Who should treat chronic ITP? Initial treatment (with corticosteroids or splenectomy) of uncomplicated cases of ITP can be managed by a family practitioner or general internist who has experience in this area. If the patient is actively bleeding or if corticosteroid therapy and splenectomy are ineffective, a hematologist or oncologist should be contacted. A general relation exists between the platelet count and symptoms: Patients with platelet counts greater than 50 x 10⁹/L are asymptomatic; those with counts of 30 to 50 x 10⁹/L have bruising on trauma; those with counts of 10 to 30 x 10⁹/L have spontaneous bruising, menorrhagia, and prolonged bleeding with injury; and those with counts less than 10 x 10⁹/L have mucosal bleeding (epistaxis, gastrointestinal, and genitourinary) and risk for central nervous system bleeding.

Patients who can maintain safe platelet counts (>30 x 10⁹/L) do not require therapy; all others should be treated. Good evidence in the literature [13] supports the use of corticosteroids [14-20] as initial therapy for ITP; corticosteroids should be followed by splenectomy [14-19] in nonresponders. Prednisone, 1 mg/kg · d^–1, is started at diagnosis. If the platelet count normalizes, the dose should be continued for 1 to 2 weeks and then tapered by 10 mg/wk until the dose reaches 0.5 mg/kg of body weight and by 5 mg/wk thereafter. Some patients with ITP maintain normal platelet counts, but most have relapse when the dose is tapered. Splenectomy is indicated if any of the following conditions exist: 1) the platelet count is less than 50 x 10⁹/L after 4 weeks (one study noted that all patients who attained complete remission had a platelet count >50 x 10⁹/L within 4 weeks [16]), 2) the platelet count has not normalized after 6 to 8 weeks (because side effects become problematic), or 3) the platelet count normalizes but decreases when therapy is tapered.

A recent study [50] suggests an alternate method for administering corticosteroids before splenectomy. Dexamethasone (40 mg/d for 4 days every 4 weeks for 6 cycles) was given to four patients before splenectomy. All patients achieved platelet counts greater than 100 x 10⁹/L that persisted during follow-up. However, these results require confirmation.

Patients for whom corticosteroid therapy is ineffective should have splenectomy. Before surgery, patients should be vaccinated with pneumococcal, Haemophilus influenzae, and meningococcal vaccines [51]. In severely thrombocytopenic patients, the corticosteroid dose should be increased or intravenous IgG should be given to increase the platelet count before surgery. Platelet transfusions should not be given prophylactically or during surgery unless excessive bleeding occurs. Laparoscopic splenectomy may be preferable in ITP [52]. If relapse occurs, patients should be evaluated for an accessory spleen [16, 53-57], which should be removed if present. Splenic radiation [58] or dapsone therapy [59-61] should be considered for patients who are poor surgical candidates because of age or concomitant disease.

Therapy for Refractory Patients

Refractory patients (who comprise about 25% to 30% of patients with ITP) are defined as those in whom treatment with standard-dose corticosteroids and splenectomy fails and who require further therapy because of unsafe platelet counts or clinical bleeding. This group responds poorly to subsequent treatment, has significant morbidity from the disease and its treatment, and has a mortality rate of about 16% [6]. Therapy should be tailored to the individual patient, but this is not always possible. For example, safe platelet counts are different in sedentary persons and those with active lifestyles, and certain agents (such as alkylating agents) should be avoided in patients who wish to have children. Comorbid conditions, such as cardiac disease and bone disease, may be exacerbated by some treatments. In some patients, the risk associated with a lower platelet count may be less than that of the treatment.

Suggestions for the treatment of patients with refractory ITP are primarily based on my experience because clinical reports that support the use of the treatments to be discussed describe few patients and incorporated varied durations of disease and follow-up. Results vary widely among studies. Therapy is arbitrarily divided into four levels: Levels 1 through 3 represent increasing severity of side effects; level 4 is only used when the other three levels have failed. The risk presented by the patient's disease must always be weighed against the risk of treatment and the likelihood of success. Patients who develop severe mucosal bleeding may require concurrent therapy with intravenous IgG, corticosteroids, or platelets. Table 2 lists the doses, response times, and side effects of each treatment, and Table 3 gives the costs of the various treatment options at one institution. As noted above, patients who can maintain safe platelet counts (>30 x 10⁹/L) do not require therapy.

