Thrombocytopenia (TCP) is reported in 3-20% of human immunodeficiency virus-1 (HIV-1)-infected subjects (1-5) but, although low platelet (PLT) counts are often observed, severe hemorrhaging is rare (6). The cause of HIV-1-associated TCP is still under investigation, but most studies suggest an immunological pathogenesis sustained by immune complexes or specific antiplatelet autoantibodies (7-10), probably generated by cross-reactivity between the common determinants of HIV-1 and human PLT (11,12). A direct role of the virus in inducing TCP is supported by the observation that the antiretroviral agent, zidovudine, is effective in increasing PLT counts and reducing the occurrence of hemorrhages in HIV-1-infected patients (13-17). Finally, the drug toxicity and infections that are common in HIV-1-infected subjects can at least partially justify the TCP observed in clinical practice (18). Because none of these potential causes excludes another, the origin of TCP might be multifactorial in some patients.
The complexity of the problem is further highlighted by the contradictory results reported concerning the relation between TCP and the progression of HIV-1 infection. Various studies have concluded that TCP is a predictor of progression to overt AIDS (4,19,20) and that low PLT counts are associated with low levels of CD4+ cells in peripheral blood (PB) (21), but others have failed to show any relation between PLT counts and the progression of HIV-1 infection (22-24). This disagreement may be partially explained by differences in the definition of TCP and the lack of homogeneity of the included patients in terms of disease severity and risk factors. For these reasons, we performed a longitudinal study of a large cohort of patients at different stages of disease progression and with different risk factors of infection to investigate some of the potential determinants of TCP and its relation with the development of acquired immune deficiency syndrome (AIDS).
Since 1983, all HIV-1-infected patients attending the outpatient department of the University of Milan's Clinic of Infectious Diseases have been consecutively enrolled in a cohort study and invited to undergo examinations and laboratory tests every 3 months. The patients were first classified according to the way in which they acquired the infection: intravenous drug use (IVDU), male homosexual intercourse, heterosexual contacts (heterosexual steady partners of HIV-1-infected people, subjects reporting multiple heterosexual partners or sexual contacts with prostitutes), others (blood and blood derivative recipients and transplanted individuals), and unknown (subjects not reporting any specific risk factor). At enrollment and during follow-up, the subjects were further classified according to demographic variables and the 1987 Centers for Disease Control (CDC) criteria (25). CD4+ cell and PLT counts were determined at enrollment and at each visit. The PLT counts were grouped into four categories: <50 × 109/L, 51-100 × 109/L, 101-150 × 109/L; and >150 × 109/L. TCP was defined as a PLT count ≤100 × 109/L; PLT counts <50 and counts from 51 to 100 × 109/L were considered indicative of severe and moderate TCP, respectively. PLT counts >150 × 109/L were considered normal, and counts from 100 and 150 × 109/L, were considered of borderline significance. The CD4+ cell counts at enrollment were grouped into four categories: ≤200/μl, 201-400/μl, 401-600/μl, and >600/μl.
Subjects enrolled after July 1991 and those with AIDS at enrollment were excluded from the analysis. The end date of this study was July 1992, and none of the events occurring after that date were considered in the analysis. The statistical significance of differences in proportions was evaluated by Chi-square test, the cumulative probabilities of the occurrence of time-dependent variables (AIDS and TCP) were evaluated by the Kaplan-Meier method, and statistical significance was calculated by the log-rank test. To take into account all potential covariates involved in the occurrence of AIDS and TCP, a multivariable analysis was performed with Cox's proportional hazard model.
Since 1988, the year in which zidovudine (ZDV) became available in Italy, attendant physicians have considered the possibility of ZDV treatment for all patients with CD4+ cell counts <500/μl. The appearance of HIV-1-related symptoms, the persistence of HIV p24 antigen positivity, and a consistent decrease in CD4+ cell counts led to the decision to initiate the treatment; moreover, ZDV was administered to most patients with TCP regardless of any signs or symptoms of the progression of HIV-1 infection. Because ZDV slows the progression to overt AIDS, it might confound the true relation between a reduced PLT count and progression to AIDS; consequently, the analysis of the time-dependent probability of the occurrence of AIDS was also made separately for patients who had been treated with ZDV >3 months during follow-up and for those who had not received this treatment or who had received it <3 months.
The study involved 1,533 patients followed for a median of 21 months. The main features of the study population are shown in Table 1. Most of the enrolled subjects were males (70.8%) aged <25 years (51.1%) who had acquired the infection as a result of IVDU (71.4%). The four classes of CD4+ cell counts were equally represented, with about one fourth of the population in each category. The majority of subjects had normal PLT counts (1,122 of 1,533, 73.2%), 81 (5.9%) had severe and 91 had (5.3%) moderate TCP, and 239 patients (15.6%) had borderline PLT counts.
