revwJ July 1, 1990.
Address reprint requests to: James J. Goedert,
0 1990 Elscvirr Science PuhllahlngCcl.. Inc.
MD. Sectmn,
Natwnal
Cancer
1047~2797/9O/$il~.50
130
Goedert PROGNOSTIC MARKERS FOR AIDS
BACKGROUND OF NATIONAL COHORT STUDIES
Pilot Cohort In August tive pilot
following
of a social
renamed,
eight
of the that
a strong
inhalants.
on the
15 men
In retrospect,
in Manhattan, named
immune
nitrite
inhalant
1990).
As of June
unknown
cause
valuable have
(unpublished
for testing
not been
In December deficiency
ratios
testing
diseases
sexual AIDS
Biggar
persuasive
recreational surrogate
that
We
use of nitrite
marker
for recep1981, four
all 15 men were alive and
1990).
and Aarhus,
colleagues
1 .O, a finding
be widespread.
This
renal failure of cohort
has been
assays, but the data
from our group.
pursued
evidence
below
might
forms and laboratory
men by going,
He and our other
provided
(2), we found
men who has chronic JJ, et al,
in the larger reports
Together
report
the
Denmark
emerging
epidemic
of immune
not to New York or San Francisco,
assembled
a cohort
CD4/CD8
ratios
of 259 homosexual
could
be acquired
by
intercourse with men in or from the United States or with the early cases of in Denmark (4). Additional follow-up of this cohort provided some of the
earliest
and best evidence
HIV-related
clinical
Homosexual
that HIV was the cause of AIDS
and immunologic
tive male patients
who entered
that
low CD4+
parallel
lymphocyte
(5) and of the evolution
New York, and Washington, prospective
cohorts
by recruiting
care physicians
New York. As with the Danish counts
could
of
(6).
the offices of two primary
DC, and one in Manhattan,
strated
abnormalities
Men in Manhattan,
In the spring of 1982, we initiated ton,
Goedert
men were all Men
of the sera from August
9 years of follow-up,
epidemiologic
in homosexual
with
of
a prospec-
one additional man seroconverted by (unpublished data Goedert JJ, et al,
HIV-positive
data
Dr. Robert
ratios
use was a strong
Men in Copenhagen 1981,
but to Denmark. that
prototype
included
Homosexual
men
1990, with nearly
for one of the original
and White
syndrome
(27%) of the 15 men were HIV seropositive, 1984, and no others seroconverted by I990 well, except
New York. The
T-lymphocyte
(3). On retrospective
first cases
we initiated
In the initial
deficiency
of low CD4/CD8
of the
men,
Black
Together).
had CD4/CD8
association
tive anal intercourse
men
New York
identification
in homosexual
organization
Men of Color
a subclinical
PROSPECTIVE
Men in Manhattan,
immediately
study of 15 homosexual
suggesting noted
INSTITUTE
and I’. carinii pneumonia
volunteers
(subsequently that
CANCER
of Homosexual
1981,
Kaposi’s sarcoma healthy
AEP Vol. 1, No. 2 L)rcembu 1990: 129-139
be acquired
cohort,
DC consecu-
in Washingwe demon-
by homosexual
contact
with men in New York, San Francisco, or Los Angeles, as well as further showing that this acquired deficiency of CD4+ lymphocytes was strongly associated with frequency of receptive
anal intercourse
(7, 8). With
the discovery
of HIV and the development
of the HIV antibody assay, univariate and multivariate analyses of the Manhattan cohort provided evidence that HIV seropositivity was directly related to frequent receptive anal intercourse with numerous homosexual partners, as well as being highly correlated with lymph node enlargement and CD4 + lymphopenia (3). This report also provided a 1982 to 1983 HIV annual incidence rate of 1.2% per month and an annual AIDS incidence of 6.9% per year among the HIV-seropositive subjects (3). Further follow-up of both the Manhattan and the Washington cohorts provided the foundation for analyses of markers of AIDS that are discussed below.
Goedert F’ROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 December 1990: 129-139
Persons with Hemophilia
in Hershey,
Pennsylvania,
131
and Elsewhere
Dr. Elaine Eyster started a comprehensive center for patients with hemophilia iri the early 1970s. By the mid-1970s, she was active in multicenter studies of hemophilia and was storing plasma samples from her patients who consented to participate in these studies. Dr. Eyster and I met in the summer of 1982 at a meeting called to report the first three cases of AIDS in patients with hemophilia. We initiated the prospective cohort study of her patients in September 1982. In 1984, her previously frozen plasma samples were tested for HIV antibodies. HIV was shown to have entered her cohort in 1979 and to have spread very rapidly between 1981 and 1983 (9). Pevalence of lymph node enlargement and deficiency of CD4+ lymphocytes were both related to longer duration of HIV seropositivity (9). In collaboration with Drs. Olivia Preble and ManChu Poon, we found that detection of acid-labile alpha interferon predicted AIDS development in three patients with hemophilia (lo), following which we expanded to the multicenter hemophilia study reported in more detail below.
Parenteral
Drug Users in Queens, New York
This small cohort, started in 1982, was most important for the demonstration that New York City parenteral drug users had high prevalence rates of human T-cell lymphotropic virus type I (HTLV-I) and type II (HTLV-II) antibodies, as well as HIV antibodies, and that these infections were strongly associated with black race (11). Along with the homosexual men in Denmark, Washington, DC, and Manhattan, and the hemophilia subjects in Hershey, and Queens drug users are the fifth cohort for which we estimated cumulative AIDS incidence rates, as discussed below.
AIDS INCIDENCE
FOLLOWING
HIV INFECTION
As highlighted, sometimes glaringly, by books like And the Band Played On, early in the epidemic there was a considerable underestimate of the risk of AIDS, even among many researchers, clinicians, and the affected communities of homosexual men and persons with hemophilia. Such denial began to erode only after publication of a comparison of these five cohorts of HIV-infected subjects, which demonstrated a minimum incubation period between infection and AIDS of 2 years and a strong association between the duration of the virus in the community and the 3-year incidence of AIDS, which reached 34% in our cohort of homosexual men in Manhattan, New York, who had HIV antibodies when they were enrolled (Figure 1) (12). The comparable 3-year AIDS incidence rates for the other four cohorts with prevalent HIV infections were 25% for the Queens drug users, 17% for the Washington homosexual men, 13% for the Hershey hemophiliacs, and 8% for the Danish homosexual men. For the seroconverting subjects, the HIV-incident cohorts, AIDS incidence was 0 for the first 2 years after infection but increased to 15% by 6 years after seroconversion in the one evaluable cohort, the Hershey hemophilicas (12). Cumulative AIDS incidence rates with longer follow-up are discussed in the summary at the end.
PREDICTIVE
MARKERS With this early understanding of AIDS incidence, plus the appearance of more and more friends, study subjects, and patients with the disease, there soon was considerable interest in the ability to identify those within each community at highest risk.
132
Goedert
AEP Vol. 1, No. 2 December 1990: 129-139
PROGNOSTIC MARKERS FOR AIDS
0.5
g
0.4
a
r-
z g ii zi n
-
-+-Manhattan
homosexuals
r-’ r’
0.3
I
-
l-2
is 0.2 i=
Queens drug users
+ -Washington
homosexuals
g 8 K
-
Hershey
hemophiliacs
Denmark
homosexuals
0.1
I-’
0.0
r I (I
-J
I 12 (Months
I I
I , J
I 24 TIME
after HTLV-III
antibodies
I 36
I 48
detected)
FIGURE 1 Actuarial incidence of AIDS among all 276 HIV-seropositive (formerly known as HTLV-III) study subjects in five different cohorts. Data are computed by the Kaplan-Meier survival technique, with the first seropositive specimen (or September 1982 for hemophiliacs with positive historical specimens) being used as time 0. Among the cohorts of homosexual men, the 3-year AIDS incidence was related to the duration of the virus in the community8% in Denmark, 17% in Washington, DC, and 34% in Manhattan, New York. From Goedert et al. (12), with permission.
Identifying both
and quantifying
the value of predictive
markers
required
consideration
of
the HIV epidemic and the effects of HIV in infected individuals. The former is
illustrated by the five cohorts in Figure 1 and the latter by Figure 2A, in which the AIDS epidemic in the Hershey hemophilia cohort was paralleled or mirrored by epidemics
of HIV
p24 antigenemia,
loss of p24 antibody,
and loss of T4
(CD4)
lymphocytes (13). Taking these dynamic complexities into account, three general classes of AIDS markers have emerged: clinical signs and symptoms; immunologic measures; and assays for HIV virus, HIV antigens, or anti-HIV antibodies. In general, both the number of abnormalities
and their severity tend to increase as AIDS
ap-
proaches. Unfortunately, few prospective multivariate analyses have been conducted, so the precise relationships of markers with each other and with AIDS have not been fully defined.
