In the last 3 years, rapid advances have been made in the management of HIV infection. Numerous clinical trials have shown greater virological effect of multiple therapy over monotherapy and some have demonstrated that combination therapies improved survival and slowed the progression to disease. Highly active antiretroviral therapy (HAART) including a protease inhibitor (PI) has now become the gold standard recommended by international guidelines [1–4] and is commonly used in wealthy countries. The main treatment goals for clinicians now is to reduce and maintain the plasma HIV RNA below the limit of detection to avoid drug resistance emergence and clinical failure. Although clinical impact is the ultimate goal for treatments, most studies of HAART evaluation have focused on CD4 T-cell counts and plasma HIV RNA measurements as endpoints. Only a few studies are available on prognostic factors of clinical outcomes in the era of HAART [5,6] probably because of the dramatic decline of opportunistic infections leading to insufficient number of events for statistical analyses. Many important questions about the long-term efficacy and the safety of PI and of multiple drug combinations still remain to be answered.
Using observational data to assess the effect of treatment is rather controversial but some studies have shown that it might be helpful [7–12] especially in situations where no data from randomized trials are available because trials are unethical or not feasible. Because of the substantial number of patients included and the large variety of treatment combinations prescribed, observational studies can provide the opportunity to address questions which have not yet been answered in clinical trials, such as the comparison between PI, the impact of changing nucleoside analogues at the time of initiation of PI or long-term clinical benefits.
In this paper we report, based on an observational study population, both clinical and virological impact of various PI prescribed for the first time as part of a triple drug regimen. Each of these PI has demonstrated significant virological and clinical efficacy in clinical trials [13–15] in which a triple drug regimen including PI were compared with nucleoside analogue dual therapy.
The French Hospital Database on HIV is one of the largest cohorts of HIV patients. To date, 68 French University Hospitals are participating in this hospital cohort and 76 736 patients have been included since 1989. Three criteria are necessary for enrolment: a confirmed HIV-1 or HIV-2 infection; the patient's signed informed consent; and follow-up in one of the participating hospitals. Trained research assistants collect data prospectively from patient medical records using the French Ministry of Health software DMI2. The standardized data collection form includes characterization of transmission group, value of biological markers (plasma viral load since July 1996 only), clinical manifestations, nature and starting date of treatments prescribed and notion of intake within a clinical trials. At each hospital visit or admission or at least every 6 months all variations that occurred in patient's clinical state or treatment are reported in a follow-up form.
The study population was restricted to PI-naïve adults who started a combination therapy with one PI and two nucleoside analogues from 1 July 1996 to 31 March 31 1997. Exclusion criteria were the following: HIV-2-positive patients; patients with no follow-up after PI introduction; patients whose first PI was nelfinavir (n = 11); patients who had participated in an antiretroviral double-blinded clinical trial; patients who had ever taken a non-nucleoside reverse transcriptase inhibitor prior to triple therapy (n = 260); patients who had no CD4 cell count or HIV RNA data reported prior to commencing PI. Finally, 4946 individuals met these criteria; of these 1402 had a measured HIV RNA reported at baseline and at 12 months.
Principles of analyses and risk factors
The first part of the analysis deals with the clinical endpoints; the second part concerns virological failure as measured by plasma HIV RNA. The baseline date was the date of PI initiation. As the median time of follow-up was 14.1 months, virological failure was analysed after a close duration of treatment i.e. 12 months. All analyses were conducted as intent-to-treat analyses and patients were left in the group of their first PI prescription whatever the changes in their treatment. The same prognostic factors were considered in both analyses.
All models were adjusted for the same potential confounding factors at PI initiation: sex, transmission group (heterosexual, injecting drug user, other versus homosexual), AIDS status, age, CD4 cell counts, viral load, associated nucleoside reverse transcriptase inhibitor (NRTI) (zidovudine + didanosine, zidovudine + zalcitabine, stavudine + didanosine, stavudine + lamivudine, other versus zidovudine + lamivudine), the type of the first PI [ritonavir, saquinavir-hard gel capsule (hgc) versus indinavir], previous antiretroviral treatment (ART) experience (pre-treated changing no NRTI, pre-treated changing one NRTI, pre-treated changing two NRTI versus ART-naïve).