PLT counts were more often normal in females (80.6%) than in males (70.1%) and in younger patients. TCP was significantly more frequent in IVDU patients (12%) and male homosexuals (11.7%) than in subjects who acquired the infection through heterosexual contacts (4.5%). Moderate and borderline TCP were clearly associated with CD4+ cell levels, occurring more frequently in patients with lower counts; on the contrary, severe TCP was equally represented in patients with different CD4+ cell counts at enrollment.
During follow-up, eight patients with normal PLT counts died without having developed overt AIDS of causes unrelated to HIV-1 infection (3 overdoses, 1 car accident, 1 suicide, and 1 of liver cirrhosis, and 2 of undetermined, probably drug-related, causes).
Overt AIDS occurred in 22.8% (350 of 1,533) of the patients, being more common in those with moderate TCP (31.9%) or borderline PLT counts (33.1%) (Table 2). The AIDS-defining diseases according to PLT counts at enrollment are shown in Table 3. Although Candida esophagitis and AIDS dementia complex (ADC) were slightly more frequent in subjects with low PLT counts and although Kaposi's sarcoma (KS) was more frequent in subjects with normal PLT, no clear evidence of a different distribution in AIDS-defining diseases was observed among the four groups.
Of the 1,361 patients who had PLT counts >100 × 109/L at enrollment, moderate or severe TCP was observed during follow-up in 137 (10.1%). None of these occurrences was attributable to acute drug toxicity. The occurrence of TCP during follow-up was more common in subjects with borderline TCP at enrollment (27.4%) than in those with normal PLT counts (6.5%). ZDV was administered to ≈40% of the patients during follow-up; 541 (33.5%) received the drug >3 months. ZDV more commonly was given to patients with severe TCP and those with borderline PLT counts (44.5 and 46.9%, respectively), followed by those with moderate TCP or normal PLT counts (34.1 and 29.9%, respectively).
For the entire cohort, the cumulative probability of AIDS was 5% after 1 year, 16% after 2 years, 24% after 3 years, and 35% after 4 years of follow-up. The probability of AIDS during follow-up according to PLT count at enrollment is shown in Fig. 1. After 24 months, patients with normal PLT counts and those with severe TCP at enrollment had the same probability of developing overt AIDS (13 and 14%, respectively). The risk of AIDS was significantly greater in patients with moderate TCP and in those with borderline PLT counts (32 and 24%, respectively). These differences were even more evident after 48 months: the patients with normal PLT counts and those with severe TCP had a cumulative probability of developing overt AIDS of 30 and 24%, respectively, those with moderate TCP and those with borderline values had probabilities of 64 and 62%, respectively, and these differences were also marked after 60 months of follow-up. With the subjects with normal PLT counts as the reference group, the relative risk of developing AIDS was 0.8 (95% CI 0.5-1.3, p = 0.4) for subjects with severe TCP, 2.1 for those with moderate TCP (95% CI 1.4-3.1, p = 0.002), and 1.6 for those with borderline PLT values (95% CI 1.2-2.1, p = 0.0004).
The cumulative probabilities of developing overt AIDS according to PLT counts at enrollment in patients not treated with ZDV and those treated for at least 3 months are shown in Fig. 2. In untreated individuals, the probability was higher in patients with moderate TCP or borderline PLT counts than in those with severe TCP or normal PLT counts. After 2 years, the probability was 19% in patients with moderate TCP, 17% in those with borderline PLT values, 10% in those with severe TCP, and 6% in those with normal PLT counts. After 48 months, the corresponding figures were 39, 35, 21, and 16%. With the subjects with normal PLT levels at enrollment as the reference category, the risk of AIDS was 1.1 (95% CI 0.5-2.5, p = 0.8) in subjects with severe TCP, 2.9 (95% CI. 1.6-5.5, p = 0.0009) in subjects with moderate TCP, and 1.9 (95% CI. 1.2-3.0, p = 0.0043) in subjects with borderline PLT values. Among the subjects who were treated with ZDV, those with severe TCP had the lowest and those with moderate TCP had the highest risk of progression to AIDS. After 2 years, the probability of AIDS was 50% in patients with moderate TCP, 28% in those with borderline PLT values, 17% in those with severe TCP, and 25% in those with normal PLT counts. After 48 months, the corresponding figures were 70, 63, 27, and 52%. With the subjects with normal platelet levels at enrollment as the reference category, the risk of AIDS was 0.5 (95% CI 0.3-0.9, p = 0.0426) in subjects with severe TCP, 1.5 (95% CI 0.9-2.5, p = 0.1) in subjects with moderate TCP, and 1.4 (95% CI 0.8-1.6, p = 0.3) in subjects with borderline PLT values.