Clinical Signs and Symptoms Clinical studies have clearly shown that HIV-infected patients with oral candidiasis (14) (and perhaps vaginal candidiasis [15]) not due to antibiotic use are at high risk of developing an AIDS-defining opportunistic infection during the subsequent 6 months.
133
Cbedert
AEP Vol. I, No. 2 December1990: 129-l 39
PROGNOSTIC MARKERS FOR AIDS
B. 1.0 .9 .8
1
.6
1
2
4
3
5
Time since antibody
6
7
8
seroconversion
9
10
(years)
Time since first T4 count < 200 cells/pL (years)
FIGURE 2 (A) Actuarial cumulative probabilities of developing p24 antigen, retaining more than 200 T4 (CD4) cells/$ and retaining p24 antibody among 87 hemophilic patients measured from individual dates of HIV antibody seroconversion. The stippled area indicates actuarial cumulative incidence of AIDS. Follow-up was available for 73 patients free of AIDS at 4 years, 40 at 6 years, 17 at 8 years, and 2 at 10 years after seroconversion. (B) Actuarial cumulative AIDS incidence measured from time since first T4 (CD4) count fell below 200 cells//.J in 87 hemophilic patients. Cumulative AIDS incidence was 67% (confidence interval, 32 to 100%) at 2 years among the 8 patients who were positive for p24 antigen, compared with only 22% (confidence interval, 6 to 38%) among the 36 patients who were p24 antigen negative at the time the T4 (CD4) count first fell below 200 cells/$ (log rank P = .004). From Eyster et al. (13), with permission. In our cohort study of homosexual
men who were HIV seropositive
when they en-
rolled in 1982, the 3-year incidence of AIDS was significantly increased for those with fever or weight loss (16). These symptoms were associated with the development of either P. carinii pneumonia or Kaposi’s sarcoma. Men with diarrhea were at increased risk of Kaposi’s sarcoma but not P. curinii pneumonia. AIDS Cohort Study (MACS) recently confirmed but not diarrhea,
Immunologic
of homosexual
The much larger Multicenter
men with prevalent HIV infections
that subjects with oral candidiasis, have an increased 3-year incidence
has
weight loss, fever, or fatigue, of I’. cminii pneumonia
(17).
Measures
In our cohort of homosexual men with prevalent HIV infections, the 3-year incidence of I’. carinii pneumonia or Kaposi’s sarcoma was strongly related to lower CD4’ lymphocyte counts at enrollment, fewer than 300 CD4+ pneumonia
with AIDS developing in 10 (44%) of 23 men with
lymphocytes/J
(16).
In the MACS
study (17),
P. carinii
developed within 3 years in 33% of those who enrolled with fewer than
134
Goedert PROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 Decembrr 1990: 129-139
200 CD4+ lymphocytes/pi, 23% of those with 201 to 350 lymphocyte/pi, and only 4 to 9% of those with more than 350 lymphocytes/pi. The latter study also demonstrated a twofold to fourfold increased risk of I’. carinii pneumonia for subjects with poor-prognosis symptoms as well as a CD4+ lymphocyte count below 200/~1, often termed AIDS-related complex (ARC). The prognostic value of CD4 counts has also been confirmed in our multicenter cohort of subjects with hemophilia, for whom risk could be measured from HIV seroconversion rather than from enrollment. This is a key finding, as several potential biases, including onset and survival biases, are inherent in studies of prevalent cohorts (18). We prospectively evaluated 319 subjects with hemophilia who had dates of HIV seroconversion documented by retrospectively testing previously frozen sera (Figure 3) (19). Time-dependent covariate analysis demonstrated that within 2 months of testing the risk of AIDS for those with a CD4+ lymphocyte count below 200//.~1 was 16fold higher than for those with a count of 200 to 499/~1 and 256-fold (that is, 16*) higher than for those with a count of at least 5001111 (Table 1). Up to a year after testing, the risks with comparable CD4+ lymphocyte counts were 4.4-fold and 19fold, and, up to three years after testing, comparable risks were 3-fold and 9-fold. Several other immunologic abnormalities appear to be useful in predicting AIDS risk. The greatest interest was focused on betal-microglobulin. Three studies of cohorts of homosexual men with prevalent HIV infections have reported a doubling to tripling of the risk of AIDS with elevated levels of beta*-microglobulin, even after stratification or multivariate adjustment for CD4 counts (20-22). We noted similar findings in our smaller cohort of homosexual men with incident HIV infections (23). FIGURE 3 Prevalence of HIV-l antibodies from 1978 through 1988 among subjects with hemophilia or related disorders. Thick lines are the maximum likelihood estimates (with shaded 95% confidence intervals) derived from all 1128 subjects whose clotting disorder, type, and severity were known. Thinner lines are the midpoint estimates derived from the subset of subjects with a sample documented to be HIV-1 negative. HIV-1 antibodies appeared earlier and in a higher proportion among subjects with severe hemophilia A than among subjects with other clotting disorders, including moderate or mild hemophilia A and hemophilia B. From Goedert et al. (19), with permission. Severe Hemophilia A
0.8
-
J1
0.7
-
‘s
0.8
-
a '; 0.5
Maximum likelihood estimate (N = 659)
Mid-point estimate
-
1 I
0.4
$
0.3 -
.es
0.2 -
69
0.1 -
-
Other Clotting Disorders
Maximum likelihood estimate (N = 469)
0.0 -
I
I
I
78
I
79
I
80
82
I
I
1
I
81
83 YEAR
84
I
85
I
86
I
87
I
88
135
Goedert PROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 December 1990: f29-I 39
TABLE
1 Time-dependent covariate analysis of AIDS markers during three intervals before diagnosis of AIDS in 319 subjects with hemophilia who were studied beginning with their midpoint dates of HIV seroconversion”
0-2
Marker CD4 uwnt
Actual
months
Expected
2-12
Actual
months
Expected
12-36
Actual
months
Expected
(cellsi~l)
500 200-499 < 200 I’ value (trend) Relatwe hazard Serum interferon level (IU) o- 7 S-19 2 20 P value (trend) Relative hazard Serum HIV p24 antigen Negative Positive I’ value (trend) Rrlatlve hazard 2
0 2 20
4 6 7
12 II
8.0 7.2 6.8 < .OOOOl I6.O
1 8 13
12.7 3.0 I.3 : .00001 4.7
12 4 6
20.5 2.5 i .OOOO1 9.5
17 6
8.9 8.8 4.2 < .00001 4.4
2 2 4
20.0 1.4 0.6 < .OO@Ol 4.1
24 4 2
21.5 1.5
15 4
.a002 5.2
1.6 3.3 1.2 .02 1.0 28.5 0.9 0.5 .0005 2.5 18.2 0.8 .0003 6.2
aFrom Goedert et al. (19), with methods described elsewhere (35). Th E interval closest to the diagnosis of AIDS Includes the first measurement after diagnosis, d none was performed during the 60-day “window” before diagnosis. Earlier mtervals were limited to evaluations before AIDS, with lags of 2 and 12 months and windows of 10 and 24 months, respect&y, for the 2- to 12-month and 12- to 36,month intervals. P values were obtained by Mantel-Haenszel tests. R&we hazard was calculated from the Cox model. The levels m the assays were recoded as x = 0 for the first level (e.g., CD4 count, h 500), x = 1 for the second level (e.g., CD4 count. 200-499). and x = 2 for the third level (e.g., CD4 count, < ZOO). The relative hazard corresponds to a umt mcrease m x. Thus, for the CD4 count 2 to 12 months before diagnosis. the r&we hazard for x = 0 as compared with x = I IS 4.4, and the relatwe hazard for x = 0 as compared with x = 2 IS (4.4)2 = 19.4.
