Those pesky platelets

An expert hematologist explains how to recognize and treat heparin-induced thrombocytopenia, idiopathic thrombocytopenic purpura and thrombotic thrombocytopenic purpura.


In the hospital, hematologists often feel more like gatekeepers than highly trained specialists, according to Joel Bennett, MD, a professor of medicine at the Hospital of the University of Pennsylvania in Philadelphia.

Colored scanning electron micrograph of platelets that have been activated Photo by PhotoResearchers
Colored scanning electron micrograph of platelets that have been activated. Photo by PhotoResearchers.

“Whenever a patient has a low platelet count, the physician assumes it's HIT [heparin-induced thrombocytopenia] and requests an HIT assay,” said Dr. Bennett during an Internal Medicine 2010 talk on thrombocytopenia. “No one wants to see a hematologist; they just want an assay.”

While HIT is common, there are many reasons a patient might be thrombocytopenic—defined as having a platelet count of less than 150,000/µL. The condition occurs in three ways:

  • decreased production of platelets by bone marrow megakaryocytes,
  • accelerated removal of circulating platelets—i.e., the usual platelet lifespan of 10 days is shortened—and
  • platelet sequestration in an enlarged spleen.
Dr Bennett If you stay in the surgical ICU long enough you are going to get thrombocytopenic Photo by Kevin Berne
Dr. Bennett: “If you stay in the surgical ICU long enough, you are going to get thrombocytopenic.” Photo by Kevin Berne.

When decreased platelet production is the cause, there is “usually something wrong with the bone marrow,” Dr. Bennett said. This could be a bone marrow stem cell disorder, like acute leukemia, aplastic anemia, myelodysplasia or paroxysmal nocturnal hemoglobinuria. It could also be a myelophthisic process, in which fibrosis, carcinoma or multiple myeloma is displacing the bone marrow. Chemotherapy or radiation therapy can cause decreased production, as well.

“If you drink too much alcohol at one time, that can also do it. HIV—you should always think of HIV when you see thrombocytopenia—but other viruses can do it, too,” Dr. Bennett said.

In thrombocytopenia due to an enlarged spleen, the physician can often feel the spleen. Platelet counts in these patients typically range from 40,000 to 50,000/µL.

Bleeding due to hypersplenism alone is unusual, he said. “If a patient with hypersplenism is bleeding, and has the usual platelet count of 50,000/µL, you need to look for something else,” Dr. Bennett said. “And if the platelet count is lower than 40,000/µL in someone with hypersplenism, there may be some other process superimposed upon it.”

The top three thrombocytopenia diagnoses for which a hematologist is called are idiopathic thrombocytopenic purpura (ITP), HIT, and thrombotic thrombocytopenic purpura (TTP), Dr. Bennett said. Each is diagnosed and managed in its own way.

ITP: A diagnosis of exclusion

ITP presents as isolated thrombocytopenia, i.e., there are no apparent causes of the condition. It's an autoimmune process, which means ITP involves the production of antiplatelet antibodies, Dr. Bennett said. “About half the time it is associated with other problems; 50% of the time, there is autoimmunity against platelets for no particular reason. Sometimes it is associated with collagen-vascular disease or low-grade lymphoproliferative disorders,” he said.

ITP incidence is 100 cases per million persons per year, so it's relatively common, Dr. Bennett said. There is no specific test for ITP, but it is usually the correct diagnosis for those patients who come in asymptomatic, and who have a normal history, physical exam, blood count and blood smear, except for low platelets. A bone marrow test typically isn't necessary, he said.

“On the other hand, a bone marrow test could be useful in older people, because part of the differential diagnosis of this disorder is myelodysplasia, which gets more and more common as people get older. It can also be useful when you treat people who look like they have ITP but don't respond to treatment; in that case, they could have something else,” Dr. Bennett said.

Another differential diagnosis for ITP is pseudothrombocytopenia, in which people have antibodies that cause their platelets to clump when there's no calcium present. These clumps won't be measured by an automated platelet counter, because they won't go through the aperture, and the result is a falsely low platelet count. A blood smear will reveal this condition, which is why it's important to do one on patients who appear to have a low platelet count, he said.

