Papers

Is the clinical course of HIV-1 changing? Cohort study

BMJ 1997; 314 doi: https://doi.org/10.1136/bmj.314.7089.1232 (Published 26 April 1997) Cite this as: BMJ 1997;314:1232
  1. A Sinicco, senior lecturer in infectious diseasesa,
  2. R Fora, registrara,
  3. R Raiteri, research statisticiana,
  4. M Sciandra, research statisticiana,
  5. G Bechis, registrara,
  6. M M Calvo, registrara,
  7. P Gioannini, directora
  1. a Department of Medical and Surgical Sciences, Section of Infectious Diseases, University of Turin, Amedeo di Savoia Hospital, 10149 Turin, Italy
  1. Correspondence to: Dr Sinicco
  • Accepted 23 January 1997

Abstract

Objective: To assess whether the clinical course of HIV infection has changed from 1985 to 1995.

Design: Cohort study.

Setting: Infectious diseases clinic.

Subjects: 285 patients recruited from September 1985 to January 1995 with ≤12 months between the dates of their last seronegative and first seropositive test result and with first follow up visit in the six months after seroconversion and at least 12 months' follow up. Patients were grouped according to the date of seroconversion.

Main outcome measures: Time to CD4 cell count of <500, 400, and 200 x 106 cells/l and clinical outcome defining AIDS; variation in cell count per day between consecutive visits, and ratio between this variation and time from estimated date of seroconversion at each visit.

Results: The groups were similar in age, number with acute primary HIV infection, CD4 cell count at intake, and cell count at the beginning of antiretroviral treatment; they differed in sex ratio, risk factors for HIV, probability of CD4 cell decline to <500, 400, and 200x106 cells/l, and risk of developing AIDS. Acute infection, seroconversion after December 1989, and serum ß2 microglobulin >296 nmol/l were independent predictors of poor clinical course. The speed of CD4 cell decline, expressed as cell variation divided by the number of days between consecutive visits, increased with more recent seroconversion (P=0.02). Ratio between the speed of CD4 cell decline and time from estimated date of seroconversion at each visit was also higher in the patients who seroconverted after December 1989.

Conclusions: The faster disease progression and the higher speed of CD4 cell decline at early stages in the patients with recently acquired HIV infection suggest changes in the clinical course of HIV infection.

Key messages

  • Interest in possible changes in the course of HIV infection has recently increased

  • Previous research has shown no clear trend for changes in CD4 cell count by interval after HIV seroconversion

  • Results from a large and heterogeneous cohort of patients who seroconverted between September 1985 and January 1995 showed that the patients who seroconverted after December 1989 had a higher probability of decline in CD4 cell count and progression to AIDS than did patients who had seroconverted before this date

  • The overall rate of decline in CD4 cell count was higher in patients who seroconverted after December 1989; 180 days after seroconversion the rate was highest in those who seroconverted after December 1989, and 360 days after seroconversion it was highest in those who seroconverted after December 1989 and before January 1991

  • Repeated monitoring within the first months after HIV seroconversion is needed to identify those patients who could benefit from early antiretroviral treatment

Introduction

Increasing interest has recently focused on possible secular changes in the course of HIV infection.1 2 3 4 5 6 7 In particular, researchers are interested in whether the calendar year of HIV seroconversion is associated with a specific pattern of CD4 cell trends and clinical evolution.

Some authors have found no change.4 5 Findings in military staff with HIV infection have shown a rapid fall in CD4 cell count soon after seroconversion but no clear trend for changes in cell count by fixed time after seroconversion.2 3 Similarly, a study of patients who seroconverted between 1984 and 1991 showed neither change in the course of infection nor an association between calendar year of seroconversion and time to CD4 cell count of <500x106 cells/l.4