Nearly comparable predictive value may be obtained by measuring levels of serum neopterin, which is highly correlated with beta*-microglobulin levels (21, 23). Over the course of years after HIV seroconversion, neopterin levels tend to rise steadily, while beta*-microglobulin levels tend to rise more slowly, with a relatively steep increase in the year before an initial AIDS diagnosis (23). With these differences, elevated neopterin levels pose a twofold to threefold relative hazard for AIDS for each year, whereas the relative hazard of AIDS within the next year is more than sixfold with an elevated beta*-microglobulin level (23). Elevated serum levels of interferon also appear to be strong, albeit relatively insensitive, markers of AIDS (see Table 1) (19). It remains to be seen whether interferon or other potential immunologic markers of AIDS, such as immunoglobulin levels, tumor necrosis factor, or soluble interleukin 2, improve the predictive value of CD4 counts plus betaz-microglobulin or neopterin.
Viral Assays One or more viral markers appear to be useful in predicting AIDS onset. These assays are of even greater interest as effective antiretroviral drugs such as zidovudine (formerly called azidothymidine, AZT) enter clinical practice. The best viral marker of AIDS risk thus far reported is presence or absence of detectable p24 (core) antigen, as measured by the Abbott enzyme immunoassay. In our multicenter study of hemophiliacs with incident HIV seroconversions, those with detectable p24 antigen in the
136
Goedert PROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 December 1990: 129-139
serum were at fivefold to 9.5fold increased risk of AIDS (see Table 1) (19). At our flagship hemophilia center, in Hershey, Pennsylvania, the prediction of AIDS with p24 antigen was shown to be independent of CD4 counts (13). With risk measured from the time that the CD4 count first fell below 200 cells/pi, the 2-year AIDS incidence was 22% for the antigen-negative subjects and 67% for the antigen-positive subjects (see Figure 2B, P = 0.004) (13). Although p24 antigen has high specificity for AIDS, its sensitivity is well below 50% (13). Improved sensitivity may be possible as standardized assays are developed to quantitate p24 antibody, as this declines prior to the appearance of p24 antigen due to antigen-antibody complexing (24). It also appears that identification of plasma viremia, that is, detection of infectious HIV in fresh plasma, may also be a useful marker of progressive HIV disease. Specifically, two studies noted that HIV could be isolated from fresh peripheral blood mononuclear cells of virtually all seropositive individuals, irrespective of clinical category (25, 26). In contrast, the frequency of detecting free virus in the plasma and the titer of virus in the plasma were related to severity of symptoms and presence or absence of frank AIDS. Without question, this observation provides important insight into the biology of HIV and may have considerable importance for the evaluation of antiviral therapies. However, it remains to be seen whether it can be translated into a clinically useful marker of AIDS or other HIV diseases. Considerable progress has been made in standardizing HIV isolation procedures, but this approach still requires a very high degree of coordination and virologic sophistication to promptly deliver fresh blood samples to a laboratory that is prepared and qualified to establish and interpret titered plasma cultures of HIV.
Reproducibility For clinical applications, markers need to be reproducible within and between laboratories. Unfortunately, unlike the moderately large scale, intermittent quality assurance program for HIV antibody testing (27), there exists no national quality assurance program for the various markers of AIDS risk, discussed above. The greatest experience with reproducibility issues has been with T-cell subset testing. Not surprisingly, the variance increases multiplicatively with the number of computations required for the measurement. As CD4+ lymphocyte counts are the product of the total white blood-cell count, the proportion of lymphocytes in the differential count, and the proportion of mononuclear cells staining with anti-CD4 monoclonal antibodies, the variance is high. Experience has shown that the proportion of CD4+ lymphocytes may be as good a predictor as the count, and the variance can be reduced considerably by omitting the total white blood-cell count calculation. Less variance would be expected for assays that involve only a single measurement, such as betaz-microglobulin or neopterin determinations. Even with simplification, as is true for virtually all clinical assays, some variance must be tolerated in clinical application. Trial and error may be required for a physician to identify a laboratory that provides reproducible results. To conform to standards of good medical practice, repeated sampling and testing of individual patients are mandatory before major therapeutic interventions are begun. As a rule, eligibility for investigational protocols of pre-AIDS therapeutic drugs requires at least two consecutive CD4+ lymphocyte counts (or proportions) that fall below a threshold, such as 500 cells/PI (28).
AEP Vol. I, No. 2 December 1990: 129-139
Use of Markers for Development
Goedert
137
PROGNOSTIC MARKERS FOR AIDS
of Public Health Policy
Variance in the measurement of markers is less critical for epidemiologic research, and its application in the development of public health policy, than for the evaluation and treatment of individual patients. This is because statistical estimates are applied to populations, in which case precision and reproducibility are highly desirable but not critical. A good example of population-based marker data that can be used for policy development is the recent estimate of the distribution of CD4+ lymphocyte counts in the HIV-infected US Army personnel (29). Through a combined approach of direct measurement and application of transition rates, the investigators estimated that during 1989 17% of the infected population had fewer than 200 CD4+ lymphocytes and an additional 41% had 200 to 499 CD4+ lymphocyteslpul. Thus, 58% of the infected US population, perhaps as many as 600,000 m d’IVI‘dua 1s, would likely benefit from treatment with zidovudine (28), and 170/o, perhaps 100,000 individuals, should receive prophylaxis against I’. carinii pneumonia (30). The costs of such a program, particularly as many infected individuals have no health insurance, or inadequate insurance, have sparked considerable debate and proposed legislation.
CONCLUSION While there is controversy about the role of HIV in the pathogenesis of immune deficiency with various direct and indirect mechanisms postulated (31), it is clear that the progressive depletion of the CD4+ lymphocyte population is the hallmark of progressive HIV disease. The factors that limit replication of the virus and destruction of cells are complex and the subject of intensive laboratory study. It also is now clear that progression toward AIDS is steady, perhaps even inexorable, if risk is measured over decades. The cumulative, 8-year AIDS incidence was approximately 40% among men with hemophilia aged 35 years or older at the time of infection (19) and homosexual men who were in hepatitis B vaccine trials (32). Some recent analyses suggest a median time to AIDS of 7 to 8 years after infection for homosexual men (23, 33). Boys with hemophilia who were infected with HIV before age 18 have a substantially lower AIDS incidence rate of 13% over 8 years (19). What host, environmental, or host-viral interactive factors account for the lower risk in children remain to be determined. HIV infection and HIV disease are sure to be with us for the foreseeable future, probably for our lifetimes. The development of effective treatment modalities must be expeditious. Treatment has already resulted in more favorable AIDS incidence trends, particularly among homosexual men, but no such benefit has been noted among the many infected individuals who have poor access to such therapy (34). Thus, improved access to optimal medical care is also urgently needed. A definitive solution will depend on a safe and effective vaccine to prevent HIV infection. Although there is progress toward this end, such a product is unlikely to be in widespread use before 1995, at the earliest. Behavior modification is the immediate answer. Because virtually all transmissions of HIV in the United States occur during sexual intercourse, through sharing of contaminated drug-injection paraphernalia, or during pregnancy and childbirth, virtually all could be prevented. How to behave is rather straightforward, but very difficult to realize. The benefits of prevention have been noted among cohorts of homosexual men, among whom the incidence of new
138
Goedert PROGNOSTIC
AEP Vol. 1, No. 2 December 1990: 129-139
MARKERS FOR AIDS
HIV infections was falling before the virus was discovered in 1984. Such reductions in HIV incidence will be noted in drug-abusing and heterosexual populations only when there is internalized recognition
of risk and sufficient motivation
both of which will require profound psychologic
to change behavior,
and societal interventions
and sup-
port.
REFERENCES Goedert JJ, Blattner WA. The epidemiology
1.
deficiency virus. In: Devita VT, Jr, Hellman Treatment
and Prevention.
2.
2nd ed. Philadelphia:
Goedert JJ, Neuland,
in homosexual 3.
men, Lancet.
CY, Wallen
4.
Biggar RJ, Melbye
5.
Melbye
Danish
conditions
M, Ebbesen
evidence men:
et al. Amy1 nitrite
in homosexual
Diagnosis,
may alter T lymphocytes of retrovirus
men, Lancet.
P, et al. Low T-lymphocyte
for a transmissible
M, Bigar RJ, Ebbesen
homosexual
WC,
Etiology,
1988:33-60.
MG, Biggar RJ, et al. Determinants
and immunodeficiency
men: epidemiologic
AA, eds. AIDS:
J. B. Lippincott;
1982;1:412-6.
Goedert JJ, Samgadharan
III) antibody
and natural history of human immuno-
S, Rosenberg
agent, JAMA.
transmission,
and
l-5.
ratios in homosexual
1984;25 1: 1441-6.
