Abstract Book Vol. 2 [International Conference on AIDS (14th: 2002: Barcelona, Spain)]

XIV International AIDS Conference

Abstracts WePeA5785-WePeA5788 23

to minimize it. Various types of statistical models including the Bayesian model
have been tried for the data and none of them seems to give good predictions
and some of them depend too much on simulation studies.
Methods: Statistical modeling through various types of growth curves have been
found in the literature for the prevalence of HIV/AIDS in various countries. We
have tried a modified version of a growth curve model known as Gompertz curve
model. This model has fitted very well for the prevalence data of Botswana, and
provides a good prediction model. This model works very well for various age
groups under consideration except for the age group 15-19.
Results: The estimates of the parameters involved in the modified Gompertz
model for various age groups have been found, which help in predicting the prevalence in those age groups, at least for the next five years. The same model is fitted
for the data of whole country irrespective of the age groups, and it fits so well as
indicated by the multiple correlation coefficient which is almost around 98%.
Conclusions: We recommend that this modified version of Gompertz curve
model should be tried to fit the prevalence data of any country before trying any
other growth curves model. We have tried the model for South African Data, and
it fits remarkably well, beating even the Bayesian model for the same data.
Presenting author: Mahadevayya V Muddapur, Dept. of Statistics, University of
Botswana, PR Bag. 0022, Gaborone, Botswana, Tel.: [phone], Fax: [phone], E-mail: [email protected]
WePeA5785 Using the Bayesian approach to assess whether
the ANRS 059 trial (early cotrimoxazole
prophylaxis in Abidjan) could have been stopped
earlier
A. Kpozehouen1, A. Alioum1, X. Anglaret1, P Van De Perre2, G. Chdne',
T. N'Dri-Yoman3, R. Salamon 1. 1INSERM U.330, Universite Victor Segalen
Bordeaux 2, Bordeaux Cedex, France; 2Centre Muraz, Bobo Dioulasso, Burkina
Faso; 3Programme PAC-CI, Abidjan, Cote d'lvoire
Background: From 1996 to 1998, a placebo-controlled trial was done in Cote
d'lvoire to assess the efficacy of cotrimoxazole (CMX) prophylaxis in reducing
severe morbidity in adults at early stages of HIV-infection. The Data Safety Monitoring Board recommended to stop this trial when the second interim frequentist
analysis performed 618 days after the trial was begun showed a 43% reduction of
severe morbidity in the CMX group. We used the real data of this trial to simulate
three Bayesian interim analyses, to assess whether this approach would have led
to propose to stop this trial earlier.
Methods: We used the theory of the counting process for the survival model. We
considered three prior distributions: a non-informative one, a skeptical one and
a prior distribution based on external information. The posterior distribution was
calculated by using directing acyclic graphs and Gibbs sampling. We calculated
the predictive probability D(t) that the function of survival under placebo would be
inferior to that under CMX at t = 253 days (6 months before the first frequentist
analysis really performed in the trial), t = 465 days (first frequentist analysis) and
t = 618 days (second frequentist analysis) after the first patient was recruited. We
decided that the trial would be stopped if D(t) > 0.95.
Results: The number of included patients at 253, 465 and 618 days were 300,
394 and 511, respectively. The probabilities D(t) that the event-free survival
was lower in the CMX group compared with the placebo group using the noninformative, skeptical and external prior distributions were: 0.92, 0.94 and 1.00
for D(253); 1.00, 0.99 and 1.00 for D(465); 1.00, 1.00 and 1.00 for D(618), respectively.
Conclusion: The results of the analyses based on the three prior distributions
would have led to propose to stop this trial 6 months before the trial was really
stopped and to reduce the number of included patients from 511 to 394.
Presenting author: Alphonse Kpozehouen, INSERM U.330 (Case 11), Universit6 Victor Segalen Bordeaux 2, 33076 Bordeaux Cedex, France, Tel.: +33
557571767, Fax: [phone], E-mail: [email protected]
WePeA5786 Risk factors associated with cases of pulmonary
tuberculosis coinfected by HIV with a smear
positive for Mycobacterium tuberculosis
P Godoy1, A. Dominguez2, J. Alvarez2, N. Camps3, J.M. Jansa2, J. Alcaide2,
S. Minguell4. I/nstitution, godoy@s/e.scs, Spain; 2/nstitution, Barcelona, Spain;
3l/nstitution, Girona, Spain; 'Institution, Thrragona, Spain
Background: The main factor responsible for producing new cases of tuberculosis by exogen transmission is the existence of Mycobacterium tuberculosis in the
sputum of tuberculosis patients. The objective of this study was to determine the
risk factors associated with cases of pulmonary tuberculosis coinfected by HIV
with positive smear in Catalonia (Spain).
Methods: We studied new cases of pulmonary tuberculosis coinfected by HIV
over the period: May 1996-April 1997. The variables were: age, gender, injection

drug use (IDU), consumption of alcohol, existence of caverns in thorax x-rays,
and laboratory results.
The association between the dependent variable -cases of pulmonary tuberculosis coinfected by HIV with a positive smear outcome for Mycobacterium
tuberculosis- and the rest of independent variables was assessed using crude
(OR) and adjusted (ORa) odds ratio applying unconditional logistic regression at
95% confidence intervals (CI).