There are, however, several limitations in these studies. Firstly, the study period was limited and included only the early 1990s. Consequently, recent changes in the course of HIV disease were not examined. Secondly, the study population usually comprised homosexual or bisexual men, with little or no information on other risk groups; analysis restricted to only one transmission category could lead to biased conclusions. Thirdly, most investigations have been based on CD4 cell counts determined in different laboratories. Hence, despite the improvements obtained by adjustment, interlaboratory variations might be responsible for measurement bias. Finally, little is known about CD4 cell counts before seroconversion and CD4 counts at the start of antiretroviral treatment. Several factors may complicate the estimate of the impact of the treatment interventions in patients infected with HIV (for instance, use of different drugs and regimens, use of associated treatments, dissimilar duration of treatment), but analytical approaches that do not consider the CD4 cell count relative to the initiation of antiretroviral treatment might lead to biased results concerning the progression of HIV disease.

Because changes in the course of HIV infection could affect clinical management of newly infected patients, as well as having repercussions on the global epidemiology of HIV infection and on healthcare resources, we examined whether the course of HIV disease has changed in recent years. We divided our 10 year cohort of 285 patients positive for HIV antibody with known date of seroconversion into four groups according to the calendar date of their seroconversion, and we then compared the probabilities of CD4 lymphocyte count of <500, 400, and 200x106 cells/l, progression to AIDS, and the speed of decline of CD4 cell count.

Patients and methods

Recruitment and inclusion criteria

From September 1985 to January 1996, 4134 patients at risk for hepatitis, sexually transmitted diseases, and HIV infection were tested for antibody to HIV at the clinic of infectious diseases, University of Turin. Those with positive test results who had had negative results at some time in the previous 12 months were enrolled in a prospective study to evaluate the course of HIV infection. Additional criteria for inclusion were a first CD4 cell count within six months of the first positive test results and a follow up of at least 12 months after entry. The 285 cohort members were divided into four groups according to their date of seroconversion: 75 subjects seroconverted between September 1985 and December 1987, 60 between January 1988 and December 1989, 69 between January 1990 and December 1991, and 81 between January 1992 and January 1995.

Definitions and study outcomes

In patients who had no symptoms at conversion we assumed the date of entry (estimated date of seroconversion) to be the midpoint between the dates of the last negative and the first confirmed positive test results. In the patients with acute primary HIV infection the entry was considered as the date of the beginning of the symptoms.

The end points were the dates of the first confirmed CD4 count of <500, 400, and 200x106cells/l and the date of the first clinical outcome defining AIDS. Confirmed decline of CD4 cell count was defined as a decrease to below the relevant cell count determined in three consecutive samples. AIDS patients recorded in this report fulfilled the revised AIDS case definition.8 To implement the analysis and to limit time lag effects we performed a cross check with the national AIDS registry.

Laboratory analysis

Antibody to HIV was assessed by enzyme immunoassay, with confirmation by western blot. HIV antigen (p24) was detected by commercially available enzyme immunoassay. CD4 lymphocyte subsets were measured by flow cytometry in a laboratory that used identical analytic procedures.

Statistical analysis

To assess differences between the four groups positive for HIV we used Kruskal-Wallis one way analysis of variance for the continuous variables and χ2 tests for the categorical variables. Results for continuous variables are expressed as median (range). To estimate the progression rates we used the Kaplan-Meier survival method, and comparison between progression curves was tested for significance with log rank test for two sample comparison and with Gehan's test or Breslow's test for multiple sample comparison.9 A multivariate analysis was carried out to identify the independent cofactors of disease progression by using Cox's proportional hazards model to calculate the hazard ratio of reaching the end points. The variables included in the hazards analysis were sex, age, education, occupation, transmission category, smoking status, drinking status, life style, annual income, use of minor tranquillisers, duration of injecting drug use, history of sexually transmitted diseases, history of primary acute HIV infection, calendar time of seroconversion to HIV, baseline laboratory measurements (CD4 and CD8 cell count, serum IgA, ß2 microglobulin and 5′-neopterin concentrations, and HIV p24 concentration). A backward stepwise selection of the covariates was used in constructing the model. The statistical criterion used to select the best model was the partial likelihood ratio test. The variable “calendar time of seroconversion” was divided into four dummy variables representing the four study periods of seroconversion. As a preliminary step, only the resulting four groups were included in the Cox's proportional hazards analysis, and we tested three groups against the reference group. Afterwards we included all the study variables in the analysis, using the September 1985 to December 1987 group as the reference group. Data were processed with statistica version 5.010 and spss version 6.0.11