P, et al. Seroepidemiology
Prevalence,
(HTLV-
1984;2:71
of HTLV-III
disease
antibody
outcome,
Br.
in
Med
J.
1984;289:573-5. 6.
Melbye
lymphotropic
M,
drome: Development Intern
7.
of immunologic
Goedert
contact
in high-incidence
areas for the acquired
immunodeficiency
syn-
1985;121:629-36.
JJ, Biggar RJ, Winn practices,
Eyster ME, Goedert
DM, et al. Decreased
Am J Epidemiol. JJ, Samgadharan
and early natural
in homo-
helper T lymphocytes
1985;121:637-644. MC;, Weiss
history of HTLV-III
SH,
antibodies
Gallo
RC,
Blattner
WA.
in persons with hemophilia,
1985;253:2219-23.
10.
Eyster ME,
possible preclinical
Goedert
JJ,
Poon
MC,
Preble
OT.
Acid-labile
marker for the acquired immunodeficiency
alpha
interferon:
syndrome in hemophilia,
A
N Engl
1983;309:583-6.
11.
Robert-Guroff
intravenous 12. 13.
M, Weiss SH, Giron JA, et al. Prevalence
drug abusers from an AIDS
risk group members,
region,
JAMA.
Science.
of HTLV-I,
and low T4 cell count,
syndrome (AIDS)
Ann Intern Med.
Klein RS, Harris CA,
initial
manifestation
incidence
of AIDS
in five cohorts
1986;231:992-5. m h emophiliacs:
JJ. Predictive Persistence
markers for
of p24 antigen
1989;110:963-9.
Butkus Small C, et al. Oral candidiasis
of the
-II, and -III in
1986;255:3133-7.
Eyster ME, Ballard JO, Gail MH, D rummond JE, Goedert
the acquired immunodeficiency 14.
endemic
Goedert JJ, Biggar RJ, Weiss SH, et al. Three-year
of HTLV-III-infected
the
Tsyn-
in homosex-
Development
J Med.
abnormalities:
for human
immunodeficiency
Goedert JJ, Biggar RJ, Winn DM et al. Decreased helper T lymphocytes
sexual men: II. Sexual
JAMA.
and clinical
the acquired
study, Ann
drome, Am. J Epidemiol.
9.
P, et al. Long term seropositivity men without
1986; 104:496-500.
I. Sexual
8.
Ebbesen
A longitudinal
Med.
ual men:
Biggar RJ,
virus type III in homosexual
acquired
immunodeticiency
in high-risk
syndrome,
N
patients
Engl
J
as
Med.
1984;311:354-8. 15.
Rhoads JL, Wright C, Redfield RR, et al. Chronic
human immunodehciency
virus infection,
16.
Goedert
JJ, Biggar RJ, Melbye
incidence
of AIDS
in homosexual
JAMA.
vaginal candidiasis
M, et al. Effect of T4 count
men infected
in women with
1987;257:3105-7. and cofactors
with human immunodeficiency
on the
virus, JAMA.
1987;257:331-4. 17.
Phair J, Mutioz A, Detels R, et al. Th e risk of Pnewnocystis
men infected
with human immunodeficiency
curinii pneumonia
virus type 1, N Engl J Med.
among
1990;322:161-5.
AEP Vol. I, No. 2 December 1990: 129- I39
Goedert PROGNOSTIC
139
MARKERS FOR AIDS
18. Brookmeyer R, Gail MH, Polk BF. The prevalent cohort study and the acquired immunodeficiency syndrome, Am J Epidemiol. 1987;126:14-24. 19. Goedert JJ, Kessler CM, Aledort LM, et al. A prospective study of human immunodeficiency virus type 1 infection and the development of AIDS in subjects with hemophilia, N Engl J Med. 1989;321:1141-8. 20. Moss AR, Bacchetti P, Osmond D, et al. Seropositivity for HIV and the development of AIDS or AIDS related conditions: three year follow up of the San Francisco General Hospital cohort, Br Med J. 1988;296:745-50. 21. Fahey JL, Taylor JMG, Detels R, et al. The prognostic value of cellular and serologic markers of infection with human immunodeficiency virus type 1, N Engl J Med. 1990;322:16672. 22. Anderson RE, Lang W, Shiboski S, Royce R, Jewel1 N, Winkelstein W, Jr. Use of &-microglobulin level and CD4 lymphocyte count to predict development of acquired immunodeficiency syndrome in persons with human immunodeficiency virus infection, Arch Intern Med. 1990;150:73-7. 23. Krgmer A, Biggar Rj, Fuchs D, et al. Levels of CD4’ lymphocytes, neopterin and beta-2-microglobulin are early predictors of AIDS, Monogr Virol HIV. 1990;18:61-73. 24. Pedersen C, Neilsen CM, Vestergaard BR, Gerstoft J, Krogsgaard K, Nielsen JO, Temporal relation of antigenaemia and loss of antibodies to core antigens to development of clinical disease in HIV infection, Br Med J. 1987;295:567-9. 25. Coombs RW, Collier AC, Allain J-P, et al. Plasma viremia in human immunodeficiency virus infection, N Engl J Med. 1989;321:1626-31. 26. Ho DD, Moudgil T, Alam M. Quantitation of human immunodeficiency virus type 1 in the blood of infected persons, N Engl J Med. 1989;321:1621-5. 27. Centers for Disease Control. Update: Serologic testing for HIV-1 antibody-United States, 1988 and 1989, MMWR. 1990;39:380-3. 28. Volberding PA, Lagakos SW, Koch MA, et al. Zidovudine in asymptomatic human immunodeficiency virus infection: A controlled trial in persons with fewer than 500 CD4positive cells per cells cubic millimeter, N Engl J Med. 1990;322:941-9. 29. Brundage JF, McNeil JG, Miller RN, et al. The current distribution of CD4+ Tlymphocyte counts among adults in the United States with human immunodeficiency virus infections: Estimates based on the experience of the U.S. Army, J AIDS. 1990;3:92-4. 30. Centers for Disease Control. Guidelines for prophylaxis against Pneumocystis catinii pneumonia for persons infected with human immunodeficiency virus, MMWR. 1989;38(suppl S5): l-9. 31. Asher MS, Sheppard HW. AIDS is immune system activation. II. The panergic imnesia hypothesis, J AIDS. 1990;3:177-91. 32. Hess01 MA, Lifson AR, O’MalIey PM, Doll LS, Jaffe HW, Rutherford GW, Prevalence, incidence, and progression of human immunodeficiency virus infection in homosexual and bisexual men in hepatitis B vaccine trials, 1978-1988, A J Epidemiol. 1989;130:1167-75. 33. Bigger RJ. Annual AIDS incidence following HIV infection, submitted for publication). 34. Gail MH, Rosenberg PS, Goedert JJ. Therapy may explain recent deficits in AIDS incidence, J AlDS,‘1990;3:296-306. 35. Gail MH. Evaluating serial cancer marker studies in patients at risk of recurrent disease, Biometrics 1981;37:67-78.
MD
Human immunodeficiency virus (HIV ) mf ec tton causes a number of clinical syndromes and many laboratory abnormalities, often heralding the development of the life-threatening opportunistic infections or malignancies thaat are known as the acquired immunodeficiency syndrome (AIDS ). Drawing heavily on the results ofprospectivecohort studies, particularly those that my colleagues at the National Cancer Institute and I have conducted, this paper reviews the relationship ofAIDS to clinical signs and symptoms, immunologic measures, and viral assays. The risk ofAIDS in the next 3 years is at least 25 to 50% for HIV-infected subjects who hawe oral candidiasis, unexplained fewer, unexplained weight loss, a CD4’ lymphocyte count below 200 cells/p.l, or combinations of these. Elevated serum levels of beta* microglobulin and neopterin also appear to be strong predictive markers of AIDS, but further work is needed in diverse HIV-infected populations, such as intravenous drug users and persons in pattern 11 countries, such as Haiti and central Africa. Elevated levels of interferon or HIV-p24 antigen in the serum are insensitive but highly specific AZDS markers that may have pedictiue value independent of CD4 lymphocyte levels. Several potentially oraluabk immunologic (immunoglobulin kvels, tumor necrosis factor, soluble interkukin 2) and virologic (HIV uiremia) assays remain to be thoroughly evaluated or technically
simplified. Datu from prospectiwecohort studies have provided clinical and laboratory markers of AIDS risk that have proved essential for therapeutic trials and other clinical drcisions. As effective treatments for HIV infection and its complications begin to emerge, these marker data will also prove inv&abk for mathematic modeling of the scolle, course, and public health response epidemic. Ann Epidemiol f990; J : 129- 139. KEY WORIX:
to the
HIV, AIDS, disease markers.
