In the 1980s and early 1990s, the prevalence of thrombocytopenia was reported as approximately 5%-15% in patients with HIV infection and 20%-45% in patients with AIDS.1 The epidemiology of and risk factors for thrombocytopenia in HIV-infected individuals in the era of potent HIV antiretroviral therapy has not been studied; however, thrombocytopenia remains an important issue because there is a risk of bleeding events and the presence of significant thrombocytopenia may complicate the treatment of HIV-related illnesses, particularly hepatitis C virus (HCV) infection. The presence of more severe thrombocytopenia may also predict decreased survival in HIV-infected patients.2 We conducted a retrospective case-control study of HIV-infected patients with and without thrombocytopenia receiving care at an urban HIV clinic to elucidate the contributions of HCV infection, HIV viremia, and other potential risk factors for thrombocytopenia in the era of potent antiretroviral therapy.
PATIENTS AND METHODS
Study Population and Matching
This study included HIV-infected outpatients from the 2 HIV clinics (the Centers for Special Studies) of the New York Presbyterian Hospital-Weill Cornell Center. The most recent 100 adult patients seen before July 1, 2005, who had a complete blood count (CBC) demonstrating thrombocytopenia, defined as platelet counts <100 × 109/L, were considered for inclusion in the study. The cut-off of 100 × 109/L was recently selected by an international group standardizing definitions of immune thrombocytopenic purpura (ITP).3 Only patients with thrombocytopenia persisting for >3 consecutive months were eligible for the study. Patients with no outpatient medical records or thrombocytopenia resulting from interferon therapy, chemotherapy, or malignancy were excluded. Incident and prevalent cases of thrombocytopenia were included in the study. Controls matched 1:1 on the following criteria were generated randomly: age ± 5 years, same sex, and first appointment date ± 6 months. Controls without outpatient records were excluded.
Medical records were reviewed retrospectively to ascertain demographic, clinical, and laboratory characteristics at the time of detection of thrombocytopenia, which was defined for the purpose of this study as onset of thrombocytopenia in incident cases or first clinic visit in prevalent cases. HIV- and HCV-associated variables included risk factor for transmission, years since HIV diagnosis, AIDS diagnosis, CD4 cell count and nadir, HIV RNA level (using a limit of detection of <400 or <50 copies/mL as available), antiretroviral regimen and duration, chronic HCV infection status (based on HCV antibody and HCV RNA results), and liver-related tests. Assessment of cirrhosis was based on provider documentation and imaging results (nodular liver or portal hypertension) or liver biopsy results describing histopathologic diagnosis of cirrhosis (Metavir Fibrosis Stage 4). Alcohol use was elucidated from the patient's medical record based on documentation provided by medical, psychiatric, and social work providers and quantity interpreted as none, mild-moderate, heavy, and qualified as past or current use. Outcomes including treatments for thrombocytopenia, major bleeding events (defined as gastrointestinal, intracranial, or events requiring hospitalization), and deaths during the duration of patient's follow up in the clinic were also recorded. The study was approved by the Institutional Review Board of the Committee on Human Rights in Research at Weill Cornell Medical College (New York, NY), and a waiver of need for informed consent was obtained.
A sample size calculation estimated that 53 cases would provide 80% power to detect an odds ratio (OR) of 3.0 for the association between HCV infection and thrombocytopenia based on an estimated 40% prevalence of HCV infection in controls (alpha of 0.05). HIV RNA level (detectable/undetectable), HCV infection status, and the other variables described above were compared between cases and controls using conditional logistic regression, and ORs with 95% confidence intervals (CIs) were calculated. Variables with a skewed distribution were dichotomized around the upper limit of normal or the median depending on clinical relevance, because transformation of data did not correct skewedness. Multivariate models were created using forward, stepwise conditional logistic regression; the criterion to enter the model was P ≤ 0.2. Only covariates with P values of <0.05 in univariate analyses were considered in the multivariate analysis. Data were analyzed using Stata (version 9.0) software, (Stata Corp., College Station, TX).