Results

From September 1985 to January 1995, 285 patients were included in the study. The median age at seroconversion was 27 (range 17-61) years; 208 (73%) were men. According to the risk factors 163 (57%) were injecting drug users, 64 (22%) male homosexuals, 56 (20%) heterosexuals, and 2 (1%) had other risk factors. Fifty one patients (18%) had acute primary HIV infection. The median time in days was 182 (range 21-337) between the last negative and the first positive test result; 1710 (363-3672) for length of follow up; 91 (1-168) between the first positive test result and first follow up visit; and 83 (36-182) for the interval between visits, which was similar in the different risk groups.

Table 1) shows the baseline characteristics of the four groups of patients divided according to their date of seroconversion date. When we compared the baseline characteristics of the four groups we found differences in sex and in transmission by injecting drug use and by heterosexual contacts (P=0.019, P=0.0010, and P=0.00003 respectively). HIV infection through drug addiction and heterosexual sex were symmetrically distributed throughout the study period. In the mid-1980s most of those who seroconverted were men, but the number of subjects infected heterosexually and the number of women, particularly those infected heterosexually, increased progressively. At the first visit, 54 patients (19%) had circulating HIV p24 antigen >10 ng/l. The groups did not differ in the intervals between the last negative and the first positive test result or in the temporary loss to follow up or in socioeconomic, lifestyle, and laboratory measurements. Overall, 110 (60%) patients began a course of antiretroviral treatment at some time during follow up. The four groups did not differ in CD4 cell count at the start of antiretroviral treatment, and all subjects used nebulised pentamidine as prophylaxis against Pneumocystis carinii pneumonia when the CD4 count fell to <200 x 106 cells/l. CD4 cell counts from before seroconversion were available in 53 patients (19%), and the groups did not differ in the CD4 cell count before seroconversion. Thirty patients (46%) who seroconverted between September 1985 and December 1987, 14 (22%) between January 1988 and December 1989, 17 (26%) between January 1990 and December 1991, and 4 (6%) between January 1992 and January 1995 developed AIDS.

Table 1

Baseline characteristics of the four groups of patients with HIV infection according to different time of seroconversion. Values are numbers (percentages) of patients unless stated otherwise

View this table:

When we compared the probability curves of CD4 lymphocyte count falling to <500, 400, and 200x106cells/l and AIDS progression, patients who seroconverted after December 1989 showed earlier declines and faster progression than did those who seroconverted before (P<0.0001, P=0.0001, P=0.0037, and P=0.19, respectively). Figure 1) shows the probability curves of CD4 lymphocyte count falling to <500, 400, and 200 x 106cells/l and AIDS progression in the four groups. Table 2) shows the cumulative estimates of the cell count falling to the end points and the progression to AIDS by time intervals from the seroconversion in the four groups. Table 3 and Table 4 show the log rank test P value and Cox's model P value relative to the group by group comparison of the probability of CD4 cell decline and progression to AIDS.

Fig 1

Survival curves for subjects according to period since estimated time of seroconversion to HIV expressed as probability of not having condition

Fig 1

Survival curves for subjects according to period since estimated time of seroconversion to HIV expressed as probability of not having condition

Fig 1

Survival curves for subjects according to period since estimated time of seroconversion to HIV expressed as probability of not having condition

Fig 1

Survival curves for subjects according to period since estimated time of seroconversion to HIV expressed as probability of not having condition

Table 2

Cumulative estimates of decline in CD4 cell count and progression to AIDS in four groups of patients with HIV infection according to time of seroconversion.* Values are percentages of patients

View this table:
Table 3

Matrix of P values in log rank test and Cox's proportional hazards analysis with hazards ratios (95% confidence intervals)* for group by group comparison of probability of decline to CD4 cell count <400 × 106/l (top right) and <500 × 106/l (bottom left)