INTRODUCTION
Acquired immunodeficiency syndrome (AIDS) is the final disease manifestation of the destruction of the human cellular immune system caused by infection with human immunodeficiency virus (HIV). Th ere is a prolonged and highly variable latency period between HIV infection and AIDS d’lagnosis. Interval disorders range from asymptomatic lymphadenopathy with immunologic aberrations, to acute and chronic neurologic and muscular disorders, to hematologic and clinical syndromes including immune thrombocytopenia, herpes zoster, and oral candidiasis. Finally, AIDS, in its conventional serveillance definition, is one or more specific opportunistic infections, most often Pneumocystis cur&ii pneumonia, Kaposi’s sarcoma, or a high-grade B-cell non-Hodgkin’s lymphoma. These features have been previously reviewed in detail, as have the special features of AIDS in infants (1). Early in the epidemic, we recognized that the best way to establish associations and quantify risks was through the prospective cohort study design. This paper reviews our approaches to quantifying the risk of AIDS and to establishing and assessing the predictive value of markers of AIDS risk.
From
the Viral
Epidemiology
Section,
Natwnal
Cancer Instttute. Bethesda, MD, Viral Epidemiology Institute, 434 Executive Plaza North. Rm. 434, Bethesda, MD 20892. Received Aprd 11, 1990; revwJ July 1, 1990.
Address reprint requests to: James J. Goedert,
0 1990 Elscvirr Science PuhllahlngCcl.. Inc.
MD. Sectmn,
Natwnal
Cancer
1047~2797/9O/$il~.50
130
Goedert PROGNOSTIC MARKERS FOR AIDS
BACKGROUND OF NATIONAL COHORT STUDIES
Pilot Cohort In August tive pilot
following
of a social
renamed,
eight
of the that
a strong
inhalants.
on the
15 men
In retrospect,
in Manhattan, named
immune
nitrite
inhalant
1990).
As of June
unknown
cause
valuable have
(unpublished
for testing
not been
In December deficiency
ratios
testing
diseases
sexual AIDS
Biggar
persuasive
recreational surrogate
that
We
use of nitrite
marker
for recep1981, four
all 15 men were alive and
1990).
and Aarhus,
colleagues
1 .O, a finding
be widespread.
This
renal failure of cohort
has been
assays, but the data
from our group.
pursued
evidence
below
might
forms and laboratory
men by going,
He and our other
provided
(2), we found
men who has chronic JJ, et al,
in the larger reports
Together
report
the
Denmark
emerging
epidemic
of immune
not to New York or San Francisco,
assembled
a cohort
CD4/CD8
ratios
of 259 homosexual
could
be acquired
by
intercourse with men in or from the United States or with the early cases of in Denmark (4). Additional follow-up of this cohort provided some of the
earliest
and best evidence
HIV-related
clinical
Homosexual
that HIV was the cause of AIDS
and immunologic
tive male patients
who entered
that
low CD4+
parallel
lymphocyte
(5) and of the evolution
New York, and Washington, prospective
cohorts
by recruiting
care physicians
New York. As with the Danish counts
could
of
(6).
the offices of two primary
DC, and one in Manhattan,
strated
abnormalities
Men in Manhattan,
In the spring of 1982, we initiated ton,
Goedert
men were all Men
of the sera from August
9 years of follow-up,
epidemiologic
in homosexual
with
of
a prospec-
one additional man seroconverted by (unpublished data Goedert JJ, et al,
HIV-positive
data
Dr. Robert
ratios
use was a strong
Men in Copenhagen 1981,
but to Denmark. that
prototype
included
Homosexual
men
1990, with nearly
for one of the original
and White
syndrome
(27%) of the 15 men were HIV seropositive, 1984, and no others seroconverted by I990 well, except
New York. The
T-lymphocyte
(3). On retrospective
first cases
we initiated
In the initial
deficiency
of low CD4/CD8
of the
men,
Black
Together).
had CD4/CD8
association
tive anal intercourse
men
New York
identification
in homosexual
organization
Men of Color
a subclinical
PROSPECTIVE
Men in Manhattan,
immediately
study of 15 homosexual
suggesting noted
INSTITUTE
and I’. carinii pneumonia
volunteers
(subsequently that
CANCER
of Homosexual
1981,
Kaposi’s sarcoma healthy
AEP Vol. 1, No. 2 L)rcembu 1990: 129-139
be acquired
cohort,
DC consecu-
in Washingwe demon-
by homosexual
contact
with men in New York, San Francisco, or Los Angeles, as well as further showing that this acquired deficiency of CD4+ lymphocytes was strongly associated with frequency of receptive
anal intercourse
(7, 8). With
the discovery
of HIV and the development
of the HIV antibody assay, univariate and multivariate analyses of the Manhattan cohort provided evidence that HIV seropositivity was directly related to frequent receptive anal intercourse with numerous homosexual partners, as well as being highly correlated with lymph node enlargement and CD4 + lymphopenia (3). This report also provided a 1982 to 1983 HIV annual incidence rate of 1.2% per month and an annual AIDS incidence of 6.9% per year among the HIV-seropositive subjects (3). Further follow-up of both the Manhattan and the Washington cohorts provided the foundation for analyses of markers of AIDS that are discussed below.
Goedert F’ROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 December 1990: 129-139
Persons with Hemophilia
in Hershey,
Pennsylvania,
131
and Elsewhere
Dr. Elaine Eyster started a comprehensive center for patients with hemophilia iri the early 1970s. By the mid-1970s, she was active in multicenter studies of hemophilia and was storing plasma samples from her patients who consented to participate in these studies. Dr. Eyster and I met in the summer of 1982 at a meeting called to report the first three cases of AIDS in patients with hemophilia. We initiated the prospective cohort study of her patients in September 1982. In 1984, her previously frozen plasma samples were tested for HIV antibodies. HIV was shown to have entered her cohort in 1979 and to have spread very rapidly between 1981 and 1983 (9). Pevalence of lymph node enlargement and deficiency of CD4+ lymphocytes were both related to longer duration of HIV seropositivity (9). In collaboration with Drs. Olivia Preble and ManChu Poon, we found that detection of acid-labile alpha interferon predicted AIDS development in three patients with hemophilia (lo), following which we expanded to the multicenter hemophilia study reported in more detail below.
Parenteral
Drug Users in Queens, New York
This small cohort, started in 1982, was most important for the demonstration that New York City parenteral drug users had high prevalence rates of human T-cell lymphotropic virus type I (HTLV-I) and type II (HTLV-II) antibodies, as well as HIV antibodies, and that these infections were strongly associated with black race (11). Along with the homosexual men in Denmark, Washington, DC, and Manhattan, and the hemophilia subjects in Hershey, and Queens drug users are the fifth cohort for which we estimated cumulative AIDS incidence rates, as discussed below.
AIDS INCIDENCE
FOLLOWING
HIV INFECTION
As highlighted, sometimes glaringly, by books like And the Band Played On, early in the epidemic there was a considerable underestimate of the risk of AIDS, even among many researchers, clinicians, and the affected communities of homosexual men and persons with hemophilia. Such denial began to erode only after publication of a comparison of these five cohorts of HIV-infected subjects, which demonstrated a minimum incubation period between infection and AIDS of 2 years and a strong association between the duration of the virus in the community and the 3-year incidence of AIDS, which reached 34% in our cohort of homosexual men in Manhattan, New York, who had HIV antibodies when they were enrolled (Figure 1) (12). The comparable 3-year AIDS incidence rates for the other four cohorts with prevalent HIV infections were 25% for the Queens drug users, 17% for the Washington homosexual men, 13% for the Hershey hemophiliacs, and 8% for the Danish homosexual men. For the seroconverting subjects, the HIV-incident cohorts, AIDS incidence was 0 for the first 2 years after infection but increased to 15% by 6 years after seroconversion in the one evaluable cohort, the Hershey hemophilicas (12). Cumulative AIDS incidence rates with longer follow-up are discussed in the summary at the end.
PREDICTIVE
MARKERS With this early understanding of AIDS incidence, plus the appearance of more and more friends, study subjects, and patients with the disease, there soon was considerable interest in the ability to identify those within each community at highest risk.
