Although the response to treatment of HIV-1-related thrombocytopenia does not appear to differ substantially from that observed in adults with idiopathic chronic thrombocytopenia, there are limitations when using conventional therapeutic approaches. Steroids and other immunosuppressive agents (for example, vincristine, cyclophosphamide) [11, 12] can facilitate the development of opportunistic infections or can lead to the flaring up of infectious hepatitis, which is often present in these patients [13, 14]; steroids have also been reported to accelerate the progression of Kaposi sarcoma [15, 16]. Splenectomy enhances the risk for subsequent infection, especially severe sepsis [17], and in some patients a late relapse of thrombocytopenia may occur [18]. High-dose intravenous immunoglobulins represent an effective treatment in most patients, but the effect is short-lasting in almost all of them, and the high costs and the need for intravenous administration limit their use [9, 19]. Finally, alternative approaches, such as dapsone [20], protein A immunoadsorption [21], or anti-rhesus factor IgG [19], have been tested in limited series and uncontrolled studies (or both).

It is now well established that zidovudine (azidothymidine) can effectively enhance the platelet count of patients with HIV-1-related thrombocytopenia. A remission of thrombocytopenia during zidovudine administration has been shown in several case series [22-24] and in one placebo-controlled, prospective clinical study [25]. However, between 30% and 40% of patients with thrombocytopenia are resistant to full-dose (1000 mg/d) zidovudine treatment [23, 24, 26, 27]. It is not clear that higher doses of zidovudine affect thrombocytopenia [28, 29]; in addition, higher doses are potentially toxic [30].

Some anecdotal reports and small, uncontrolled case series have recently suggested that interferon- can correct the thrombocytopenia of patients with HIV-1 infection [9, 14, 31-33], even in patients not responding to zidovudine treatment [31, 33]. However, no studies have analyzed, under carefully controlled conditions, the efficacy, toxicity, and kinetics of the response to interferon- in patients with zidovudine-resistant, HIV-1-related thrombocytopenia.

We tested whether interferon- can increase the platelet count in patients with zidovudine-resistant, HIV-1-related severe thrombocytopenia and assessed the kinetics of the interferon- effects and the possible toxicity of interferon in association with zidovudine in a randomized, placebo-controlled clinical study.

Methods

Top Methods Results Discussion Author & Article Info References

Selection of Patients

Patients were considered eligible for the study if they were seropositive for HIV-1 by the enzyme-linked immunosorbent assay and the Western-blot technique and had a platelet count less than 25 x 10⁹/L on at least two occasions separated by more than 2 weeks. In addition, thrombocytopenia had to be refractory to a 1-month course of full-dose (1000 mg/d) zidovudine treatment and to previous therapeutic attempts with one of the following: corticosteroids, splenectomy, vincristine, danazol, high-dose intravenous IgG, or anti-ID immunoglobulins.

The main exclusion criteria were pregnancy or breast-feeding; the acquired immunodeficiency syndrome; concomitant serious diseases not related to HIV-1 infection (cardiomyopathy, seizure, stroke, psychosis); presence of antinuclear, anti-smooth-muscle, antimitochondrial, or anti-liver-kidney microsome 1 antibodies; and granulocyte counts less than 1 x 10⁹/L or hemoglobin levels less than 80 g/L. Women of childbearing potential were advised to practice effective methods of birth control.

Assessment before study included a medical history, physical examination, electrocardiogram, chest radiograph, and the following laboratory tests: measurement of hemoglobin, hematocrit, mean corpuscular volume, leukocyte count, differential leukocyte count, platelet count, serum glucose, blood urea nitrogen, serum creatinine, bilirubin, aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, serologic tests for hepatitis B and C, prothrombin time, activated partial thromboplastin time, bleeding time, and urinalysis.

Fifteen consecutive patients satisfying the criteria for entry into the study were enrolled between April and October 1992 in four centers belonging to four different hospitals in Central Northern Italy. Written, informed consent was obtained from each patient before enrollment in the study; the study protocol was approved by the Ethics Committee of the Umbria Region.

Treatment

Patients were enrolled in a double-blind, crossover, placebo-controlled, randomized study in which they were treated with either subcutaneous lymphoblastoid interferon- (Wellferon, Burroughs Wellcome, Italy) or 1 mL of subcutaneous saline solution (placebo). Therapy was administered to the patients in hospital by specially trained nurses. The treatment code was known only to the nurse, and the doctors, nurses involved in patient management, and the patients were all blinded to the study code.