Lymphocyte count (log2 CD4), viral load (log10 HIV RNA) and age at PI initiation were analysed as continuous variables. If no biological data were available at the date of PI initiation, the smallest value within the 6 months prior to PI initiation was used. AIDS-defining events (ADE) were defined according to the 1993 Centers for Disease Control and Prevention clinical definition. During the inclusion period, only the hgc formulation of saquinavir (Invirase) was available.
For primary analysis, multivariate models, using Cox proportional hazard models, were performed for time to first new ADE or to death. Disease recurrences were not included for patients with AIDS at baseline. Relative risks of new ADE or death and their corresponding 95% confidence intervals (CI) were estimated while controlling for potential confounding variables. To take into account the notification delay, a censoring method first developed for incubation time estimation was used [16,17]. Individuals seen with no event in the 6 months before the end of study (i.e., from 30 June 1997 to 31 December 1997) were censored at the date of the end of study (31 December 1997). Patients with no follow-up after 30 June 1997 were considered to be lost to follow-up and were censored at the last date seen.
In order to homogenize the change of PI first prescription over time, Cox models were stratified on starting periods at 3 month intervals. In addition, and to address a potential lack of power of the analyses due to small numbers of events, several sensitivity analyses were performed to confirm the consistency of the results. These analyses included analyses on a larger population as well as the use of a multivariate extension to the proportional hazards model based on a marginal approach [18,19]. This later method allows not only the first clinical event (ADE or death) that occurs to be taken into account, as in standard Cox models, but also all those occurring during the patients' follow-up. Hence, some power gain could be expected. This analysis was also stratified on starting periods of triple therapy and adjusted for the same covariates as the Cox analyses.
Because of the multicentric origin of the data, various assays were used to quantify HIV RNA in plasma. In order to overcome the heterogeneity of the assay detection levels and the unreliability of low measurements, the value of 1000 copies/ml was chosen as the threshold value. The virological treatment failure was defined as plasma viral load > 1000 copies/ml 12 months (i.e., between 11.5 and 12.5 months) after the initiation of triple drug therapy. Patients with HIV-1 RNA levels below this limit at initiation of triple drug therapy were included in the analysis. A multivariate logistic regression model was used to identify baseline predictors of virological treatment failure and to assess the impact of first prescribed PI on virological response, after adjusting for other risk factors.
Statistical analyses were performed using SAS package version 6.12 (SAS Institute, Cary, North Carolina, USA).
Of the 1402 patients on triple therapy who had an available HIV RNA measurement at baseline and 12 months after PI initiation, 866 (61.7%) were taking indinavir as the first PI, 200 (14.3%) ritonavir and 336 (23.9%) were taking saquinavir-hgc (Table 1). Most of the patients were previously treated, only 18.4% were ART-naïve. For the 1144 pre-treated patients, the median time on ART prior to initiating PI was 20.6 months [interquartile range (IQR), 9–40]; half were previously on dual therapy and stayed on the same NRTI combination when starting PI. The most frequently prescribed NRTI combination was zidovudine + lamivudine. At enrolment, 467 (33.3%) had AIDS. Compared with other groups, patients with saquinavir-hgc were in less advanced clinical stages: only 24.1% of them had AIDS. Patients on saquinavir-hgc had lower median viral load and higher median CD4 cell count (201 × 106/l) compared with patients on ritonavir (125 × 106/l) and indinavir (107 × 106/l).
The median time of follow-up was 14.1 months (IQR, 12.5–15.4). At the end of follow-up, most of the patients (72.1%) were still on triple therapy (two NRTI and one PI) and 8.8% were on four drug regimens mainly combining saquinavir-hgc and ritonavir. Patients with indinavir were more likely to stay on triple therapy with the same PI as at the beginning (84.3%) than patients starting with saquinavir-hgc (44.8%) or ritonavir (47.3%).