Other differential diagnoses include occult liver disease, disseminated intravascular coagulation (DIC), TTP, HIT and gestational thrombocytopenia. As for the latter, platelet counts do fall in pregnancy, sometimes as low as 70,000/µL, but come up again several months after the pregnancy has ended.

In about one case per 100,000 persons per year, thrombocytopenia is drug-induced. Drugs can impair platelet production or accelerate platelet removal. This usually happens within a week or two of starting a drug, “so if someone comes in and you think this is the process, it's good to take a history and find out what drugs have recently been prescribed,” Dr. Bennett said. Usually people get better within several days of stopping the offending drug, although sometimes it can take a couple of weeks. Although it is less common, there have also been instances of this condition appearing from a drug that was taken several years prior, he said.

“If you stay in the surgical ICU long enough, you are going to get thrombocytopenic. It's a rule of mine. Sometimes it is obvious what the cause might be. Other times it's not, and you have to stop all unnecessary medication or switch to another drug. It is a help that a lot of the drugs people are on are unnecessary,” Dr. Bennett said.

Tegretol, GPIIb/IIIa antagonists, sulpha drugs, rifampin and vancomycin are all common drug culprits for thrombocytopenia, but “you can't be sure with most drugs, so it's a good idea to change or stop a patient's drugs if you think they could be a problem,” he said.

Though it's rare, people can be thrombocytopenic on a congenital basis, as well. Several years ago, one of Dr. Bennett's patients looked “for all the world like he had ITP,” but nothing his doctors tried made him better.

“His platelets were uniformly big, but I wasn't smart enough to realize that was a little strange. When the patient woke up after splenectomy he said, ‘Interestingly, doc, my sister has the same thing,’” Dr. Bennett recalled. “So, taking a history is important, and often there are clues on the blood smear.”

Managing ITP

For patients with suspected ITP whose platelet count is above 30,000/µL and who have no or minimal bleeding, it's fine to just watch them, Dr. Bennett said. Those with a count below 30,000/µL and/or symptoms and bleeding should be put on corticosteroids; Dr. Bennett said he typically starts with a prednisone dose of 1 milligram per kilogram of body weight.

Patients who have a platelet count below 10,000/µL and/or have major bleeding should start corticosteroids and also get intravenous immunoglobulin G (IVIgG) to get the platelet count up more quickly while waiting for the corticosteroids to kick in.

“For patients who respond to steroids—and most patients will—I keep them on the drugs until their platelet count is normal, then do a slow taper,” Dr. Bennett said.

Above all, be patient, he added.

“I have seen patients whose platelet counts increase to 100,000 or more and their hematologists rapidly cut the steroids to half, then half again, then half again—and they relapse. So I'm in no hurry to decrease the prednisone dose. I cut it by 20 [mg], then 20 [mg] again, and when I get down to 20 mg, I cut it by 5 mg or 10 mg each week, until I get down to nothing,” he said.

With some patients, one can get the platelet counts above 30,000/µL at a dose of 10 mg of prednisone or less. Others require a higher dose in order to keep the platelet count at that level, and if one cuts their dosage, they relapse. Because steroids can be toxic over time, the latter patients need another kind of treatment—either rituximab or a splenectomy, Dr. Bennett said.

While there are no clinically useful predictors of response, splenectomy has shown itself to be “very effective” in patients who fail steroids, with a response rate of 67% to 75%, and a low rate of relapse, Dr. Bennett said. Still, “it's an operation, and there is morbidity and mortality, so it's not trivial,” he said. Fewer than 1% of patients who have their spleens out will develop infectious complications, which can be minimized by giving pre-surgery vaccinations, he said.

The response rate to rituximab is 33% to 50%, but many patients will relapse, Dr. Bennett said. “I tell patients, ‘Listen, we can take your spleen out, and it works most of the time, but you could also die or bleed. Or, I can give you this drug [rituximab]. You get four injections, there is no surgery, but it doesn't work quite as well as splenectomy.’ And no surprise, most people opt for the rituximab. So patients tend to like this better, but I think splenectomy is the better choice.”

Platelet transfusions don't tend to increase platelet counts for patients with ITP and thus are only indicated for serious or life-threatening hemorrhage, he said. In those cases, it is useful to give a concomitant IVIgG infusion with the platelet transfusion, he added.