Level 1 Therapy

Agents used for level 1 therapy have moderate risk and should be tried in the order described.

Corticosteroids

Prednisone, administered as described above, is the drug of choice if safe platelet counts can be maintained with doses that are acceptable for long-term use (5 to 10 mg/d). If patients require unacceptable doses of prednisone, pulsed dexamethasone (40 mg/d for 4 days every 28 days for 6 cycles) should be tried. Preliminary results of small studies have ranged from excellent to poor [50, 68, 69].

If the dexamethasone course is also ineffective, patients should be treated with other agents. Many patients require concomitant corticosteroids to maintain their platelet counts while other agents are tried. If a response occurs, corticosteroid therapy should be tapered and withdrawn.

Vinca Alkaloids

Vincristine (1 to 2 mg/wk intravenously) is suggested because occasional patients achieve complete remission [17, 18, 21-28]. If a permanent response is not achieved, vincristine therapy should be discontinued after 4 to 6 doses before peripheral neuropathy becomes a problem.

Danazol

Danazol is given at a dose of 200 mg orally four times daily for at least 6 months [29-37] because responses are often slow [36]. Liver function should be checked monthly. If a response occurs, doses should be continued at full levels for at least 1 year and then tapered by 200 mg/d every 4 months.

Colchicine

In patients who do not respond to vinca alkaloids and danazol, 0.6 mg of colchicine should be administered orally three times daily for at least 2 months [70, 71]. If a response occurs, the dose may be tapered to the lowest level that results in safe platelet counts; withdrawal causes relapse. Diarrhea may require a reduction in dose and may be minimized with diphenoxylate or loperamide.

Dapsone

The dose for dapsone therapy should be 75 mg orally per day; responses occur within 2 months [59-61]. Patients should be screened for erythrocyte glucose-6-phosphate dehydrogenase because persons with low values can have serious hemolysis.

Level 2 Therapy

Level 2 treatments have greater toxicity than those of level 1 and should be given only in cases of dangerously low platelet counts (10 to 15 x 10⁹/L), mucosal bleeding, or both. The role of staphylococcal-A immunoadsorption is unclear, and some believe that the side effects outweigh the benefits [72]. If used, staphylococcal-A immunoadsorption should be tried before cyclophosphamide or azathioprine because the best results occur in patients with higher platelet counts and more recent disease [73]. It is advisable to use cyclophosphamide before azathioprine because responses are more rapid. If the patient does not respond to cyclophosphamide, azathioprine should be administered.

Staphylococcal-A Immunoadsorption

Patients receiving staphylococcal-A immunoadsorption should be treated 3 times per week for 2 weeks, according to manufacturer's instructions; responses may occur after 2 to 3 cycles of therapy [73]. If no response occurs, therapy should be discontinued.

Cyclophosphamide

The dosage for treatment with cyclophosphamide should begin with 150 mg/d orally and should be adjusted with the aim of maintaining mild neutropenia [18, 38-43]. Responses occur within 8 weeks. If the platelet count becomes normal, patients should receive the full dose for 3 additional months, and then therapy should be stopped. If relapse occurs, such long-term risks as secondary malignancy must be weighed against the benefits of resuming therapy. During treatment, patients should drink at least 2 L of liquid daily to prevent hemorrhagic cystitis, and the blood count should be checked at least weekly.

Azathioprine

Responses occur slowly over 3 to 6 months with azathioprine therapy, and many physicians stop therapy before giving it a fair trial [18, 44-49]. The dosage should begin at 150 mg/d orally and should be adjusted to result in mild neutropenia. If a response occurs, the full dose should be continued for 18 months, and therapy should be discontinued thereafter. Many patients who have relapse respond if therapy is repeated. The decision to continue therapy for the long term (>18 months) is determined by weighing the risk–benefit ratio. Because neutropenia is the major complication, serial blood counts should be monitored.

Level 3 Therapy

Level 3 treatments should be reserved for patients who are refractory to the first two levels of therapy and who have life-threatening symptoms or extremely low platelet counts (<10 x 10⁹/L). High-dose cyclophosphamide should be given first because it is easier to administer and costs less; if high-dose cyclophosphamide therapy is unsuccessful, combination chemotherapy is indicated. Toxicity and response rates are similar in the few patients who have received these forms of treatment [43, 74]. If severe neutropenia (leukocyte counts <0.5 x 10⁶/L) occurs, prophylactic ciprofloxacin, 750 mg twice daily, and rifampin, 300 mg twice daily, should be administered until counts reach safe levels.