During the follow-up, four patients died, 92 patients (6.6%) experienced at least one new ADE of whom two experienced an ADE before death. Overall, 114 new ADE were diagnosed, including: oesophageal candidiasis (20), herpes simplex infection (13), cytomegalovirus infection (12), Mycobacterial avium complex (MAC) infection (10), tuberculosis (10), wasting syndrome (10), Kaposi's sarcoma (10), non-Hodgkin's lymphoma (six), cerebral toxoplasmosis (six), Pneumocystis carinii pneumonia (five), cryptosporidiosis (five) and HIV encephalopathy (three). Most of the opportunistic infections were notified shortly after PI initiation (median interval, 5.9 months for candidiasis, 5.6 months for cytomegalovirus, 1.7 months for MAC) in contrast with other ADE (8.9 months for Kaposi's sarcoma, 10 months for lymphoma). Overall, 39.1% of the ADE were experienced within the first 3 months of triple therapy.
The occurrences of virological and clinical failure according to patients' baseline characteristics are given in Table 2. Patients on saquinavir-hgc experienced fewer clinical events than patients starting triple therapy with other PI (2.9 per 100 person-years versus 7.3 and 6.7 for patients starting with ritonavir and indinavir, respectively). In the mutivariate analysis (Table 3), after adjusting for confounding factors, independent significant predictors were baseline CD4 cell count (the risk is 25% lower for every twofold higher CD4 cell count at baseline) and HIV RNA (a 1 log10 higher plasma viral load at baseline corresponded to a 52% risk increase). The type of the first PI received was not found to be associated with ADE or death. Saquinavir-hgc and ritonavir compared to indinavir had no significantly different clinical effect [risk ratio, 0.70; 95% CI, 0.36–1.35 and risk ratio, 1.29; 95% CI, 0.74–2.20]. AIDS status, sex, age, transmission group, associated NRTI and ART experience were not significantly prognostic of clinical events.
The analysis of multivariate failure time data corroborated the previous results of Cox models showing no differences between PI. These trends were also confirmed by all of the sensitivity analyses (Table 4). A model adjusted for the same factors was run considering the population with HIV RNA measurement at baseline regardless of the presence of a second HIV RNA measurement at 12 months (n = 4946). In this population, where patients' characteristics were very similar to those of the 1402 patients, 365 (7.4%) were lost to follow-up with no clinical events and less than 6 months follow-up. Within a median follow-up of 12 months, 375 patients experienced a new ADE or died and again, no significant differences between PI were found. The relative risk of saquinavir-hgc versus indinavir was 0.91 (95% CI, 0.67–1.23). Similar results were found when considering ART naïve patients and pre-treated patients separately. `On treatment analyses' in contrast with `intent-to-treat' analyses were performed on patients who did not discontinue the initial PI during their follow-up (n = 717 patients): similar results were observed (Table 4), the relative risk of saquinavir-hgc versus indinavir was 0.29 (95% CI, 0.04–2.19).
Virological treatment failure
At 12 months, the CD4 cell count increased from baseline by a median 123 × 106/l (IQR, 46–203) and the median viral load reduction was −0.75 log10 copies/ml (IQR, 0–1.66). A significant negative correlation was found between responses in plasma viral load and CD4 cell count, as measured by the differences between baseline and month 12 (r, −0.32;P = 0.0001). The overall rate of virological treatment failure at 12 months was 50%. It was 46.5% among subjects whose first PI was ritonavir, 59.8% for saquinavir-hgc, 46.9% for indinavir (P = 0.001) (Table 2). The median CD4 cell count changes from baseline between PI were similar (P = 0.19): +129, +112, +124 for patients taking ritonavir, saquinavir-hgc and indinavir, respectively.
The results of the multivariate analyses are shown in Table 3. The adjusted odds ratios (ORa) for treatment virological failure were lower for older patients, and for patients with a higher baseline CD4 cell count (7% lower for a twofold increase in CD4 cell count). In contrast, the ORa were higher for patients with a higher baseline plasma HIV-1 RNA measurement (a 1 log10 increase in plasma viral load corresponded to a 44% increase in risk). Neither the combination of NRTI nor sex, transmission groups or AIDS status was associated with risk of virological failure. After controlling for these risk factors, the ORa indicated that use of saquinavir-hgc as first PI was associated with a higher risk of virological failure (ORa, 1.96; 95% CI, 1.48–2.29) compared with indinavir whereas no difference was found between ritonavir and indinavir.