All patients received subcutaneous injections of placebo or interferon-, 3 million units, three times a week for 4 weeks. During the following 4 weeks, they did not receive any injections (washout period); they were then switched to the opposite treatment for another 4 weeks. After treatment discontinuation, patients were followed for a final 4-week period (washout period). Patients were randomly assigned to either sequence of the crossover design.

The study lasted 16 weeks; during this time patients were seen weekly for clinical evaluations including a physical examination, a review of subjective complaints, and standard hematologic tests including platelet counts. All patients received, throughout the study period, zidovudine therapy at 200 mg three times a day. Bleeding times, immune studies, and p24 antigen, hematologic, renal, and hepatic tests were done at weeks 0, 4, 8, 12, and 16.

Evaluation Criteria

The primary end point, established before the beginning of the study, was the variation in the number of circulating platelets. A secondary end point was the clinical response to interferon- treatment, defined as follows: 1) complete response: platelet count more than 100 x 10⁹/L; 2) partial response: platelet count more than 50 x 10⁹/L and less than 100 x 10⁹/L; 3) failure: platelet count less than 50 x 10⁹/L. A cut-off level of 50 x 10⁹ platelets/L was chosen to define treatment failure because conventionally hemorrhage is rare when counts are greater than this limit. Other secondary end points included the bleeding time, p24 antigen levels in serum, CD4 and CD8 counts, ß₂-µglobulin levels in serum, and platelet-associated IgG.

Laboratory Data

Platelet counts were determined on ethylenediaminetetraacetic acid-anticoagulated venous blood samples by using electronic particle counters (Coulter Counter, STKR, Hialeah, Florida for the four centers). Platelet counts were measured on each patient using the same instrument throughout the study. Bleeding time was carried out according to Mielke and colleagues [34] by the use of a standardized template (Simplate II, General Diagnostics, Morris Plains, New Jersey). The registration of the bleeding time was followed for a maximum of 30 minutes; when the bleeding time exceeded 30 minutes, it was considered to be equal to 30 minutes for the statistical analysis.

CD4 and CD8 lymphocyte counts were determined by two-color flow cytometry using monoclonal antibodies (Leu 2-3+, Leu 2+3-; Leu 4+Dr-; Becton Dickinson, Franklin, New Jersey) on whole blood samples. Serum p24 core antigen was determined by enzyme immune assay (New England Nuclear, du Pont, Rockville, Maryland); the least detectable amount with this assay is 4.4 pg/mL of serum. Serum ß₂-µglobulin was assayed by enzyme immune assay (Behring, Scoppito, Italy). Serum platelet antibodies were detected with the indirect platelet suspension immunofluorescence test [35].

Determinations of hemoglobin, complete blood counts, serum glucose, blood urea nitrogen, creatinine, liver function tests (aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, -glutamyltranspeptidase, and total bilirubin), prothrombin time, and activated partial thromboplastin times were carried out according to standard procedures.

Data Collection and Statistical Analysis

All investigators, nurses, and laboratory personnel involved in patient and sample management were blinded to treatment groups until all end points were determined at the end of the study. First-order carryover identifies the residual effect of a given treatment at the time of the second baseline (before the beginning of the second treatment), whereas second-order carry-over indicates any residual effect of the first treatment at the time of the second treatment.

Data required by the study protocol were collected and recorded in case report forms by the investigators at each of the participating centers. The case report forms were then submitted to clinical research personnel at the Institute of Internal and Vascular Medicine of the University of Perugia for review and entry of data into computer programs.

The calculation of the best study sample size was done before the study as follows: It was estimated that the average baseline platelet count of enrolled patients would be about 20 x 10⁹/L and that an increase in platelet count of at least 150% (from 20 to 50 x 10⁹/L) produced by the active treatment would be worth detecting. Given an -error of 0.05, a ß-error of 0.20, and a predicted standard error of 21 x 10⁹/L, we calculated that a sample size of at least 10 patients was necessary to detect any changes.

Statistical analysis was carried out, for primary and secondary end points, according to the method described elsewhere for crossover design with baseline [36]. The difference between treatment effects was the main estimate, and it was analyzed only after the first and second-order carryover effects were not significant. The method we adopted does not include a specific analysis of the sequence effect because this is contained in the overall analysis (sequence by period interaction) [36]. When using binary response variables as the end point of the crossover analysis, we represented the effects in terms of logarithms of odds ratios according to a model previously described [36].