132
Goedert
AEP Vol. 1, No. 2 December 1990: 129-139
PROGNOSTIC MARKERS FOR AIDS
0.5
g
0.4
a
r-
z g ii zi n
-
-+-Manhattan
homosexuals
r-’ r’
0.3
I
-
l-2
is 0.2 i=
Queens drug users
+ -Washington
homosexuals
g 8 K
-
Hershey
hemophiliacs
Denmark
homosexuals
0.1
I-’
0.0
r I (I
-J
I 12 (Months
I I
I , J
I 24 TIME
after HTLV-III
antibodies
I 36
I 48
detected)
FIGURE 1 Actuarial incidence of AIDS among all 276 HIV-seropositive (formerly known as HTLV-III) study subjects in five different cohorts. Data are computed by the Kaplan-Meier survival technique, with the first seropositive specimen (or September 1982 for hemophiliacs with positive historical specimens) being used as time 0. Among the cohorts of homosexual men, the 3-year AIDS incidence was related to the duration of the virus in the community8% in Denmark, 17% in Washington, DC, and 34% in Manhattan, New York. From Goedert et al. (12), with permission.
Identifying both
and quantifying
the value of predictive
markers
required
consideration
of
the HIV epidemic and the effects of HIV in infected individuals. The former is
illustrated by the five cohorts in Figure 1 and the latter by Figure 2A, in which the AIDS epidemic in the Hershey hemophilia cohort was paralleled or mirrored by epidemics
of HIV
p24 antigenemia,
loss of p24 antibody,
and loss of T4
(CD4)
lymphocytes (13). Taking these dynamic complexities into account, three general classes of AIDS markers have emerged: clinical signs and symptoms; immunologic measures; and assays for HIV virus, HIV antigens, or anti-HIV antibodies. In general, both the number of abnormalities
and their severity tend to increase as AIDS
ap-
proaches. Unfortunately, few prospective multivariate analyses have been conducted, so the precise relationships of markers with each other and with AIDS have not been fully defined.
Clinical Signs and Symptoms Clinical studies have clearly shown that HIV-infected patients with oral candidiasis (14) (and perhaps vaginal candidiasis [15]) not due to antibiotic use are at high risk of developing an AIDS-defining opportunistic infection during the subsequent 6 months.
133
Cbedert
AEP Vol. I, No. 2 December1990: 129-l 39
PROGNOSTIC MARKERS FOR AIDS
B. 1.0 .9 .8
1
.6
1
2
4
3
5
Time since antibody
6
7
8
seroconversion
9
10
(years)
Time since first T4 count < 200 cells/pL (years)
FIGURE 2 (A) Actuarial cumulative probabilities of developing p24 antigen, retaining more than 200 T4 (CD4) cells/$ and retaining p24 antibody among 87 hemophilic patients measured from individual dates of HIV antibody seroconversion. The stippled area indicates actuarial cumulative incidence of AIDS. Follow-up was available for 73 patients free of AIDS at 4 years, 40 at 6 years, 17 at 8 years, and 2 at 10 years after seroconversion. (B) Actuarial cumulative AIDS incidence measured from time since first T4 (CD4) count fell below 200 cells//.J in 87 hemophilic patients. Cumulative AIDS incidence was 67% (confidence interval, 32 to 100%) at 2 years among the 8 patients who were positive for p24 antigen, compared with only 22% (confidence interval, 6 to 38%) among the 36 patients who were p24 antigen negative at the time the T4 (CD4) count first fell below 200 cells/$ (log rank P = .004). From Eyster et al. (13), with permission. In our cohort study of homosexual
men who were HIV seropositive
when they en-
rolled in 1982, the 3-year incidence of AIDS was significantly increased for those with fever or weight loss (16). These symptoms were associated with the development of either P. carinii pneumonia or Kaposi’s sarcoma. Men with diarrhea were at increased risk of Kaposi’s sarcoma but not P. curinii pneumonia. AIDS Cohort Study (MACS) recently confirmed but not diarrhea,
Immunologic
of homosexual
The much larger Multicenter
men with prevalent HIV infections
that subjects with oral candidiasis, have an increased 3-year incidence
has
weight loss, fever, or fatigue, of I’. cminii pneumonia
(17).
Measures
In our cohort of homosexual men with prevalent HIV infections, the 3-year incidence of I’. carinii pneumonia or Kaposi’s sarcoma was strongly related to lower CD4’ lymphocyte counts at enrollment, fewer than 300 CD4+ pneumonia
with AIDS developing in 10 (44%) of 23 men with
lymphocytes/J
(16).
In the MACS
study (17),
P. carinii
developed within 3 years in 33% of those who enrolled with fewer than
134
Goedert PROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 Decembrr 1990: 129-139
200 CD4+ lymphocytes/pi, 23% of those with 201 to 350 lymphocyte/pi, and only 4 to 9% of those with more than 350 lymphocytes/pi. The latter study also demonstrated a twofold to fourfold increased risk of I’. carinii pneumonia for subjects with poor-prognosis symptoms as well as a CD4+ lymphocyte count below 200/~1, often termed AIDS-related complex (ARC). The prognostic value of CD4 counts has also been confirmed in our multicenter cohort of subjects with hemophilia, for whom risk could be measured from HIV seroconversion rather than from enrollment. This is a key finding, as several potential biases, including onset and survival biases, are inherent in studies of prevalent cohorts (18). We prospectively evaluated 319 subjects with hemophilia who had dates of HIV seroconversion documented by retrospectively testing previously frozen sera (Figure 3) (19). Time-dependent covariate analysis demonstrated that within 2 months of testing the risk of AIDS for those with a CD4+ lymphocyte count below 200//.~1 was 16fold higher than for those with a count of 200 to 499/~1 and 256-fold (that is, 16*) higher than for those with a count of at least 5001111 (Table 1). Up to a year after testing, the risks with comparable CD4+ lymphocyte counts were 4.4-fold and 19fold, and, up to three years after testing, comparable risks were 3-fold and 9-fold. Several other immunologic abnormalities appear to be useful in predicting AIDS risk. The greatest interest was focused on betal-microglobulin. Three studies of cohorts of homosexual men with prevalent HIV infections have reported a doubling to tripling of the risk of AIDS with elevated levels of beta*-microglobulin, even after stratification or multivariate adjustment for CD4 counts (20-22). We noted similar findings in our smaller cohort of homosexual men with incident HIV infections (23). FIGURE 3 Prevalence of HIV-l antibodies from 1978 through 1988 among subjects with hemophilia or related disorders. Thick lines are the maximum likelihood estimates (with shaded 95% confidence intervals) derived from all 1128 subjects whose clotting disorder, type, and severity were known. Thinner lines are the midpoint estimates derived from the subset of subjects with a sample documented to be HIV-1 negative. HIV-1 antibodies appeared earlier and in a higher proportion among subjects with severe hemophilia A than among subjects with other clotting disorders, including moderate or mild hemophilia A and hemophilia B. From Goedert et al. (19), with permission. Severe Hemophilia A
0.8
-
J1
0.7
-
‘s
0.8
-
a '; 0.5
Maximum likelihood estimate (N = 659)
Mid-point estimate
-
1 I
0.4
$
0.3 -
.es
0.2 -
69
0.1 -
-
Other Clotting Disorders
Maximum likelihood estimate (N = 469)
0.0 -
I
I
I
78
I
79
I
80
82
I
I
1
I
81
83 YEAR
84
I
85
I
86
I
87
I
88
135
Goedert PROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 December 1990: f29-I 39
TABLE
1 Time-dependent covariate analysis of AIDS markers during three intervals before diagnosis of AIDS in 319 subjects with hemophilia who were studied beginning with their midpoint dates of HIV seroconversion”
0-2
Marker CD4 uwnt
Actual
months
Expected
2-12
Actual
months
Expected
12-36
Actual
months
Expected
(cellsi~l)
500 200-499 < 200 I’ value (trend) Relatwe hazard Serum interferon level (IU) o- 7 S-19 2 20 P value (trend) Relative hazard Serum HIV p24 antigen Negative Positive I’ value (trend) Rrlatlve hazard 2
0 2 20
4 6 7
12 II
8.0 7.2 6.8 < .OOOOl I6.O
1 8 13
12.7 3.0 I.3 : .00001 4.7
12 4 6
20.5 2.5 i .OOOO1 9.5
17 6
8.9 8.8 4.2 < .00001 4.4
2 2 4
20.0 1.4 0.6 < .OO@Ol 4.1
24 4 2
21.5 1.5
15 4
.a002 5.2
1.6 3.3 1.2 .02 1.0 28.5 0.9 0.5 .0005 2.5 18.2 0.8 .0003 6.2
aFrom Goedert et al. (19), with methods described elsewhere (35). Th E interval closest to the diagnosis of AIDS Includes the first measurement after diagnosis, d none was performed during the 60-day “window” before diagnosis. Earlier mtervals were limited to evaluations before AIDS, with lags of 2 and 12 months and windows of 10 and 24 months, respect&y, for the 2- to 12-month and 12- to 36,month intervals. P values were obtained by Mantel-Haenszel tests. R&we hazard was calculated from the Cox model. The levels m the assays were recoded as x = 0 for the first level (e.g., CD4 count, h 500), x = 1 for the second level (e.g., CD4 count. 200-499). and x = 2 for the third level (e.g., CD4 count, < ZOO). The relative hazard corresponds to a umt mcrease m x. Thus, for the CD4 count 2 to 12 months before diagnosis. the r&we hazard for x = 0 as compared with x = I IS 4.4, and the relatwe hazard for x = 0 as compared with x = 2 IS (4.4)2 = 19.4.