Compared with ART-naïve patients, pre-treated patients were at higher risk of virological failure. Among pre-treated patients those who introduced, concomitantly with the PI, two nucleoside analogues that had not been used previously had a lower risk of virological failure compared with patients who remained on the same NRTI combination (ORa, 0.71; 95% CI, 0.49–1.04).
This observational study on a large number of highly pre-treated patients who initiated triple therapies soon after PI became available, and with an advanced immunosuppression, showed that occurrence of clinical events and virological failure are related to the baseline level of CD4 cell count and viral load at which they start PI. No differences between the effects of the different PI on clinical events were found despite a contrast in virological effect. Similarly, no difference in clinical events was found between patients according to their previous ART experience, again despite a significant contrast in virological endpoints.
Recent studies have underlined the importance of adherence on treatment efficacy in patients receiving HAART. The lack of data concerning tolerance and compliance in our cohort is certainly a drawback, but it would have been very difficult to properly document these aspects on such a large population. Interestingly the rate of patients switching their initial PI therapy is much more important for patients starting with ritonavir or saquinavir-hgc than for those starting with indinavir. The reasons for treatment discontinuation are probably varied (viral rebound or adverse events) and may be different according to the treatment  but are not notified in the database.
Because clinical trials have been conducted mainly on ART-naïve highly selected patients and with a relatively short follow-up, observational data can provide insights into the long-term efficacy of PI in an unselected population, better reflecting the usual clinical practice in the `real world'. In other words, clinical trials may reflect treatment efficacy whereas observational data illustrate treatment effectiveness at a population level [12,20]. This may lead to distinct findings such as the relatively high rate of virological failure reported in observational studies [5,21,22] : 49.9% at 12 months in this study, compared with the rates obtained in trials.
In an observational study, however, drug allocation is not random and biases might affect results in many other ways than in trials [7,23,24], particularly the bias by treatment indication. In our population, patients' profiles were very different according to PI. Patients on ritonavir and indinavir were in much more advanced stages of HIV infection than were patients on saquinavir-hgc. Because our study period coincides with the first period of PI prescriptions in France, the choice of PI was influenced by the restricted criteria retained for providing compassionate access to PI before the approval of the European Agency for the Evaluation of Medicinal Products. Saquinavir-hgc protocols accepted (or included) less immunosuppressed patients than did ritonavir or indinavir protocols.
In this context, although multivariate analyses were adjusted for known prognostic factors of clinical progression (AIDS status, CD4 cell count and viral load) residual confounding biases cannot be excluded and control of every source of confounding remains uncertain. Therefore, the results should be examined with caution. It might be that adjustments on baseline characteristics were unable to control for the imbalance between groups .
As in any observational study, biases caused by informative censoring due to `lost to follow-up' patients for clinical progression is a concern. In particular this might affect the sensitivity analysis on the 4946 patients in which 365 patients were lost to follow-up. In terms of CD4 cell count and viral load at baseline these lost to follow-up patients were similar to those who remained in the analysis. Nevertheless, to address this potential bias, we performed an analysis with the assumption that the patients considered as lost to follow-up died the day after their last visit. In this analysis of `the worst case', 740 patients (15%) experienced an ADE or died; again, no difference was found between PI.
Nonetheless, our results on virological endpoints confirm those expected from clinical trials: indinavir and ritonavir are more effective on viral load than saquinavir-hgc; patients starting HAART with ritonavir or indinavir more frequently achieved undetectable levels than patients on saquinavir-hgc. The low bioavailability of the saquinavir-hgc probably explains these findings. Hence, a more favourable clinical course might have been expected in patients taking indinavir or ritonavir. However, no difference between PI was found on clinical progression. These results are consistent with the comparative clinical trial on indinavir versus ritonavir  in which no difference in clinical outcomes between the two PI was found. No trial has ever compared saquinavir-hgc with other PI on clinical endpoints and probably no trial ever will, now that trials have shown a better virological effect of saquinavir-sgc over saquinavir-hgc . A recently published report from the Swiss HIV Cohort Study  that also considered clinical outcomes in the era of HAART found no effect of the choice of PI.