For patients with ITP who fail splenectomy, rituximab and prednisone, one can try additional immunosuppression, or the new thrombopoietin agonists (TPO-R agonists). Romiplostim (brand name: Nplate) works in about 80% of patients: Platelet counts go up, bleeding stops, and steroids can be decreased. Some patients do develop bone marrow reticulin fibrosis, which goes away once the drug is stopped, so the significance of that side effect isn't clear, Dr. Bennett said.

“Nplate is an option in patients who are really refractory,” he said. “I can tell you I have never used it; I've been waiting for a patient to try it. I have a number of refractory patients but somehow they have been well enough that I haven't had to use it yet.”

HIT the diagnostic mark

Heparin is, of course, used often in the hospital, and several percent of patients who get unfractionated heparin may develop HIT, which is the most common cause of drug-induced thrombocytopenia in the hospital, Dr. Bennett said. Unlike ITP, HIT causes thrombosis instead of bleeding.

“It's a really frequent problem, and a bad one, because up to 50% of patients with HIT can develop arterial or venous thrombosis, and that can be life-threatening. So it really needs to be evaluated,” he said.

In about half the patients who have untreated HIT, one can detect new thromboses. In some series, about 21% of patients end up with amputations, and about 30% die from it, Dr. Bennett said.

HIT comes in two varieties. Type I usually occurs within the first one to four days of giving heparin, and the platelet count rarely dips below 100,000/µL. The heparin is stopped, and there are no consequences. It is not antibody-mediated. Type 2 HIT is immune-mediated, occurs in 1% to 4% of patients receiving unfractionated heparin, and usually begins five to 10 days after starting heparin, though it can be faster in patients who have been exposed to heparin before.

There are also several reports of delayed-onset HIT, in which patients stop heparin, go home, and come back to the hospital days to weeks later with thromboses or thrombocytopenia, Dr. Bennett said.

HIT happens most frequently with unfractionated heparin, but it can occur with low-molecular-weight heparin as well. Dosage or route of administration doesn't matter in terms of susceptibility, nor does the type of patient (cardiac, orthopedic, medical or surgical), he said.

“There are reports of people developing HIT from heparin eluting off Swan-Ganz catheters. So if someone develops HIT, you need to look them over and see what kind of tubes are stuck in them,” Dr. Bennett said.

HIT should be strongly considered in patients getting heparin whose platelet count drops under 150,000/µL or falls below 50% of baseline, or patients in whom new thrombotic or thromboembolic events occur. Basically, it should be suspected any time there is an unexplained decrease in platelet count, he said.

“As a house officer, I used to see people who would get heparin and keep having clots, and my response was, ‘Oh well, they are not getting enough heparin.’ I'm sure these patients had HIT, so we obviously did the wrong thing,” Dr. Bennett said.

Clearly, patients with suspected HIT should have their heparin stopped, he said. “You need to look at the patient and make sure they aren't getting heparin in a way that's not obvious, like it isn't in a catheter or heparin flushes or IVs or things like that.”

Next, one should send for a blood test for anti-heparin antibodies, and start the patient on an alternative anticoagulant until the diagnosis is clear and the platelet count has returned to normal, Dr. Bennett said.

Heparin antibody assays are either functional or immunological. The functional assays depend on the effect of heparin antibody complexes on platelets. One, the serotonin release assay, or SRA, is very sensitive and specific. “If somebody has a positive SRA, I feel really good about saying this person has HIT,” Dr. Bennett said. It can be falsely negative on occasion, however, he added.

SRAs require radioactivity, so they often aren't available. Most often, patients get the enzyme-linked immunosorbent assay (ELISA), which is very sensitive, such that many have positive results without having HIT, he said. That's why the clinical diagnosis is key.

“The other thing about the ELISA is that the degree of positive is quite important, so the more positive the ELISA, the more likely the patient is to have HIT and to get into trouble with thrombosis. [Patients with] ELISA measurements greater than one are more likely to have HIT or thrombosis,” Dr. Bennett said. “So if someone's ELISA is just slightly positive and the history isn't clear, then they probably don't have HIT. But if the ELISA is whoppingly positive and the SRA negative, then they probably do have HIT.”