Analyses of variance with repeated measures and the paired or unpaired t-test were applied to continuous variables according to the standard criteria. When appropriate, the chi-square test with correction of continuity was used in comparisons implying categorical variables. Data are presented as mean ±SD or, when specified, ±SE. All confidence intervals (CIs) are 95%, except for carryover effects where we used 90% CIs to improve the power of the test.

Results

Top Methods Results Discussion Author & Article Info References

Characteristics of the Study Population

Fifteen patients were enrolled in the study at the four participating centers. The baseline characteristics at enrollment are reported in Table 1. Fourteen patients had a history of intravenous drug abuse and 2 of them were also homosexual; for 1 patient, the risk factor or factors for HIV-1 infection were unknown. All patients had stopped drug abuse at the time of enrollment. The mean age of the patients was 29 years (range, 24 to 40 years), and two were women. The average stage of the disease, based on Walter Reed classification [37], was 4.1 (range, 1 to 5). Thirteen patients were positive for antibodies to hepatitis C virus; 2 of them were also positive for hepatitis B virus, whereas 1 did not show serologic signs of hepatitis infection; 1 patient was not tested. Nine of 15 patients had signs of chronic hepatitis. The CD4 lymphocyte count was 0.237 x 10⁹/L (range, 0.012 to 0.534), and p24 antigen was detected in 4 patients at baseline.

Thrombocytopenia had been present for an average of 2.5 years (range, 1 to 6 years). All the patients had been treated, with no or only partial effect, with one or more of the following: prednisone, high-dose IgG, and anti-rhesus factor IgG. In addition, at the time of enrollment all had completed at least 1 month of treatment with zidovudine, 1000 mg/d, with no increase of the platelet count above 25 x 10⁹/L. Actually, for patients 1 to 12, 14, and 15, the duration of previous zidovudine therapy ranged between 12 and 28 months, whereas it was 1 month for patient 13. Only one of the patients had palpable splenomegaly. Six patients had a positive bleeding history (epistaxis in 2, bleeding gums in 1, bruising in 2, and hemoptysis in 1). None of the patients showed signs of opportunistic infections at enrollment or during the study.

Two patients (patients 14 and 15) were lost at follow-up (immediately after enrollment and after 2 weeks of follow-up, respectively) because they did not come back to the center for the control visit. One patient (patient 13) was withdrawn from the trial when a bleeding episode (bleeding gums) occurred concomitantly with a decrease in platelet count that took place at the end of the washout period that followed interferon- treatment. This patient had responded partially to interferon- (platelet count: baseline level, 20 x 10⁹/L; after interferon-, 55 x 10⁹/L), had a further increase at the end of the first week of washout (77 x 10⁹/L), and then showed a progressive decrease (second week: 16 x 10⁹/L; third week: 9 x 10⁹/L; and fourth week: 2 x 10⁹/L). After withdrawal from the study, the patient was treated with interferon- (3 million units, three times per week) and had a slight response (after 2 weeks: 27 x 10⁹/L). Additional therapeutic attempts (full-dose zidovudine; intravenous IgG; anti-rhesus factor IgG; high-dose prednisone) were unsuccessful. This patient did not differ from the others in our series except for receiving a course of only 1 month of full-dose (1000 mg/d) zidovudine before entering the trial.

Platelet Counts

The mean baseline platelet count for the 12 patients who completed the trial was 15.6 x 10⁹/L (95% CI, 11.1 to 20.1 x 10⁹/L). No difference was observed between the average baseline platelet count of the patients who received either sequence of treatments (16.5 ±3.1 x 10⁹/L compared with 15.1 ±8.1 x 10⁹/L; P > 0.2). After the fourth week of treatment, the estimated change in platelet count induced by interferon- with respect to placebo was 60.0 ±16.5 x 10⁹/L (mean ±SE; CI, 23.2 to 96.8 x 10⁹/L; P = 0.0045). No significant residual effects of the first treatment were detected either before the beginning of the second treatment (second baseline) or during the second treatment (baseline: 90% CI, –10.1 to 7.2 x 10⁹/L, P > 0.2; second treatment: 90% CI, –14.1 to 39.6 x 10⁹/L, P > 0.2). Interferon- treatment increased the platelet count to 40 x 10⁹/L (CI, 19.4 to 60.6 x 10⁹/L), 58.8 x 10⁹/L (CI, 31.3 to 86.4 x 10⁹/L), 66.6 x 10⁹/L (CI, 35.0 to 98.1 x 10⁹/L; P < 0.05), and 82.2 x 10⁹/L (CI, 48.9 to 115.3 x 10⁹/L; P < 0.005) at the end of 1, 2, 3, and 4 weeks of treatment, respectively (Figure 1). After interferon- treatment, the average maximal increase in the platelet count was 66.6 x 10⁹/L (CI, 35.4 to 97.8 x 10⁹/L).