The fact that no difference on clinical progression was observed between patients receiving saquinavir-hgc and those on other PI could be explained by too short a follow-up period and incomparability between groups at baseline. Most of the patients taking saquinavir-hgc had a CD4 cell count > 200 × 106/l and were consequently at a low risk of opportunistic infections prior to receiving PI.
Recent reports have suggested that an increase in CD4 cell count is possible despite a moderate viral load decrease in patients with discordant responses to PI-containing regimens [28–30]. Although patients with saquinavir-hgc more frequently experienced a discordant response in this study (data not shown), the absence of differences in terms of immunological recovery – measured by CD4 cell count at 12 months – might also explain the findings on clinical progression.
Introducing two new nucleoside analogues when initiating HAART was found to reduce the risk of virological failure, thus confirming recently published findings , whereas no difference was found for clinical progression. Similar results were found in the Swiss study . Again, it could be that the follow-up is too short for the impact on clinical outcomes to become apparent.
Our observational study, like that of Chêne et al. and other clinical reports of MAC and cytomegalovirus infection [32–37] observed that many ADE occurred within the first months of starting a PI. In the most immunosuppressed patients, the occurrence of ADE may not reflect a failure of new combined therapy initiated but rather the presence of underlying diseases or the persistence of a previous severely compromised immune status that does not improve despite adequate suppression of virus replication.
Although in observational studies confounding biases can never be fully excluded, our results showed that virological failure and clinical failure are not necessarily concordant. Studies conducted with a longer duration of follow-up will probably help to answer the question of whether these discordant results are real or whether they reflect the delay between virological failure and clinical manifestations. At a time when other studies have found that persistent viremia can be detected despite an improvement in CD4 cell count, these results address the question of how treatment failure should be defined and whether the notion of treatment failure (that would involve the next therapeutic options) should rely on virological outcomes only.
The authors thank all of the participants of the cohort and especially the research assistants without whom this work would not have been possible. We also thank C. Goujard for her useful comments on the first draft.
1. Guidelines for the use of antiretroviral agents in HIV- infected adults and adolescents. Department of Health and Human Services and Henry J. Kaiser Family Foundation. MMWR
2. Carpenter CCJ, Fischl MA, Hammer SM. et al. Antiretroviral therapy for HIV infection in 1998: updated recommendations of the International AIDS Society-USA Panel.
JAMA 1998, 280: 78 –86.
3. Gazzard B, Moyle G:1998 revision to the British HIV Association guidelines for antiretroviral treatment of HIV seropositive individuals. BHIVA Guidelines Writing Committee.Lancet
4. Gulick RM. HIV treatment strategies: Planning for the long term.
JAMA 1998, 279: 957 –959.
5. d'Arminio Monforte A, Testa L, Adorni F. et al. Clinical outcome and predictive factors of failure of highly active antiretroviral therapy in antiretroviral-experienced patients in advanced stages of HIV-1 infection.
AIDS 1998, 12: 1631 –1637.
6. Ledergerber B, Egger M, Opravil M. et al. Clinical progression and virological failure on highly active antiretroviral therapy in HIV-1 patients: a prospective cohort study.
Lancet 1999, 353: 863 –868.
7. Gail MH. Use of the observational data for evaluating AIDS therapies.
In AIDS Clinical Trials.
Edited by Finkelstein DM, Schoenfeld DA. New York: John Wiley & Sons; 1995: 403– 422.
8. Hogg RS, Rhone SA, Yip B. et al. Antiviral effect of double and triple drug combinations amongst HIV-infected adults: lessons from the implementation of viral load-driven antiretroviral therapy.
AIDS 1998, 12: 279 –284.