Multivariable analysis (Table 4) showed that the risk of AIDS was inversely proportional to the number of CD4+ cells in PB and that it was higher in older patients and in those who were treated with ZDV. No difference was observed between the two sexes or the different ways in which the infection was acquired. With subjects with normal PLT counts as the reference group, no significant difference was observed among the different categories of PLT counts; however, the risk of subjects with severe TCP was reduced by 40% [risk ratio (RR) = 0.6, 95% CI 0.4-1.1, p = 0.0602]. Moreover, with the subjects with severe TCP at enrollment as the reference group, a significantly higher risk of developing AIDS was noted in patients with moderate TCP (RR = 2.2, 95% CI 1.2-4.3, p = 0.0182) and borderline PLT counts (RR = 2.0, 95% CI 1.1-3.5, p = 0.0242). No significant interaction was observed between PLT counts at enrollment and the other variables considered in the multivariable model.
The cumulative time-dependent probability of developing TCP according to PLT count at enrollment is shown in Fig. 3. The risk was clearly associated with PLT level at enrollment: after 24 months, the cumulative probability of TCP was 25% in patients with borderline PLT counts and 5% in patients with normal values; the corresponding figures were 44 and 14% after 48 months. At multivariable analysis (Table 4), the risk of the occurrence of TCP was significantly associated with CD4+ cell counts and PLT level at enrollment. The risk of developing TCP in patients with borderline PLT, in relation to that of patients with normal PLT counts, was 4.7 (95% CI 2.9-6.0, p = 0.0001) and was significantly reduced in patients with CD4+ cell counts >200/μl.
Our results confirm that TCP is common in patients infected with HIV-1. We observed PLT counts ≤100 × 109/L in >10% of our 1,533 consecutive patients, and 10% of the patients with PLT counts >100 × 109/L at enrollment developed TCP after a median follow-up of 22 months. Borderline PLT counts (<150 × 109/L but >100 × 109/L) were also quite frequent, observed in ≈15% of the population at enrollment. PLT counts ≤50 × 109/L (defined as severe TCP) were observed in ≈5% of the patients at enrollment. Severe TCP was not associated with immunodepression, being equally represented in patients with both low and high CD4+ cell counts. On the contrary, moderate TCP (PLT >50 × 109/L but ≤100 × 109/L) and borderline PLT counts were more common in the patients with low CD4+ cell counts.
The risk of developing overt AIDS during follow-up was higher in the patients with moderate TCP or borderline PLT values than in those with severe TCP or normal PLT levels. Moreover, the higher risk of patients with moderate TCP or borderline PLT levels observed in the crude data was substantially lost at multivariable analysis, which also showed that older age, low CD4+ cell counts, and ZDV treatment were associated with a greater risk of developing AIDS. This result suggests that a part of the risk associated with PLT levels is due to a spurious association which is lost when other prognostic predictors of progression (such as CD4+ cell counts) are taken into account. Moreover, the probability of developing TCP during follow-up was higher in subjects with borderline PLT counts (ranging from 101 to 150×109/L) and with low CD4+ cell counts at enrollment. Other factors, such as age, sex, and ZDV treatment, had no effect on the probability of the occurrence of TCP.
An association between low CD4+ cell counts and TCP has been reported by other investigators (21,26), who have suggested that TCP may parallel the progressive immunodepression induced by the infection. On the contrary, at least one study has failed to show that severe TCP has any relevance in terms of the risk of progression to AIDS (24). One possible explanation for these discrepancies lies in the different criteria adopted for defining TCP. Peltiers et al., who considered PLT values <150 × 109/L indicative of TCP and studied a cohort of patients with a low prevalence of PLT counts <50 × 109/L (1%, 7 patients), reported a positive association between immunodepression and TCP (21). Our study showed similar results in patients with a borderline to moderate decrease in PLT counts (< 150 > 50) but, in agreement with results of Cameron and Flegg (24), we also noted that patients with severe TCP (≤50 PLT × 109/L) have the same risk of developing AIDS as patients with normal PLT values.