Of the 100 patients identified with platelet counts <100 × 109/L, 27 patients were excluded for the following reasons: thrombocytopenia did not persist for >3 months (6), no outpatient records existed (10), interferon related (4), chemotherapy or malignancy related (3), and age less than 18 years (4); this left 73 evaluable cases for whom 73 matched controls were generated.
Of the case patients, nadir platelet count was ≤50 × 109/L in 58%, ≤30 × 109/L in 38%, and ≤10 × 109/L in 7%. The thrombocytopenia prevalence among 2298 outpatients who had CBC testing during the period from which the cases were obtained (May 2004-July 2005) was calculated at 3.2% (95% CI 2.5-4.0%). Of the 73 patients identified as thrombocytopenia cases, 43 (59%) were classified as incident cases of thrombocytopenia (platelet count was normal on preceding CBCs at our clinic), whereas 30 cases (41%) were defined as preexisting (prevalent) cases in which thrombocytopenia existed at the time of initial presentation to our clinic. Thrombocytopenia had not resolved at the time of last laboratory testings in 64% of cases. In these cases, the median duration of thrombocytopenia was 3.3 years (interquartile range = 1.2-6.5 years), whereas it was 2 years (interquartile range = 1.1-3.5) in cases with resolution of thrombocytopenia.
Table 1 shows demographic characteristics of cases and controls at the time of onset of thrombocytopenia and the characteristics associated with thrombocytopenia on univariate analysis and multivariate analysis. HIV RNA detectability, HCV infection, and cirrhosis remained significantly associated with thrombocytopenia status after adjusting for the other model variables. In fact, 98% of case patients compared with 53% of controls had at least one of these risk factors, and all 3 were observed in 15% of cases compared with none of the controls. In patients without HCV infection or cirrhosis, detectable HIV RNA was strongly associated with thrombocytopenia case status [present in 21 of 22 cases versus 5 of 22 of controls with available HIV RNA data, OR = 17.5 (CI 2.3-127.4) P = 0.006]. In contrast, in patients with known HCV infection or cirrhosis, HIV RNA detectability did not differ significantly between cases and controls.
Major bleeding events defined as gastrointestinal, intracranial, or events requiring hospitalization occurred in 13 cases compared with 2 controls (OR 6.5, CI 1.5-28.8; P = 0.014). Among case patients, major bleeding events were not associated with nadir platelet count or duration of thrombocytopenia; however, there was an association with cirrhosis (OR = 11.5, CI 2.3-58, P = 0.003) and a trend toward an association with HCV infection (OR = 3.92, CI 0.8-19.3, P = 0.09). Compared with controls, cases with thrombocytopenia were more likely to suffer death from any cause or major bleeding events during their history of thrombocytopenia; however, no fatal bleeding events occurred. Nine deaths occurred among cases compared with 0 in control subjects (P = 0.002, chi-square); control subjects, however, were more likely to have unknown vital status (7 cases versus 16 controls, P = 0.04, chi-square).
Nineteen case patients (26%) were treated for thrombocytopenia with 7 documented as receiving >1 type of treatment. Treatments included the following: corticosteroids (received by 11 patients); platelet transfusions, intravenous immunoglobulin, and anti-D (each received by 5); progesterone (3); splenectomy (2); and vincristine (1). In addition, 5 patients required packed red cell transfusion due to a major bleeding event.
Our study demonstrates that thrombocytopenia remains a clinically significant problem in the current HIV treatment era. The prevalence of thrombocytopenia in HIV-infected outpatients detected in this study (approximately 3%) is decreased substantially compared with that seen during the era before potent antiretroviral therapy. Nevertheless, among those thrombocytopenic in this study, more than half experienced a platelet count less than 50 × 109/L with more than one-third experiencing a nadir below 30 × 109/L, and one-quarter requiring treatment. In addition to being severe, thrombocytopenia did not resolve quickly in most patients, often persisting for years.