9. Joffe MM, Hoover DR, Jacobson LP. et al. Estimating the effect of zidovudine on Kaposi Sarcoma from observational data using a rank preserving structural failure-time model.
Stat Med 1998, 17: 1073 –1102.
10. Egger M, Hirschel B, Francioli P. et al. Impact of new antiretroviral combination therapies in HIV infected patients in Switzerland: prospective multicentre study.
BMJ 1997, 315: 1194 –1199.
11. Mocroft A, Gill MJ, Davidson W, Phillips AN. Predictors of viral response and subsequent virological treatment failure in patients with HIV starting a protease inhibitor.
AIDS 1998, 12: 2161 –2167.
12. Muñoz A, Hoover DR. Use of the cohort studies for evaluating AIDS therapies.
In AIDS Clinical Trials.
Edited by Finkelstein DM, Schoenfeld DA. New York: John Wiley & Sons; 1995: 423– 446.
13. Clumeck N, on behalf of the saquinavir International Phase III Trial (SV14604) Group. Clinical benefit of saquinavir (SQV) plus zalcitabine (DDC) plus Zidovudine (ZDV) in untreated/minimally treated HIV-infected patients.37th Interscience Conference on Antimicrobial Agents and Chemotherapy.
Toronto, September–October 1997 [abstract LB4].
14. Hammer SM, Squires KE, Hughes MD. et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less.
:AIDS Clinical Trials Group 320 Study Team.
N Engl J Med 1997, 337: 725 –733.
15. Cameron DW, Heath Chiozzi M, Danner S. et al. Randomised placebo-controlled trial of ritonavir in advanced HIV-1 disease.
:The Advanced HIV Disease Ritonavir Study Group.
Lancet 1998, 351: 543 –549.
16. Muñoz A, Sabin CA, Phillips AN. The incubation period of AIDS.
AIDS 1997, 11: S69 –S76.
17. Cozzi Lepri A, Phillips AN, Pezzotti P. et al. Evaluating the effect of year of seroconversion on HIV progression in cohort studies.
AIDS 1998, 12: 1353 –1360.
18. Wei LJ, Lin Y, Wessfeld L. Regression analysis of multivariate incomplete failure time data by modelling marginal distributions
. J Am Stat Assoc 1989, 84: 1065 –1073.
19. Neaton J, Wentworth D. Methods of studying interventions.
:Considerations in choice of a clinical endpoint for AIDS clinical trials.
Stat Med 1994, 13: 2107 –2125.
20. Detels R, Muñoz A, McFarlane G. et al. Effectiveness of potent antiretroviral therapy on time to AIDS and death in men with known HIV infection duration.
:Multicenter AIDS Cohort Study Investigators.
JAMA 1998, 280: 1497 –1503.
21. Casado JL, Perez-Elias MJ, Antela A. et al. Predictors of long-term response to protease inhibitor therapy in a cohort of HIV-infected patients.
AIDS 1998, 12: F131 –F135.
22. Fatkenheuer G, Theisen A, Rockstroh J. et al. Virological treatment failure of protease inhibitor therapy in an unselected cohort of HIV-infected patients.
AIDS 1997, 11: F113 –F116.
23. Glesby MJ, Hoover DR. Survivor treatment selection bias in observational studies: Examples from the AIDS literature.
Ann Intern Med 1996, 124: 999 –1005.
24. Easterbrook PJ. Research potentials and pitfalls in the use of an HIV clinical database: Chelsea and Westminster hospital.
J Acq Immun Defic Synd Hum Retrovirol 1998, 17 (suppl 1): S28 –S33.
25. Phillips A, Grabar S, Tassié J. et al. Use of observational databases to evaluate the effectiveness of antiretroviral therapy for HIV infection: comparison of cohort studies with randomized trials.
AIDS 1999, 13: 2075 –2082.
26. Tomino C, Vella S, Fragola FV, et al.A multicenter randomised study comparing ritonavir and indinavir in 1251 previously treated patients with CD4 below 50/mm3 (Italian ISS-IP1Study Group).12th World AIDS Conference.
Geneva, July 1998 [abstract 12237].