The lower risk of developing overt AIDS and the higher number of CD4+ cells at enrollment (in comparison with patients with moderate TCP or borderline PLT values) suggest that severe TCP tends to occur early during HIV infection and that more than one pathogenetic mechanism probably is involved. This hypothesis is also supported by various kinetic studies of HIV-1-associated TCP, which show a pattern of PLT survival that may range from those observed in patients with idiopathic TCP to those observed when TCP patients are pooled (5,27,28). Idiopathic TCP has an autoimmune pathogenesis and is similar in clinical and laboratory terms to the severe TCP observed in HIV-1-infected subjects, a similarity which suggests that severe TCP may also have an autoimmune origin. On the contrary, in patients with advanced disease, TCP might be the result of reduced PLT production due to the inhibiting influence of the virus on bone marrow (29,30), the direct infection of megakaryocytes (31,32), or a specific inhibiting effect on megakaryocyte activity that is not sustained by direct viral infection (33,34).
In agreement with results of previous studies (5,21,35), in our study the prevalence of TCP at enrollment was higher among IVDU patients than in the other risk groups. In IVDU patients, several factors other than HIV-1 infection might contribute the development of TCP. Some studies have suggested that such infection may be caused by a direct toxic effect of heroin or adulterants (36,37); hepatitis C virus infection (which is highly prevalent in IVDU patients) may cause TCP and may also be associated with TCP in HIV-1-infected patients (38). Therefore, the higher prevalence of TCP reported among IVDU patients' can be at least partially explained by the higher prevalence of HCV infection with liver impairment.
PLT counts at enrollment highlight various probabilities of disease progression, with the highest risk of developing AIDS during follow-up observed in subjects with moderate TCP. The existence of different forms of HIV-1-associated TCP suggested by our study and other studies (5,18,29) should be considered when future investigations of the pathogenesis and therapy of these disorders are designed.
Acknowledgment: This work was supported by the Italian Ministry of Health ISS-AIDS Research Project Contracts No. 42051, 520503, 6205007, 72/0304, and by the Lombardy Region Research Project 774. S. R., A. R., and A. L. were supported by an AIDS fellowship from the Italian National Institute of Health (ISS), Rome, Italy. We thank Dr. Antonella d'Arminio Monforte for helpful discussion and Linda Vicini, Bianca Ghisi, and Tiziana Formenti for secretarial support.
1. Kaslow RA, Phair JP, Friedman HP, et al. Infection with the human immunodeficiency virus; clinical manifestations and their relationship to immunodeficiency. A report from the multicentre AIDS Cohort Study. Ann Intern Med
2. Jost J, Tauber MG, Luthy R, Sieghentaler W. HIV-assozierte Thrombocytopenie. Schweiz Med Wochenschr
3. Murphy MF, Metcalfe P, Waters AH, et al. Incidence and mechanism of neutropenia and thrombocytopenia
in patients with human immunodeficiency virus infection. Br J Haematol
4. Rossi G, Gorla R, Stellini R, et al. Prevalence, clinical and laboratory features of thrombocytopenia
among HIV-infected individuals. AIDS Res Hum Retrovir
5. Landonio G, Galli M, Nosari A, et al. HIV-1
related severe thrombocytopenia
in intravenous drug users: prevalence, response to therapy in a medium-term follow-up and pathogenetic evaluation. AIDS
6. Finazzi G, Mannucci PM, Lazzarin A, et al. Low incidence of bleeding from HIV-related thrombocytopenia
in drug addicts and haemophiliacs: implication for therapeutic strategies. Eur J Haematol
7. Walsh CM, Nardi MA, Karpatkin S. On the mechanism of thrombocytopenic purpura in sexually active homosexual man. N Engl J Med
8. Yu JR, Lennette ET, Karpatkin S. Anti-F(ab') antibodies in thrombocytopenic purpura patients at risk for acquired immunodeficiency syndrome. J Clin Invest
9. Karpatkin S, Nardi MA, Lennette ET, Byrne B, Polesz B. Anti-human immunodeficiency virus type 1 antibody complexes on platelets of seropositive thrombocytopenic homosexuals and narcotic addicts. Proc Natl Acad Sci USA