As predicted, uncontrolled HIV and HCV replication, particularly associated with advanced liver disease, was strongly associated with thrombocytopenia in this study. Our data also indicate that it is HIV replication itself rather than advanced disease related to AIDS or low CD4 count that is associated with thrombocytopenia in the current era, as supported by epidemiologic, mechanistic, and clinical data reported before the availability of potent antiretroviral therapy.1,4-6
The strong association between HCV coinfection and thrombocytopenia in this study, likely explains why thrombocytopenia did not resolve with effective HIV treatment in some patients. Traditionally, the thrombocytopenia related to HCV has been attributed to cirrhosis and portal hypertension causing splenic sequestration of platelets, although other factors may also contribute. Advanced liver disease results in reduced liver production of thrombopoietin, the dominant humoral thrombopoietic factor.7-9 Our study found a strong association of thrombocytopenia with cirrhosis (and splenomegaly), but also found an independent association with HCV infection suggesting that mechanisms other than advanced liver disease may play a role in HIV/HCV-coinfected patients. Chronic ITP associated with HCV has been described previously,10,11 and cryoglubulins12 and antiplatelet autoantibodies13-15 have been detected in thrombocytopenic patients with HCV.
Deaths occurred more frequently in cases than in controls, although no deaths were specifically attributed to bleeding. Thrombocytopenia has previously been identified as a poor prognostic marker for AIDS and death,2,16 however, conflicting results have been described as well.17,18 Despite the absence of hemorrhagic deaths, major bleeding events occurred in 13 cases. Bleeding events were not associated with nadir platelet count or duration of thrombocytopenia but were associated with cirrhosis and trended toward an association with HCV infection. Although coagulopathy or the presence of esophageal varices may explain this association of bleeding events with cirrhosis, an increased tendency to bleed at higher platelet counts has been described as a feature of HCV-related ITP without associated coagulopathy, possibly mediated by cryoglobulin-induced vasculitis.12
Limitations of this study include its retrospective nature, which introduces possible biases related to ascertainment, documentation, and chart review. It is also possible that the presence of thrombocytopenia itself contributed to the diagnosis of cirrhosis or led to disproportionate testing to detect cirrhosis in cases compared with controls. However, an argument against this is that the cases also differed significantly in terms of other laboratory markers of cirrhosis, which were routinely assessed in all patients. In addition, both prevalent and incident cases were included to have a study population reflecting the spectrum of patients with thrombocytopenia in the current era. Because of this, cases displayed a variable duration of thrombocytopenia, including cases with thrombocytopenia predating their enrollment in the clinic. To assess the effect of these limitations, sensitivity analyses were performed (data not shown). The inclusion of only the incident cases in the multivariate models did not change the outcomes qualitatively, nor did excluding 3 subjects whose thrombocytopenia detection predated 1996. Thus, the results of this study seem to be generalizable to patients with thrombocytopenia whether it is newly diagnosed or longstanding. This study also predated the Food and Drug Administration approval of the first thrombopoietic agent, which has been shown to be an effective option for the treatment of refractory ITP.
In summary, thrombocytopenia still occurs frequently in HIV-infected patients and at levels that could predispose to bleeding and require treatment. Detectable HIV RNA, HCV infection, and cirrhosis are major independent risk factors associated with thrombocytopenia in the current antiretroviral era. When assessing patients with HIV infection, it is important to recognize that thrombocytopenia may be a sign of cirrhosis, particularly in HCV-coinfected patients, whether there is coincidentally detectable HIV RNA. Furthermore, in coinfected patients, milder degrees of thrombocytopenia may predispose to bleeding and preclude the use of pegylated interferon, which is of great relevance because a large burden of thrombocytopenia seems to be due to HCV infection and related liver disease. In patients without HCV infection or liver disease, detectable levels of HIV replication may account for most cases. The better understanding of the risk factors for and etiology of thrombocytopenia initiated here will help guide whether antiviral, immune-based, or thrombopoietic therapies are appropriate initial treatment strategies for HIV-infected individuals with thrombocytopenia in future.