View this table:
Table 4

Matrix of P values in log rank test and Cox's proportional hazards analysis with hazards ratios (95% confidence intervals)* for group by group comparison of probability of decline to CD4 cell count <200 × 106/l (top right) and progression to AIDS (bottom left)

View this table:

Cox's proportional hazards models confirmed seroconversion to HIV after December 1989, seroconversion associated with acute primary HIV infection, and serum ß2 microglobulin >296 nmol/l at entry to be the most significant independent predictors of poor evolution (table 5).5 Patients who seroconverted through intravenous drug use and those who acquired HIV sexually did not differ in the CD4 cell decline and progression to AIDS, and this finding was confirmed in the multivariate model.

Table 5

Cox's proportional hazards analysis of factors associated with different biological and clinical end points at entry among 285 patients who seroconverted between September 1985 and January 1995. Values are hazard ratios (95% confidence intervals)

View this table:

The rate of CD4 cell decline, expressed as a ratio of CD4 cell variation to the time between visits in days was different in the four groups, with higher values in those who seroconverted after December 1989 (hazard ratio 0.48 (95% confidence interval 0.0065 to 9.21)v 0.50 (0.0025 to 6.71) v 0.65 (0.0068 to 28.9) v 0.53 (0.13 to 22.15); P=0.020). Likewise, the ratio between the rate of CD4 cell decline and time from estimated date of seroconversion at each visit differed in the four groups (1.52x10−3 v 1.145x10−3 v 3.317x10−3v 4.97x10−3, P=0.000042). When we compared patients who seroconverted more recently with those who seroconverted before December 1989, the recently infected subjects showed higher rates of CD4 cell decline at the shortest intervals from the date of estimated date of seroconversion.

The initial CD4 cell loss was different in the four groups (fig 2). In particular, the difference between groups in the rate of daily decline of CD4 cell count at 180 days after estimated date of seroconversion was different, being higher in the patients who seroconverted after December 1989 (1.33 (0.45 to 5.19) cells/dayv 2.25 (0.74 to 3.86) v 3.19 (0.20 to 9.19) v 4.52 (0.10 to 22.15); P=0.025). The rate of daily CD4 cell decline at 360 days after estimated date of seroconversion also differed, being higher in the patients who seroconverted between January 1990 and December 1991 (1.04 (0.09 to 5.52) cells/day v 0.64 (0.05 to 1.99) v 2.0 (0.36 to 28.9) v 0.94 (0.01 to-3.7); P=0.0034).

Fig 2
Fig 2

Rate of daily loss of CD4 cells according to estimated time since seroconversion

Discussion

To study the possible influence of the calendar time of seroconversion on the course of HIV infection we assessed whether the subjects who seroconverted in different time periods between September 1985 and January 1995 had distinct patterns of disease progression. Although all patients had similar immunological characteristics at entry, CD4 cell counts started differing at early stages of the infection, and this finding was solely associated with the higher speed of decline of CD4 cell counts in those who seroconverted after December 1989. Consistent with previous reports, our data show that the first 12 months after seroconversion are extremely critical for the future course of HIV disease and that a higher rate of daily depletion of CD4 cells within the first year of the infection distinguishes those who seroconverted after December 1989.7

Reducing bias

To reduce potential sources of bias in the evaluation of the course of HIV disease, we accurately controlled the baseline characteristics of the patients with different calendar times of seroconversion. Similarly, to avoid selection bias we controlled the composition of the population referred to our unit in the study period. The patients who comprised the cohort did not differ in sociodemographic and behavioural factors from the other patients who were found to be positive for HIV in the same period.