27. Mitsuyasu RT, Skolnik PR, Cohen SR. et al. Activity of the soft gelatin formulation of saquinavir in combination therapy in antiretroviral-naïve patients.
AIDS 1998, 12: F103 –F109.
28. Grabar S, Costagliola D. Clinical outcomes of patients under HAART with discordant response to CD4 and viral load at 6 months.Sixth Conference on Retroviruses and Opportunistic Infections.
Chicago, January–February 1999 [abstract 169].
29. Kaufmann D, Pantaleo G, Sudre P, Telenti A. CD4-cell count in HIV-1-infected individuals remaining viraemic with highly active antiretroviral therapy (HAART) [letter].
Lancet 1998, 351: 723 –724.
30. Piketty C, Castiel P, Belec L. et al. Discrepant responses to triple combination antiretroviral therapy in advanced HIV disease.
AIDS 1998, 12: 745 –750.
31. Chêne G, Binquet C, Moreau JF. et al. Changes in CD4+ cell count and the risk of opportunistic infection or death after highly active antiretroviral treatment.
AIDS 1998, 12: 2313 –2320.
32. Jacobson MA, Zegans M, Pavan PR. et al. Cytomegalovirus retinitis after initiation of highly active antiretroviral therapy.
Lancet 1997, 349: 1443 –1445.
33. Jacobson MA, French M. Altered natural history of AIDS-related opportunistic infections in the era of potent combination antiretroviral therapy.
AIDS 1998, 12 (suppl A): S157 –S163.
34. Foudraine N, Hovenkamp E, Notermans D. et al. Immunopathology as results of highly active antiretroviral therapy in HIV-1 infected patients.
AIDS 1999, 13: 177 –184.
35. Gilquin J, Piketty C, Thomas V. et al. Acute cytomegalovirus infection in AIDS patients with CD4 counts above 100
× 106 cells/I following combination antiretroviral therapy including protease inhibitors.
AIDS 1997, 11: 1659 –1660.
36. Race EM, AdelsonMitty J, Kriegel GR. et al. Focal mycobacterial lymphadenitis following initiation of protease-inhibitor therapy in patients with advanced HIV-1 disease.
Lancet 1998, 351: 252 –255.
37. Kaspar R, DuBois D. The clinical course of severely immunocompromised patients achieving a re-expansion of CD4 cells to above 50 cells/mm3 following antiretroviral therapy for HIV-1.Fourth Conference on Retroviruses and Opportunistic infections.
Washington DC, January 1997 [abstract 358].
The Clinical Epidemiology Group of the FHDH
S. Alfandari, F. Bastides, E. Billaud, A. Boibieux, F. Boué, A. Cabié, D. Costagliola, L. Cotte, L. Cuzin, F. Dabis, J.P. Daurès, V. Garrait, J.A. Gastaut, C. Gaud, S. Grabar, A. Goudeau, C. Katlama, D. Lacoste, J.M. Lang, H. Laurichesse, P. Leclercq, C. Leport, M.E. Mars, S. Matheron, M.C. Meyohas, C. Michelet, J. Moreau, C. Pradier, D. Quinsat, C. Rabaud, W. Rozenbaum, D. Salmon-Ceron, M. Sobesky, H. Tissot-Dupont, R. Verson, J.P. Viard, A. Waldner-Combernoux.
Data co-ordinating centre
INSERM SC4 (D. Costagliola, S. Grabar, L. Lièvre, L. Marrero, M. Mary-Krause, J.M. Tassie).