Results: We detected 2508 new cases of tuberculosis. The 76.8% presented a
pulmonary tuberculosis (1926/2508). The 18.4% of pulmonary tuberculosis were
coinfected by HIV (355/1926).
The prevalence of cases with positive smear was significantly lower in those IDU
(ORa=0.6; 95% CI =0.3-1.0), and was positively associated with consumption of
alcohol (ORa=1.9; 95% CI 1.1-3.3) and caverns in thorax x-rays (ORa=2.0; 95%
Cl 1.0-3.9).
Conclusions: Study of contacts and direct observed treatment implemented to
reduce cases related to exogen transmission must be considered not only in coinfected HIV and IDU but also cases with pulmonary tuberculosis, who consume
alcohol and exhibit caverns.
Presenting author: Pere Godoy, [email protected], Spain, Tel.: [phone],
Fax: [phone], E-mail: [email protected]
WePeA5787 HIV infection progression in children: Is there a
TH1->TH2 shift?
I. Caragol, J.M. Bertran, M. Hernandez, T Espanol. Immunology Unit,
Barcelona, Spain
Background: Progression to AIDS in HIV-infected individuals has been associated to a shift from TH1 to TH2 cells.
We studied a group of HIV-infected children to investigate such association.
Methods: We studied intracellular expression of type 1 and 2 cytokines(ICC) in
CD4 and CD8T cells, CD28 and CD38 expression in CD8 cells in 44
HIV-infected children and 59 controls. Infected children were grouped according
age-specific CD4 cell counts (CDC,1994), Cat.1:n=23 (mean age:7.3 y.)
Cat.2: n=11(6.2 y.), Cat.3:n=10 (5.5 y.) and controls in G1:n=14 (2.7 y.), G2:n=40
(10.6 y.)
Whole blood was stimulated for 4 h., lysed, permeabilised and stained with fluorescence conjugated MoAb: IL-2 and IL-4 were cxombined with CD8 and CD3
and IFNg or TNFa with CD28, CD8 and CD3, CD28 and CD38 were combined
with CD8 to study activation and functional CD8 subsets.
Results: Cytokines: IL-2 intracellular expression in CD4 cells was lower in Cat.3
than in Cat.1 (p=0.04) but no significant differences were found in IFNg or IL-4
intracellular expression between any two of the three categories. IFNg-producing
CD8T-cells and subsets (CD3+CD8++,CD3+CD8+CD28-) were lower in Cat.3 tan
in Cat.1(p=0.02,0.02,0.07 respectively), the same differences were found when
comparing Cat.3 with Cat.2. CD8 absolute counts were lower in Cat.3 than in
both Cat.2 and 1 (p=0.04 and 0.01).
There were no differences in the parameters studied between patients in Cat.1
and 2. When compared ICC and T-cell subsets in each category with controls
there were statistical differences in all categories in most of the parameters.HIVinfected children had higher percentages of CD8+CD28-, CD8+CD38+ cells and
IFNg-producing cells within the subsets of CD8T cells, were also higher than controls.
Conclusions: There is not a clear shift from TH1 to TH2 within the CD4 cells,
but there are fewer IL-2 and IFNg CD8T-producing cells in later stages of HIV
infection in children.
Presenting author: Teresa ESPANOL, Immunology Unit, Hospital Vail d'Hebron,
08035 BARCELONA, Spain, Tel.: [phone], Fax: [phone], E-mail:
terespa@ hg.vhebron.es
WePeA5788 ILack of CD8+ Cell Antiviral Factor in CD8+ cell
granules
C.E. Mackewicz', B. Wang2, S. Metkar2, M. Ritchey', C. Froelich2, J.A. Levy1.
' University of California, Department of Medicine, San Francisco, CA, United
States; 2 Women's Hospital Evanston Northwestern Healthcare, Evanston, IL,
United States
Background: CD8+ lymphocytes from HIV-infected individuals produce a soluble factor in vitro that inhibits HIV replication in CD4+ lymphocytes. This CD8+
cell antiviral factor (CAF) blocks HIV transcription, but has yet to be identified.
Because several antiviral proteins reside in the granules of CD8+ T cells, we examined granule constituents for potential CAF activity.
Methods: CD8+ T cells from HIV-infected individuals showing strong CAFmediated antiviral activity were induced to release their granule constituents into
culture. The resulting culture fluids, taken one and three hours after stimulating
with PMA or anti-CD3, were tested for anti-HIV activity against the chemokineinsensitive HIV-1SF33 isolate.
Results: The one-hour and three-hour stimulated CD8+ cell culture fluids were
devoid of appreciable anti-HIV activity, yet contained high levels of granzyme B, a
marker for granule release. Culture fluids taken prior to, or 5-7 days after, stimulation showed strong CAF activity as expected for the kinetics of CAF production.
Conclusions: The findings suggest that, unlike the cytotoxic effector molecules
perforin and granzymes, CAF does not reside in granules but instead is likely directly secreted from CD8+ T cells. To confirm the above results, we are currently

assessing the potential anti-HIV activity in lysates of granules purified from the
antiviral CD8+ T cells.
Presenting author: Carl Mackewicz, University of California, San Francisco, Department of Medicine, 513 Parnassus, Box 1270, San Francisco, CA 94143-1270,
United States, Tel.: [phone], Fax: [phone], E-mail: [email protected].
edu