10. Van der Lelie J, Lange JMA, Voss JJE, et al. Autoimmunity against blood cells in human immunodeficiency virus (HIV) infection. Br J Haematol
11. Hohmann AW, Booth K, Peters V, Gordon DL, Comacchio DL. Common epitope on HIVp24 and human platelets. Lancet
12. Bettaich A, Fromont P, Louache F, et al. Presence of crossreactive antibody between human immunodeficiency virus (HIV) and platelet glycoproteins in HIV-1
related thrombocytopenic purpura. Blood
13. Flegg PJ, Jones ME, MacCallum LR, et al. Effect of zidovudine on platelet counts. Br Med J
14. Hymes KB, Green JB, Karpatkin S. The effect of azidothimidine on HIV-related thrombocytopenia
. N Engl J Med
15. The Swiss Group for Clinical Studies on AIDS. Zidovudine for the treatment of thrombocytopenia
associated with HIV. Ann Intern Med
16. Oksenhendler E, Bierlin P, Ferchal F, Clauvel JP, Seligmann M. Zidovudine for thrombocytopenic purpura related to human immunodeficiency virus (HIV) infection. Ann Intern Med
17. Montaner JSG, Le T, Fanning M, et al. The effect of zidovudine on platelet counts in HIV-infected individuals. J Acquir Immune Defic Syndr
18. Ratner L. Human immunodeficiency virus-associated autoimmune thrombocytopenic purpura: a review. Am J Med
19. Walsh C, Krigel R, Lennette E, Karpatkin S. Thrombocytopenia
in homosexual patients: prognosis, response to therapy, and prevalence of antibody to the retrovirus associated with the acquired immunodeficiency syndrome. Ann Intern Med
20. Gordswerg HG, Grossman R, William D. Thrombocytopenia
in homosexual men. Am J Haematol
21. Peltier JV, Lambin P, Doinel C, Couroucé AM, Rouger P, Lefrere JJ. Frequency and prognostic importance of thrombocytopenia
in symptom-free HIV-infected individuals: a five-year prospective study. AIDS
22. Abrams DI, Kiprov DD, Goedert JJ, et al. Antibodies to human T-lymphotropic virus type III and development of the acquired immunodeficiency syndrome in homosexual men presenting with immune thrombocytopenia
. Ann Intern Med
23. Holzman RS, Walsh CM, Karpatkin S. Risk of the acquired immunodeficiency syndrome among thrombocytopenic and non-thrombocytopenic homosexual men seropositive for the human immunodeficiency virus. Ann Intern Med
24. Cameron DA, Flegg PJ. The prognostic importance of thrombocytopenia
25. Centers for Disease Control. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR
26. Sloand EM, Klein HG, Banks SM, Vareldris B, Merritt S, Pierce P. Epidemiology of thrombocytopenia
in HIV infection. Eur J Hematol
27. Landonio G, Nosari A, Spinelli F, Vigorelli R, Caggese L, Schlacht I. HIV-related thrombocytopenia
: four different clinical subsets. Haematologica
28. Ballem PJ, Belzberg A, Devine DV, et al. Kinetic studies of the mechanism of thrombocytopenia
in patients with human immunodeficiency virus infection. N Engl J Med
29. Najean Y, Rain JD. The mechanism of thrombocytopenia
in patients with HIV infection. J Lab Clin Med
30. Spivak JL, Stuart ES, Thomas CQ. Acquired immune deficiency syndrome and pancytopenia. JAMA
31. Zucker-Franklin D, Cao Y. Megakaryocytes of human immunodeficiency virus-infected individuals express viral RNA. Prot Natl Acad Sci USA
32. Zucker-Franklin D, Seremetis S, Zheng ZY. Internalization of human immunodeficiency virus type I and other retroviruses by megakaryocytes and platelets. Blood
33. Zauli G, Re MC, David D, et al. Impaired in vitro
growth of purified (CD34+
) hematopoietic progenitors in human immunodeficiency virus-1 seropositive thrombocytopenic individuals. Blood
34. Kunzi MS, Groopman JE. Identification of a novel human immunodeficiency virus strain cytopathic to megakaryocytic cells. Blood
35. Mientjes GHC, Van Ameijden EJC, Mulder JW, Van den Hoek JAR, Coutinho RA, Von dem Borne AE. Prevalence of thrombocytopenia
in HIV-infected and non-HIV infected drug users and homosexual men. Br J Haematol
36. Adams WH, Rufo RA, Talarico L, Silverman SL, Brauer MJ. Thrombocytopenia
and intravenous heroin use. Ann Intern Med
37. Koury MJ. Thrombocytopenic purpura in HIV seronegative users of intravenous cocaine. Am J Hematol
38. Eyster ME, Diamondstone LS, Lien JM, Ehmann WC, Quan S, Goedert JI. Natural history of hepatitis C virus infection in multitransfused hemophiliacs: effect of coinfection with human immuno-deficiency virus. J Acquir Immune Defic Syndr
Keywords:© Lippincott-Raven Publishers.
HIV-1; Thrombocytopenia; Progression to AIDS