The four groups shared similar characteristics at entry but differed in sex ratio and risk factors to HIV. This intergroup discrepancy was consistent with the epidemiological changes of the HIV epidemic in Italy throughout the study period, butt these differences at intake did not seem to affect the final analysis. As shown in the multivariate analysis, male sex was a non-significant predictor of poor evolution. Although there was a higher proportion of men in those who seroconverted before January 1988, this covariate did not influence the progression of this group, and seroconversion after December 1989 remained the most important independent factor associated with faster progression of HIV disease. Likewise, sexual transmission was unrelated to the evolution of the disease in our cohort. Moreover, when we considered December 1989 as a cut off date of seroconversion, the resulting two groups of positive subjects, similar also in sex ratio and category of exposure, showed different rates of CD4 cell decline and progression to AIDS. Seroconversion after December 1989 was consistent with a more severe disease course.

The heterogeneous composition of the sample with respect to the risk factors, the adequate size of the four groups, and the satisfactory length of the follow up support the validity of the results. Further, the unchanged site and method of determination of CD4 cell count, the partial availability of cell counts from before seroconversion, the knowledge of cell counts relative to the start of antiretroviral treatment, and the regular monitoring of cell counts from the initial period after seroconversion to very late HIV disease argue for reliability of the data. Finally, the exclusion from the study of the patients with >12 months between the last negative and the first positive test result for HIV infection contributed to the prevention of any error in calculation of progression rates.

Possible limitations

Notwithstanding these attempts to contain potential biases, some limitations should be acknowledged. First and foremost, few virological data—apart from p24 antigenaemia, a rough marker of HIV attributes—were known for our patients. In addition, the value of the presence of circulating HIV antigen at entry as an early predictor of rapid progression was confounded by the association of p24 antigenaemia at entry with the history of acute primary HIV infection in our sample.

The characteristics of HIV play an important part in the course of the disease. Studies suggest that HIV isolates vary in their cellular host range, tropism, and pathogenic potential.12 13 14 15 16 Furthermore, viral changes often occur throughout the course of HIV infection.17 18 Recent research supported the association between more rapid disease progression and certain viral phenotypes, such as syncytium inducing strains as opposed to non-syncytium strains.19 20 Moreover, it is well known that the selective pressure of antiretroviral treatment encourages the appearance of resistant variations, sometimes soon after start of treatment, interfering with immune responses and coinciding with failure of treatment.21 Accordingly, there is the possibility of primary acquisition of viruses resistant to the antiretroviral treatments in newly infected patients.22 23 24 25 Finally, accumulating evidence indicates that diverse HIV subtypes are spreading to regions of previously restricted genetic diversity.26 Although data were not available, some mechanisms of those previously reported, alone or together, may account for the increased rates of immunological impairment and progression to AIDS in patients who had recently seroconverted.

With the exclusion of drugs used in antiretroviral treatment, no drugs known to have an appreciable impact on CD4 cells were extensively used in the study population, but we do not know whether methadone treatment affects HIV disease.27 28 This potential bias could be important for the final analysis because of the large proportion of injecting drug users in our cohort; however, it would have been minimised by the fact that those who seroconverted had the same access to methadone treatment as those who did not.

Thirdly, another confounder could be laboratory drift of the CD4 cell determinations over time.29 30 31 Within the laboratory, however, CD4 measurements at six month intervals in a healthy control group showed no variations, and CD4 cell counts remained unchanged among subjects acting as controls.

In the multivariate model, seroconversion after December 1989 and acute primary HIV infection were confirmed as the most important independent factors of disease progression.32 In the Italian seroconversion study, older age at seroconversion was associated with a faster progression to AIDS.33 This discrepancy with our data could be explained by the different composition of our cohort.

Conclusions

The emergence of more virulent strains due to multiple biological mechanisms may be responsible for more aggressive course of HIV disease in patients who have recently seroconverted. Our findings suggest possible changes in the course of HIV epidemic in the 1990s and raise intriguing issues on the course of HIV infection. If our data are confirmed, therapeutic approaches to the infection will need to be reviewed. In particular, if HIV disease has become more aggressive, more frequent screening would be essential to identify patients who have just seroconverted and could benefit from early antiretroviral treatment.

Acknowledgments

Funding: No external funding.

Conflict of interest: None.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
View Abstract