Nearly comparable predictive value may be obtained by measuring levels of serum neopterin, which is highly correlated with beta*-microglobulin levels (21, 23). Over the course of years after HIV seroconversion, neopterin levels tend to rise steadily, while beta*-microglobulin levels tend to rise more slowly, with a relatively steep increase in the year before an initial AIDS diagnosis (23). With these differences, elevated neopterin levels pose a twofold to threefold relative hazard for AIDS for each year, whereas the relative hazard of AIDS within the next year is more than sixfold with an elevated beta*-microglobulin level (23). Elevated serum levels of interferon also appear to be strong, albeit relatively insensitive, markers of AIDS (see Table 1) (19). It remains to be seen whether interferon or other potential immunologic markers of AIDS, such as immunoglobulin levels, tumor necrosis factor, or soluble interleukin 2, improve the predictive value of CD4 counts plus betaz-microglobulin or neopterin.
Viral Assays One or more viral markers appear to be useful in predicting AIDS onset. These assays are of even greater interest as effective antiretroviral drugs such as zidovudine (formerly called azidothymidine, AZT) enter clinical practice. The best viral marker of AIDS risk thus far reported is presence or absence of detectable p24 (core) antigen, as measured by the Abbott enzyme immunoassay. In our multicenter study of hemophiliacs with incident HIV seroconversions, those with detectable p24 antigen in the
136
Goedert PROGNOSTIC MARKERS FOR AIDS
AEP Vol. I, No. 2 December 1990: 129-139
serum were at fivefold to 9.5fold increased risk of AIDS (see Table 1) (19). At our flagship hemophilia center, in Hershey, Pennsylvania, the prediction of AIDS with p24 antigen was shown to be independent of CD4 counts (13). With risk measured from the time that the CD4 count first fell below 200 cells/pi, the 2-year AIDS incidence was 22% for the antigen-negative subjects and 67% for the antigen-positive subjects (see Figure 2B, P = 0.004) (13). Although p24 antigen has high specificity for AIDS, its sensitivity is well below 50% (13). Improved sensitivity may be possible as standardized assays are developed to quantitate p24 antibody, as this declines prior to the appearance of p24 antigen due to antigen-antibody complexing (24). It also appears that identification of plasma viremia, that is, detection of infectious HIV in fresh plasma, may also be a useful marker of progressive HIV disease. Specifically, two studies noted that HIV could be isolated from fresh peripheral blood mononuclear cells of virtually all seropositive individuals, irrespective of clinical category (25, 26). In contrast, the frequency of detecting free virus in the plasma and the titer of virus in the plasma were related to severity of symptoms and presence or absence of frank AIDS. Without question, this observation provides important insight into the biology of HIV and may have considerable importance for the evaluation of antiviral therapies. However, it remains to be seen whether it can be translated into a clinically useful marker of AIDS or other HIV diseases. Considerable progress has been made in standardizing HIV isolation procedures, but this approach still requires a very high degree of coordination and virologic sophistication to promptly deliver fresh blood samples to a laboratory that is prepared and qualified to establish and interpret titered plasma cultures of HIV.
Reproducibility For clinical applications, markers need to be reproducible within and between laboratories. Unfortunately, unlike the moderately large scale, intermittent quality assurance program for HIV antibody testing (27), there exists no national quality assurance program for the various markers of AIDS risk, discussed above. The greatest experience with reproducibility issues has been with T-cell subset testing. Not surprisingly, the variance increases multiplicatively with the number of computations required for the measurement. As CD4+ lymphocyte counts are the product of the total white blood-cell count, the proportion of lymphocytes in the differential count, and the proportion of mononuclear cells staining with anti-CD4 monoclonal antibodies, the variance is high. Experience has shown that the proportion of CD4+ lymphocytes may be as good a predictor as the count, and the variance can be reduced considerably by omitting the total white blood-cell count calculation. Less variance would be expected for assays that involve only a single measurement, such as betaz-microglobulin or neopterin determinations. Even with simplification, as is true for virtually all clinical assays, some variance must be tolerated in clinical application. Trial and error may be required for a physician to identify a laboratory that provides reproducible results. To conform to standards of good medical practice, repeated sampling and testing of individual patients are mandatory before major therapeutic interventions are begun. As a rule, eligibility for investigational protocols of pre-AIDS therapeutic drugs requires at least two consecutive CD4+ lymphocyte counts (or proportions) that fall below a threshold, such as 500 cells/PI (28).
AEP Vol. I, No. 2 December 1990: 129-139
Use of Markers for Development
Goedert
137
PROGNOSTIC MARKERS FOR AIDS
of Public Health Policy
Variance in the measurement of markers is less critical for epidemiologic research, and its application in the development of public health policy, than for the evaluation and treatment of individual patients. This is because statistical estimates are applied to populations, in which case precision and reproducibility are highly desirable but not critical. A good example of population-based marker data that can be used for policy development is the recent estimate of the distribution of CD4+ lymphocyte counts in the HIV-infected US Army personnel (29). Through a combined approach of direct measurement and application of transition rates, the investigators estimated that during 1989 17% of the infected population had fewer than 200 CD4+ lymphocytes and an additional 41% had 200 to 499 CD4+ lymphocyteslpul. Thus, 58% of the infected US population, perhaps as many as 600,000 m d’IVI‘dua 1s, would likely benefit from treatment with zidovudine (28), and 170/o, perhaps 100,000 individuals, should receive prophylaxis against I’. carinii pneumonia (30). The costs of such a program, particularly as many infected individuals have no health insurance, or inadequate insurance, have sparked considerable debate and proposed legislation.
CONCLUSION While there is controversy about the role of HIV in the pathogenesis of immune deficiency with various direct and indirect mechanisms postulated (31), it is clear that the progressive depletion of the CD4+ lymphocyte population is the hallmark of progressive HIV disease. The factors that limit replication of the virus and destruction of cells are complex and the subject of intensive laboratory study. It also is now clear that progression toward AIDS is steady, perhaps even inexorable, if risk is measured over decades. The cumulative, 8-year AIDS incidence was approximately 40% among men with hemophilia aged 35 years or older at the time of infection (19) and homosexual men who were in hepatitis B vaccine trials (32). Some recent analyses suggest a median time to AIDS of 7 to 8 years after infection for homosexual men (23, 33). Boys with hemophilia who were infected with HIV before age 18 have a substantially lower AIDS incidence rate of 13% over 8 years (19). What host, environmental, or host-viral interactive factors account for the lower risk in children remain to be determined. HIV infection and HIV disease are sure to be with us for the foreseeable future, probably for our lifetimes. The development of effective treatment modalities must be expeditious. Treatment has already resulted in more favorable AIDS incidence trends, particularly among homosexual men, but no such benefit has been noted among the many infected individuals who have poor access to such therapy (34). Thus, improved access to optimal medical care is also urgently needed. A definitive solution will depend on a safe and effective vaccine to prevent HIV infection. Although there is progress toward this end, such a product is unlikely to be in widespread use before 1995, at the earliest. Behavior modification is the immediate answer. Because virtually all transmissions of HIV in the United States occur during sexual intercourse, through sharing of contaminated drug-injection paraphernalia, or during pregnancy and childbirth, virtually all could be prevented. How to behave is rather straightforward, but very difficult to realize. The benefits of prevention have been noted among cohorts of homosexual men, among whom the incidence of new
138
Goedert PROGNOSTIC
AEP Vol. 1, No. 2 December 1990: 129-139
MARKERS FOR AIDS
HIV infections was falling before the virus was discovered in 1984. Such reductions in HIV incidence will be noted in drug-abusing and heterosexual populations only when there is internalized recognition
of risk and sufficient motivation
both of which will require profound psychologic
to change behavior,
and societal interventions
and sup-
port.