High-Dose Cyclophosphamide

High-dose cyclophosphamide should be given at 1.0 to 1.5 g/m² of body surface area intravenously and should be repeated at 4-week intervals. If patients do not respond after two courses, therapy should be discontinued. In patients who do respond, at least three courses of treatment are suggested. A high fluid intake (3 to 4 L/d orally or intravenously during and for 3 days after therapy) is recommended to prevent hemorrhagic cystitis. Blood counts should be checked frequently during the first 2 weeks and at least weekly thereafter.

Combination Chemotherapy

No data support one form of combination chemotherapy [74] over another. I use the following: cyclophosphamide, 750 mg/m² intravenously on days 1 and 8; prednisone, 40 mg/m² orally on days 1 to 14; and etoposide, 100 mg/m² intravenously on days 14, 15, and 16. If patients do not respond after two cycles, therapy should be stopped. Patients who do respond should be given 6 cycles at 4-week intervals. Other combinations have been successful [74]. At least 2 L of fluid daily should be taken for 3 to 4 days after cyclophosphamide therapy, and antacids or histamine H₂-receptor blockers should be given during and for at least 1 week after corticosteroid therapy. Blood counts should be checked at least weekly. If neutropenia is severe, etoposide doses should be decreased in subsequent courses.

Level 4 Therapy

Level 4 treatments are used when all others have failed; experience with them is limited, response rates are low, or their expense or frequency of administration weigh against their use. Many patients who receive these treatments require concomitant corticosteroid therapy.

Interferon

The recommended dose is 3 million U subcutaneously three times per week for 4 weeks [75-79]. One report [80] suggests that more prolonged therapy may be beneficial. Interferon may suppress platelet production, and this suppression was thought to have contributed to a patient's death in at least one case [81].

Gammaglobulin

Some patients in whom all other forms of therapy are ineffective continue to respond to intravenous IgG (0.5 to 1.0 g/kg as needed to prevent mucosal bleeding). Responses are almost always transient, lasting as long as a few weeks [82]. This treatment is expensive (Table 3) and inconvenient but is sometimes the only effective option.

Vinblastine

In patients who respond to vinca alkaloids, vinblastine (5 to 10 mg intravenously every 1 to 4 weeks) can occasionally be used effectively for long intervals.

Cyclosporine

The suggested dose for cyclosporine [83-85] is 1.25 to 2.5 mg/kg orally twice daily (total daily dose, 2.5 to 5.0 mg/kg). Serum creatinine and cyclosporine levels must be measured periodically, and the dose of cyclosporine should be adjusted as needed. In view of its serious side effects, this drug should be used cautiously and only when no good alternatives are available.

Special Situations

Emergency Treatment

In patients who have severe bleeding or require surgery, a rapid increase in the platelet count can usually be achieved with gammaglobulin or methylprednisolone therapy.

Gammaglobulin, 0.5 to 1.0 g/kg, should be followed by platelet transfusion (one plateletpheresis or 6 to 8 U of random platelets). This treatment may be repeated [86]. I have had anecdotal experience with three patients in whom this regimen was ineffective but who responded to gammaglobulin, 1 g/kg, and platelets, 1 U/h, given by continuous infusion for 24 hours.

Methylprednisolone, 1 g/d intravenously for 3 days, should be followed by platelet transfusion after the first dose. Serum potassium levels must be normal before this therapy can be used. If the therapy is used frequently, rapidly occurring osteoporosis with compression fractures may be a problem, particularly in young patients in whom calcium turnover is rapid.

Therapy during Pregnancy

Mother

For pregnant women, the goal is to maintain safe platelet counts (>30 x 10⁹/L) However, a higher platelet count (50 to 100 x 10⁹/L) is desirable at delivery if certain types of anesthesia (such as epidural anesthesia) are to be used. Either intravenous IgG or corticosteroids can be used to support the platelet count, although some groups prefer intravenous IgG [87]. Splenectomy is indicated only when maternal benefits outweigh the risk for increased fetal death. If splenectomy is done, it should be done in the second trimester. Immunosuppressants are contraindicated.