View larger version (17K): [in this window] [in a new window]   Figure 1. Change in platelet counts during the study. Average platelet counts in the 12 patients who completed the trial. Values represent the mean ±SD of all patients completing the study (\#9679;——\#9679;, n = 12), of patients who responded (——, n = 8), and of patients who did not respond (——, n = 4) to interferon. Asterisks represent a significant difference when compared with baseline (basal) levels (P < 0.05).

After receipt of placebo, the platelet counts were not statistically different from baseline levels. Platelet counts were as follows: after 1 week, 21.5 x 10⁹/L (CI, 14.2 to 28.8 x 10⁹/L); after 2 weeks, 28.2 x 10⁹/L (CI, 15.0 to 41.4 x 10⁹/L); after 3 weeks, 25.6 x 10⁹/L (CI, 10 to 41.3 x 10⁹/L), and after 4 weeks, 24.1 x 10⁹/L (CI, 12.4 to 35.7 x 10⁹/L) (Figure 1). None of these values differs significantly from the baseline platelet count; the maximal increase was 8.5 x 10⁹/L (CI, –1.73 to 18.7 x 10⁹/L).

After interferon- therapy, platelet counts tended to slowly return to pretreatment values, but a statistically significant increase over baseline levels was still present 1 week after treatment discontinuation (Figure 1). Four of 12 patients were unresponsive to interferon- according to the preset categories of clinical response. The patients who did not respond differed from the responders in several laboratory variables, suggesting a more advanced stage of the disease (Table 2).

View this table: [in this window] [in a new window]   Table 2. Distinctive Clinical and Laboratory Characteristics of Patients Responding, or Not, to Interferon- Treatment*

An additional analysis of the results was done to include the patient who was withdrawn from the trial because of a decrease in platelet count during the first washout period. This analysis was based on a binary variable obtained by considering complete or partial response to treatment compared with failure and by assuming the withdrawn patient as a treatment failure in all periods. The difference between interferon- and placebo was close to significance (odds ratio, 2.77 in favor of interferon-; CI, 0.96 to 7.89; P = 0.058) and was almost identical to the results obtained when the analysis was repeated without the withdrawn patient (odds ratio, 2.70 in favor of interferon-; CI, 0.95 to 7.82; P = 0.062).

Laboratory Data

Bleeding times were as follows: baseline level, 23.2 minutes (CI, 19.2 to 27.2 minutes); after interferon-, 11.7 minutes (CI, 8.7 to 14.8 minutes; P < 0.0005 compared with baseline levels); after placebo, 21.9 minutes (CI, 17.8 to 26.8 minutes; P > 0.2 compared with baseline levels); and 17.1 minutes (CI, 12.8 to 21.4 minutes) and 21.9 minutes (CI, 17.8 to 25.9 minutes), after the two washout periods, respectively (P = 0.053 and P > 0.2 compared with baseline levels, respectively). The estimated decrease in bleeding time induced by interferon- when compared with placebo was 10.3 ±3.5 minutes (mean ±SE; CI, 2.45 to 18.1 minutes; P = 0.015). No significant residual effects were detected of the first treatment on the bleeding time, measured at the end of the first washout period or at the end of the second treatment period.

The bleeding times before and after treatment were somewhat shorter than the values expected on the basis of the platelet count, but they were longer than those reported for patients with immune thrombocytopenic purpura with similar numbers of circulating platelets [38]. Considering that the relatively short bleeding times of patients with immune thrombocytopenic purpura are because of the presence in the circulation of a large proportion of young platelets with enhanced hemostatic capacity [38], our results agree with the mixed nature, central and peripheral, of HIV-related thrombocytopenia [10, 23]. Interferon- treatment did not alter any of the other laboratory variables, including the presence of platelet antibodies in serum.