REFERENCES Goedert JJ, Blattner WA. The epidemiology
1.
deficiency virus. In: Devita VT, Jr, Hellman Treatment
and Prevention.
2.
2nd ed. Philadelphia:
Goedert JJ, Neuland,
in homosexual 3.
men, Lancet.
CY, Wallen
4.
Biggar RJ, Melbye
5.
Melbye
Danish
conditions
M, Ebbesen
evidence men:
et al. Amy1 nitrite
in homosexual
Diagnosis,
may alter T lymphocytes of retrovirus
men, Lancet.
P, et al. Low T-lymphocyte
for a transmissible
M, Bigar RJ, Ebbesen
homosexual
WC,
Etiology,
1988:33-60.
MG, Biggar RJ, et al. Determinants
and immunodeficiency
men: epidemiologic
AA, eds. AIDS:
J. B. Lippincott;
1982;1:412-6.
Goedert JJ, Samgadharan
III) antibody
and natural history of human immuno-
S, Rosenberg
agent, JAMA.
transmission,
and
l-5.
ratios in homosexual
1984;25 1: 1441-6.
P, et al. Seroepidemiology
Prevalence,
(HTLV-
1984;2:71
of HTLV-III
disease
antibody
outcome,
Br.
in
Med
J.
1984;289:573-5. 6.
Melbye
lymphotropic
M,
drome: Development Intern
7.
of immunologic
Goedert
contact
in high-incidence
areas for the acquired
immunodeficiency
syn-
1985;121:629-36.
JJ, Biggar RJ, Winn practices,
Eyster ME, Goedert
DM, et al. Decreased
Am J Epidemiol. JJ, Samgadharan
and early natural
in homo-
helper T lymphocytes
1985;121:637-644. MC;, Weiss
history of HTLV-III
SH,
antibodies
Gallo
RC,
Blattner
WA.
in persons with hemophilia,
1985;253:2219-23.
10.
Eyster ME,
possible preclinical
Goedert
JJ,
Poon
MC,
Preble
OT.
Acid-labile
marker for the acquired immunodeficiency
alpha
interferon:
syndrome in hemophilia,
A
N Engl
1983;309:583-6.
11.
Robert-Guroff
intravenous 12. 13.
M, Weiss SH, Giron JA, et al. Prevalence
drug abusers from an AIDS
risk group members,
region,
JAMA.
Science.
of HTLV-I,
and low T4 cell count,
syndrome (AIDS)
Ann Intern Med.
Klein RS, Harris CA,
initial
manifestation
incidence
of AIDS
in five cohorts
1986;231:992-5. m h emophiliacs:
JJ. Predictive Persistence
markers for
of p24 antigen
1989;110:963-9.
Butkus Small C, et al. Oral candidiasis
of the
-II, and -III in
1986;255:3133-7.
Eyster ME, Ballard JO, Gail MH, D rummond JE, Goedert
the acquired immunodeficiency 14.
endemic
Goedert JJ, Biggar RJ, Weiss SH, et al. Three-year
of HTLV-III-infected
the
Tsyn-
in homosex-
Development
J Med.
abnormalities:
for human
immunodeficiency
Goedert JJ, Biggar RJ, Winn DM et al. Decreased helper T lymphocytes
sexual men: II. Sexual
JAMA.
and clinical
the acquired
study, Ann
drome, Am. J Epidemiol.
9.
P, et al. Long term seropositivity men without
1986; 104:496-500.
I. Sexual
8.
Ebbesen
A longitudinal
Med.
ual men:
Biggar RJ,
virus type III in homosexual
acquired
immunodeticiency
in high-risk
syndrome,
N
patients
Engl
J
as
Med.
1984;311:354-8. 15.
Rhoads JL, Wright C, Redfield RR, et al. Chronic
human immunodehciency
virus infection,
16.
Goedert
JJ, Biggar RJ, Melbye
incidence
of AIDS
in homosexual
JAMA.
vaginal candidiasis
M, et al. Effect of T4 count
men infected
in women with
1987;257:3105-7. and cofactors
with human immunodeficiency
on the
virus, JAMA.
1987;257:331-4. 17.
Phair J, Mutioz A, Detels R, et al. Th e risk of Pnewnocystis
men infected
with human immunodeficiency
curinii pneumonia
virus type 1, N Engl J Med.
among
1990;322:161-5.
AEP Vol. I, No. 2 December 1990: 129- I39
Goedert PROGNOSTIC
139
MARKERS FOR AIDS
18. Brookmeyer R, Gail MH, Polk BF. The prevalent cohort study and the acquired immunodeficiency syndrome, Am J Epidemiol. 1987;126:14-24. 19. Goedert JJ, Kessler CM, Aledort LM, et al. A prospective study of human immunodeficiency virus type 1 infection and the development of AIDS in subjects with hemophilia, N Engl J Med. 1989;321:1141-8. 20. Moss AR, Bacchetti P, Osmond D, et al. Seropositivity for HIV and the development of AIDS or AIDS related conditions: three year follow up of the San Francisco General Hospital cohort, Br Med J. 1988;296:745-50. 21. Fahey JL, Taylor JMG, Detels R, et al. The prognostic value of cellular and serologic markers of infection with human immunodeficiency virus type 1, N Engl J Med. 1990;322:16672. 22. Anderson RE, Lang W, Shiboski S, Royce R, Jewel1 N, Winkelstein W, Jr. Use of &-microglobulin level and CD4 lymphocyte count to predict development of acquired immunodeficiency syndrome in persons with human immunodeficiency virus infection, Arch Intern Med. 1990;150:73-7. 23. Krgmer A, Biggar Rj, Fuchs D, et al. Levels of CD4’ lymphocytes, neopterin and beta-2-microglobulin are early predictors of AIDS, Monogr Virol HIV. 1990;18:61-73. 24. Pedersen C, Neilsen CM, Vestergaard BR, Gerstoft J, Krogsgaard K, Nielsen JO, Temporal relation of antigenaemia and loss of antibodies to core antigens to development of clinical disease in HIV infection, Br Med J. 1987;295:567-9. 25. Coombs RW, Collier AC, Allain J-P, et al. Plasma viremia in human immunodeficiency virus infection, N Engl J Med. 1989;321:1626-31. 26. Ho DD, Moudgil T, Alam M. Quantitation of human immunodeficiency virus type 1 in the blood of infected persons, N Engl J Med. 1989;321:1621-5. 27. Centers for Disease Control. Update: Serologic testing for HIV-1 antibody-United States, 1988 and 1989, MMWR. 1990;39:380-3. 28. Volberding PA, Lagakos SW, Koch MA, et al. Zidovudine in asymptomatic human immunodeficiency virus infection: A controlled trial in persons with fewer than 500 CD4positive cells per cells cubic millimeter, N Engl J Med. 1990;322:941-9. 29. Brundage JF, McNeil JG, Miller RN, et al. The current distribution of CD4+ Tlymphocyte counts among adults in the United States with human immunodeficiency virus infections: Estimates based on the experience of the U.S. Army, J AIDS. 1990;3:92-4. 30. Centers for Disease Control. Guidelines for prophylaxis against Pneumocystis catinii pneumonia for persons infected with human immunodeficiency virus, MMWR. 1989;38(suppl S5): l-9. 31. Asher MS, Sheppard HW. AIDS is immune system activation. II. The panergic imnesia hypothesis, J AIDS. 1990;3:177-91. 32. Hess01 MA, Lifson AR, O’MalIey PM, Doll LS, Jaffe HW, Rutherford GW, Prevalence, incidence, and progression of human immunodeficiency virus infection in homosexual and bisexual men in hepatitis B vaccine trials, 1978-1988, A J Epidemiol. 1989;130:1167-75. 33. Bigger RJ. Annual AIDS incidence following HIV infection, submitted for publication). 34. Gail MH, Rosenberg PS, Goedert JJ. Therapy may explain recent deficits in AIDS incidence, J AlDS,‘1990;3:296-306. 35. Gail MH. Evaluating serial cancer marker studies in patients at risk of recurrent disease, Biometrics 1981;37:67-78.