Lepirudin (Refludan) and argatroban are the two U.S.-approved alternative anticoagulants for HIT patients. The former is primarily excreted by the kidneys, making it difficult to use in patients with renal failure, while the latter is metabolized by the liver. Bleeding is the most common side effect for both, and there is no antidote for either drug, Dr. Bennett said.

“If a patient starts bleeding after administration, all you can do is give blood and wait for the drug to go away,” Dr. Bennett said. Lepirudin has a half-life of 1.3 hours, while argatroban's is 39 to 51 minutes.

A major downside to lepirudin is that people who have gotten it twice have had bad allergic reactions, including death. “This makes me want to use argatroban,” Dr. Bennett said.

Though not FDA-approved for this use, pentasaccharide fondaparinux sounds “really cool” as a treatment for HIT, he added. Made from the pentasaccharide part of the heparin molecule that activates antithrombin, it is small and can be given subcutaneously. There has been at least one report of someone developing HIT from taking pentasaccharide, however, he added.

“It sounds like a drug I'd want to use, but I get nervous about it,” Dr. Bennett said.

TTP: Deadly yet treatable

The third type of thrombocytopenia is TTP, which is associated with thrombosis and characterized by five symptoms. One hundred percent of patients with TTP present with thrombocytopenia and with microangiopathic hemolytic anemia; 63% present with neurologic symptoms; 59% present with renal disease; and 24% present with fever, Dr. Bennett said.

Congenital TTP is rare, while idiopathic TTP is more common at 3.7 cases per million people per year. Drugs are one non-idiopathic cause of TTP, and culprits include ticlopidine (one case per 1,600 to 5,000 patients), clopidogrel (3.7 cases per million patients), quinine, mitomycin C, cyclosporine/tacrolimus, gemcitabine and pentostatin. Bone marrow transplant (0.5% to 63.6% incidence), pregnancy (one case per 25,000 patients), malignancy (5% to 6% incidence) and HIV are other non-idiopathic causes.

Untreated, the mortality rate from TTP is above 90%. The good news is that, when treated, 80% of patients survive TTP. “This is why the ER is so obsessed with TTP: It's bad, but if you treat it, you can save patients,” he said.

One of the top differential diagnoses is hemolytic-uremic syndrome, which is similar to TTP, but is mostly manifested by renal failure and mostly seen in children. “It's similar to TTP in that it is a microangiopathy, but its pathogenesis is different and it responds differently to therapy,” Dr. Bennett said.

Alternative diagnoses in patients who present with clinically suspected TTP include autoimmune disorders, sepsis, disseminated malignancy, malignant hypertension, HIT, or pre-eclampsia/ HELLP (hemolysis, elevated liver enzymes, low platelet count) syndrome during pregnancy or post-partum.

“You can't rely on the blood smear to diagnose TTP, because it can look like other conditions. That's why a history is important,” Dr. Bennett said. “The ADAMTS13 level doesn't work for diagnosis either. If it comes back low, it can help confirm suspected TTP, but not diagnose it.”

To treat TTP, start daily plasmapheresis until the platelet count and lactate dehydrogenase (LDH) are normal. An infusion of fresh frozen plasma can be given until plasmapheresis is instituted. At Dr. Bennett's hospital, physicians give patients steroids, he said.

“I've had bad experiences when I don't start [steroids], so based on anecdotes I give everyone prednisone and start daily plasmapheresis,” Dr. Bennett said.

Patients do relapse, he added: About 25% of patients with idiopathic TTP wind up back in the hospital with another episode.

If plasmapheresis doesn't work, there are a variety of things one can do, most of which don't have good literature to support their use, he said. These second-line therapies include increasing the frequency and volume of plasmapheresis, using cryosupernatant plasma during plasmapheresis, or trying the following drugs:

  • anti-platelet agents,
  • rituximab,
  • IVIgG,
  • vincristine,
  • cyclophosphamide,
  • azathioprine, or
  • cyclosporine A.

Splenectomy can also be an option in these cases, according to Dr. Bennett. “Splenectomy doesn't really make sense but sometimes it works. If you've been doing plasmapheresis for months, and the patient has recurrent neurologic symptoms and is not getting better, you may resort to it,” Dr. Bennett said.