Paris area: CISIH de Bichat-Claude Bernard (Hôpital Bichat-Claude Bernard: S. Matheron, J.P. Coulaud, J.L. Vildé, C. Leport, L. Belarbi, Y. Bennai, M.M. Prevot, F. L'Heriteau), CISIH de Paris-Centre (Hôpital Broussais; G.H. Tarnier-Cochin: D. Sicard, D. Salmon; Hôpital Saint-Joseph: J. Gilquin, A. Cros), CISIH de Paris-Ouest (Hôpital Necker adultes; Hôpital Laennec; Hôpital de l'Institut Pasteur), CISIH de Paris-Sud (Hôpital Antoine Béclère; Hôpital de Bicêtre: C. Goujard, M.T. Rannou; Hôpital Henri Mondor: A.S. Lascaux, J.D. Magnier; Hôpital Paul Brousse: D. Vittecoq), CISIH de Paris-Est (Hôpital Rothschild; Hôpital Saint-Antoine: M.C. Meyohas, J.L. Meynard; Hôpital Tenon: C. Mayaud), CISIH de Pitié-Salpétrière (GH Pitié-Salpétrière: C. Katlama, M. Richard), CISIH de Saint-Louis (Hôpital Saint-Louis: J. Modaï, V. Garrait, J.P. Clauvel, L. Gerard; GH Lariboisière-Fernand Widal: V. Vincent, M.C. Mazeron), CISIH 92 (Hôpital Ambroise Paré: Rouveix E, Berthé H; Hôpital Louis Mourier: A.M. Simonpoli, C. Chandemerle; Hôpital Raymond Poincarré), CISIH 93 (Hôpital Avicenne: M. Bentata, P. Berlureau, B. Jarrousse, P. Cohen; Hôpital Jean Verdier; Hôpital Delafontaine).
Outside Paris area: CISIH Auvergne-Loire (CHU de Clermont-Ferrand: F. Gourdon, L. Cormerais; CHRU de St Etienne); CISIH de Bourgogne-Franche ComtE (CH de Belfort: P. Eglinger, J.P. Fall; CHRU de Besançon: C. Drobacheff, B. Hoen; CHRU de Dijon); CISIH de Caen (CHRU de Caen: C. Bazin, R. Verdon), CISIH de Grenoble (CHU de Grenoble: O. Bouchard, P. Leclercq), CISIH de Lyon (Hôpital de la Croix-Rousse; Hôpital Edouard Herriot; Hôtel-Dieu: C. Trepo, L. Cotte; CH de Lyon-Sud), CISIH de Marseille (Hôpital de la Conception; Hôpital Houphouët-Boigny: J. Moreau; Institut Paoli Calmettes; Hôpital Sainte-Marguerite: J.A. Gastaut, I. Poizot-Martin, J.J. Grob, M.A. Richard; Hôtel-Dieu: A. Boussuges, J.M. Sainty; CHG d'Aix-En-Provence; CH d'Arles; CH d'Avignon: G. Lepeu, O. Boulat; CH de Digne Les Bains: P. Granet-Brunello; CH de Gap: L. Pelissier, J.P. Esterni; CH de Martigues: M. Nezri, J. Ruer; CHI de Toulon), CISIH de Montpellier (CHU de Montpellier: J. Reynes, V. Baillat; CHG de Nîmes), CISIH de Nancy (Hôpital de Brabois), CISIH de Nantes (CHRU de Nantes: F. Raffi, M.F. Charonnat), CISIH de Nice (Hôpital Archet 1: C. Pradier, P.M. Roger; Hôpital Archet 2; Hôpital Cimiez; Hôpital Pasteur; Hôpital Saint-Roch; CHG Antibes Juan les Pins), CISIH de Rennes (CHU de Rennes), CISIH de Rouen (CHRU de Rouen: Y. Debab, F. Caron), CISIH de Strasbourg (CHRU de Strasbourg: M. Jollinier, V. Walter; Hôpital de Haute-Pierre: P. Fraisse; CH de Mulhouse), CISIH de Toulouse (CHU Purpan: L. Cuzin; Hôpital la Grave; CHU Rangueil), CISIH de Tourcoing-Lille (CH Gustave Dron; CH de Tourcoing: Y. Mouton, S. Alfandari), CISIH de Tours (CHRU de Tours: F. Bastides, J.M. Besnier; CHU Trousseau).
Overseas: CISIH de Guadeloupe (CHRU de Pointe-à-Pitre), CISIH de Guyane (CHG de Cayenne: M. Sobesky, P. Couppie), CISIH de Martinique (CHRU de Fort-de-France: G. Comlan-Mayaud, R. Cesaire), CISIH de La Réunion (CHD Félix Guyon).