Tolerability

Treatment was well tolerated. During interferon- treatment, slight fever was present in four patients, malaise in three, and chilling in one patient; during placebo treatment, two patients had slight fever and one had malaise. No granulocytopenia or increases in the levels of alanine aminotransferase were observed. Of the four patients positive for p24 antigen, three remained positive after interferon-, whereas the fourth patient became negative for p24 antigen; one patient negative for p24 antigen became positive after interferon-.

Minor bleeding took place in four patients: in three during interferon- treatment (patients 4, 5, and 12) and in one after placebo (patient 3). Of the three patients who had hemorrhagic episodes during interferon- treatment, two (patients 5 and 12) did not respond to interferon-, and one (patient 4) had only taken the drug for less than 2 weeks.

Discussion

Top Methods Results Discussion Author & Article Info References

Our data show that interferon- treatment is able to increase the platelet count in patients with HIV-1 infection with zidovudine-resistant, severe thrombocytopenia. The increase in platelet count after treatment is rather large, and it is associated with a striking decrease in the bleeding time. The increase in levels of circulating platelets produced by interferon- is relatively prompt and persistent, with more than a doubling occurring after 1 week of treatment, and levels are still statistically significantly increased when compared with baseline levels 1 week after discontinuing interferon- treatment. Interferon- is effective in two thirds of the patients, similar to what was previously observed with this treatment in patients with immune, non-HIV-1-related thrombocytopenia [11].

Although the small number of patients studied makes it difficult to carry out subgroup analyses, results from laboratory tests suggest that the patients who did not respond to interferon- are those with the most serious alterations. In particular, before treatment, three of four patients who did not respond to interferon- were positive for p24 antigen (14.7 ±11.0 pg/mL), whereas only one of eight responding patients presented with detectable levels of p24 antigen (9 pg/mL). After the 1-month cycle of interferon-, four of four patients who did not respond were positive for p24 antigen (16.5 ±9.5 pg/mL) compared with none of the eight patients who responded to treatment. The levels of ß₂-µglobulin and circulating IgA were also higher in patients who did not respond: Both variables are considered to be indicators of disease progression [39, 40]. The baseline number of circulating platelets was not significantly lower (P = 0.075) in patients who did not respond to interferon-. Moreover, after interferon- the lymphocyte CD4/CD8 ratio of patients who did respond was significantly higher (P = 0.002) than that of those who did not respond (Table 2).

Interferon- appears to be able to interfere with virus replication [41], and interferon- and zidovudine act synergistically against HIV-1 replication in vitro [42]. This combination therapy has been proposed as an early treatment in persons with asymptomatic HIV-1 infection to stop the progression of the infection [43] and was found to be well tolerated and possibly effective [30].

Possible advantages of interferon- compared with other therapeutic measures in HIV-1-related thrombocytopenia are the relatively low cost, the lack of need for hospitalization, and the possible beneficial effect on hepatitis progression. The latter aspect may be relevant considering that a relatively large proportion of patients positive for HIV-1 are infected with hepatitis virus (B or C, or both).

The effect of interferon- on platelet count in patients responding to this treatment is clearly evident within 1 week of its institution; thus, it may be possible to carry out a short (2 to 3 weeks) therapeutic trial with interferon- in patients who are thrombocytopenic soon after an unsuccessful attempt with zidovudine and only later to test other more invasive (splenectomy) or more expensive or more unconventional (intravenous IgG; anti-rhesus factor IgG, and so forth) therapeutic modalities. Platelet and leukocyte counts should be monitored closely during the first period of interferon- treatment in view of some possible decreases in cell counts reported in sporadic patients [44].

It remains to be established whether the increase in platelet count attained in patients responding to interferon- can be sustained during a more prolonged period than that (1 month) tested in our study. It is also conceivable that a dose lower than 3 million units or that a lower frequency of administration (for example, once a week), or both, is sufficient to maintain remission in patients who respond. In addition, based on our results, it is unclear whether zidovudine contributes to the response observed after the institution of interferon- treatment.

Our data, obtained from a double-blind, randomized, placebo-controlled study, show that interferon- is a safe and effective treatment for HIV-1-related, severe, zidovudine-resistant thrombocytopenia. In our series of patients, treatment with interferon- was well tolerated; it might be possible to adopt this treatment even before using zidovudine in those patients not requiring the antiviral agent; thus, a prospective trial in this subset of patients could be of interest.