Fetus

Maternal antibody passes through the placenta and may cause fetal thrombocytopenia. It has been postulated that trauma during vaginal delivery may precipitate central nervous system bleeding and that cesarean section may obviate this problem; however, this hypothesis has been questioned [88]. Of 893 deliveries evaluated, 288 live-born infants in 11 series satisfied the following criteria: more than 10 entrants per series (to reduce worse-case reporting bias), platelet counts from the fetus and neonate, and data on morbidity and death. This group had no deaths or intracranial hemorrhages; secondary morbidity (purpura or mucosal bleeding) occurred in 3.5% of the patients. Thrombocytopenia occurred with platelet counts less than 50 x 10⁹/L in 10.1% of patients and with counts less than 20 x 10⁹/L in 4.2% of patients at delivery. In the same report [88], the neonatal platelet count was provided for 552 pregnancies but the time at which the count was done was not stated. In this group, major adverse events occurred in seven patients (1.2%; two deaths and five intracranial hemorrhages), but it could not be determined whether they were caused by the delivery or by declining platelet counts after delivery. Therefore, the effect of the method of delivery on these events was unevaluable. The results suggest that the rate of severe morbidity or death in the fetus is less than 1.2%. Neither group showed evidence that cesarean section was associated with lower rates of death or morbidity than was vaginal delivery.

Delivery

It is known that no correlation exists between maternal and fetal platelet counts [89], that treating the mother with either corticosteroids or intravenous IgG does not affect the fetal platelet count [90], and that mothers with ITP who remain severely thrombocytopenic despite splenectomy or who have delivered a thrombocytopenic infant in the past are more likely to give birth to a thrombocytopenic infant [91].

An accurate, risk-free method of determining fetal platelet count would be useful. However, the two available methods have problems. First, fetal scalp sampling is technically difficult and often inaccurate. The use of this technique by one group resulted in more incorrect than correct decisions about whether to do cesarean section [92]. Some investigators report that examination of a fetal blood smear allows an accurate estimate of the platelet count [93]. Second, cordocentesis, although it provides an accurate platelet count, is associated with a fetal mortality rate of more than 1% in most cases [94], although one group reported no complications if the procedure was done after 35 weeks [95].

Because it is difficult to obtain accurate and safe fetal platelet counts, opinions differ on the management of delivery. One group [87] recommends that the means of delivery be determined solely on the basis of obstetric indications and that sampling of fetal blood is not indicated. They believe that no evidence shows that cesarean section is safer, that scalp sampling is inaccurate, and that fetal mortality rates associated with cordocentesis are greater than those associated with vaginal delivery. Another group [95], which has extensive experience with cordocentesis, recommends fetal sampling for all mothers with ITP who are being considered for vaginal delivery. They advise sampling after 37 weeks if the mother's bleeding time is normal and her platelet count is greater than 50 x 10⁹/L. Cesarean section is done if the fetal count is less than 50 x 10⁹/L. All agree that cordocentesis is contraindicated unless the operator is very experienced.

After Delivery

The infant's platelet count often declines during the first week after delivery and must be monitored [96]. For severe thrombocytopenia or mucosal bleeding, intravenous IgG is the treatment of choice. Platelet transfusions that are cytomegalovirus negative and irradiated can be added in the event of severe bleeding.

Therapy for Central Nervous System Bleeding

If central nervous system bleeding is suspected, computed tomography should be done to localize the site of bleeding. If central nervous system bleeding is present, intravenous IgG and platelet transfusions should be given to keep the platelet count above 50 x 10⁹/L, even if continuous infusion is required. If posterior compartment bleeding is noted, splenectomy (if relevant) should be done and followed by craniotomy. For hemispheric bleeding, the need for surgery depends on neurologic status and response to therapy.

Nonspecific Measures To Control Bleeding

Aminocaproic acid has been used to control hemorrhage in immune and nonimmune patients with thrombocytopenia [97, 98]. A loading dose of 0.1 g/kg is given over 30 to 60 minutes and is followed by 6 g every 6 hours. Once bleeding is controlled, the dose is tapered to 1 to 3 g every 6 hours. Potential side effects include dizziness, hypotension, abdominal discomfort, nausea, vomiting, diarrhea, headache, arrhythmias, rash, delirium, myopathy, seizures, and thrombogenic phenomena. Despite many potential side effects (Table 2), few were reported in two studies [97, 98], and the response to therapy in these studies was impressive.