The authors thank the patients and physicians of the New York Presbyterian Hospital's Center for Special Studies for making this study possible and W.D. Johnson, Jr. and R.M. Gulick for their mentorship. We acknowledge Ximena Lavender for assistance with data entry.
1. Carbonara S, Fiorentino G, Serio G, et al. Response of severe HIV
to highly active antiretroviral therapy including protease inhibitors. J Infect
2. Sullivan PS, Hanson DL, Chu SY, et al. Surveillance for thrombocytopenia
in persons infected with HIV
: results from the multistate Adult and Adolescent Spectrum of Disease Project. J Acquir Immune Defic Syndr Hum Retrovirol
3. Rodeghiero F, Stasi R, Gernsheimer T, et al. Standardization of terminology, definitions and outcome criteria in immune thrombocytopenic purpura (ITP) of adults and children: report from an International Working Group. Blood
4. Servais J, Nkoghe D, Schmit JC, et al. HIV
-associated hematologic disorders are correlated with plasma viral load and improve under highly active antiretroviral therapy. J Acquir Immune Defic Syndr
5. Nardi MA, Liu LX, Karpatkin S. GPIIIa-(49-66) is a major pathophysiologically relevant antigenic determinant for anti-platelet GPIIIa of HIV
-1-related immunologic thrombocytopenia
. Proc Natl Acad Sci U S A
6. Oksenhendler E, Bierling P, Ferchal F, et al. Zidovudine for thrombocytopenic purpura related to human immunodeficiency virus (HIV
) infection. Ann Intern Med
7. Giannini E, Botta F, Borro P, et al. Relationship between thrombopoietin serum levels and liver function in patients with chronic liver disease related to hepatitis C
virus infection. Am J Gastroenterol
8. Giannini E, Borro P, Botta F, et al. Serum thrombopoietin levels are linked to liver function in untreated patients with hepatitis C
virus-related chronic hepatitis. J Hepatol
9. Adinolfi LE, Giordano MG, Andreana A, et al. Hepatic fibrosis plays a central role in the pathogenesis of thrombocytopenia
in patients with chronic viral hepatitis. Br J Haematol
10. Bauduer F, Marty F, Larrouy M, et al. Immunologic thrombocytopenic purpura as presenting symptom of hepatitis C
infection. Am J Hematol
11. Pockros PJ, Duchini A, McMillan R, et al. Immune thrombocytopenic purpura in patients with chronic hepatitis C
virus infection. Am J Gastroenterol
12. Rajan SK, Espina BM, Liebman HA. Hepatitis C
: clinical and laboratory characteristics compared with chronic immune thrombocytopenic purpura. Br J Haematol
13. Sanjo A, Satoi J, Ohnishi A, et al. Role of elevated platelet-associated immunoglobulin G and hypersplenism in thrombocytopenia
of chronic liver diseases. J Gastroenterol Hepatol
14. Pereira J, Accatino L, Alfaro J, et al. Platelet autoantibodies in patients with chronic liver disease. Am J Hematol
15. Kajiwara E, Akagi K, Azuma K, et al. Evidence for an immunological pathogenesis of thrombocytopenia
in chronic liver disease. Am J Gastroenterol
16. Pearce CL, Mack WJ, Levine AM, et al. Thrombocytopenia
is a strong predictor of all-cause and AIDS-specific mortality in women with HIV
: the women's interagency HIV
17. Rieg G, Yeaman M, Lail AE, et al. Platelet count is associated with plasma HIV
type 1 RNA and disease progression. AIDS Res Hum Retroviruses
18. Holzman RS, Walsh CM, Karpatkin S. Risk for the acquired immunodeficiency syndrome among thrombocytopenic and nonthrombocytopenic homosexual men seropositive for the human immunodeficiency virus. Ann Intern Med