DITP causation: 3 methods better than 1?

In this issue of Blood, Reese and colleagues describe 3 different methods to identify drugs that can cause acute—presumably immune-mediated—thrombocytopenia, assembling a list of 2 dozen well-documented culprits, but pointing to many more.¹ 

Clinicians evaluating patients with acute thrombocytopenia must consider the possibility of drug-induced immune thrombocytopenia (DITP). Unlike antibody-mediated thrombocytopenia caused by heparin—a relatively common prothrombotic syndrome with platelet-activating antibodies—DITP is rare, usually prohemorrhagic, and caused by non–platelet-activating drug- (or metabolite-) dependent, platelet-reactive antibodies.^2,3  Nevertheless, as thrombocytopenia often occurs in patients receiving numerous drugs, DITP is a common—and vexing—diagnostic consideration: clinicians want to know which drugs are plausible culprits, because indiscriminate discontinuation of all drugs can have serious consequences.

Reese and co-investigators have used 3 approaches to identify drugs that might cause DITP: (1) assessment of published case reports using defined clinical criteria; (2) laboratory evaluation for drug-dependent, platelet-reactive antibodies in patients with clinically suspected DITP; and (3) use of a data-mining algorithm to search for drugs within the Food and Drug Administration's Adverse Event Reporting System (AERS) database.¹  Together, these approaches identified 1468 drugs suspected of causing thrombocytopenia; however, only 23 drugs were identified by all 3 methods (see figure).

The first approach (University of Oklahoma) evaluated published cases using defined clinical criteria.^4,5  Drugs were implicated if at least one report merited classifying a drug as having a “definite” causal association, or if 2 or more reports met minimal criteria for a “probable” causal association. Although 253 drugs were thus evaluated, only 87 met the criteria for “definite/probable” causes of DITP.

The second approach used detection of drug-dependent, platelet-reactive antibodies by the Platelet and Neutrophil Immunology Laboratory of the BloodCenter of Wisconsin. This group, which provides comprehensive referral services for evaluating DITP, recently reported its experience using laboratory detection of drug-dependent antibodies as the key criterion for judging DITP causation.⁶  Although 202 different drugs were investigated by the Milwaukee lab, only 67 yielded supportive in vitro data. Combining these complementary clinical and laboratory approaches of the Oklahoma and Wisconsin investigators, 130 drugs had evidence of causality for DITP by either approach, but only 24 drugs were common to both methods (see figure).

A third, more novel approach used algorithmic “data mining” of the AERS database. Here, the authors obtained quarterly extracts of AERS data over 4 decades provided in response to a Freedom of Information request. The authors assessed for different drugs whether the ratio of observed reports exceeded the expected number of reports (“proportional reporting ratio”), using predefined thresholds of “signal of disproportionate reporting.” They found 1444 currently approved drugs for which AERS reports of thrombocytopenia existed; 573 of these agents had a statistically significant reporting association with thrombocytopenia.

As mentioned, only 23 drugs appear on all 3 lists (a 24th agent—amiodarone—was found using only the first 2 approaches). There seems little doubt that these 2 dozen drugs cause DITP, although 588 others identified by at least 1 of these 3 approaches remain possible causes of DITP. Most of these additional drugs were identified by data mining, but this method's specificity for DITP is unknown.

Physicians should be cautious about imputing causation simply because any particular drug being given to a thrombocytopenic patient appears on a “list.” A patient who really has DITP should exhibit 3 features.^2,7  First, the implicated drug should bear a temporal relationship with thrombocytopenia consistent with an immune-mediated etiology (usually, onset approximately 1 week after beginning a drug for the first time, or more quickly when exposed intermittently). Second, the platelet count nadir should be less than 30 × 10⁹/L, and usually less than 20 × 10⁹/L—this is because drug-dependent antibodies usually recognize glycoprotein (GP) IIb/IIIa and/or GPIb/IX antigen targets,^7,8  and these high copy number membrane surface GPs ensure that a high proportion of sensitized platelets are cleared. Finally, there should not be an alternative explanation for thrombocytopenia more highly probable than DITP.

The clinician should also be aware that many drugs beyond the 2 dozen listed do cause DITP. For example, the Milwaukee laboratory has identified at least 10 patients with suspected DITP as having ceftriaxone-dependent platelet-reactive antibodies,^1,6  a causal relationship also supported by data mining.¹  Inexplicably, however, no individual case reports of ceftriaxone-induced DITP exist.^4,5 

Another example: 4 years ago, the authors of this Inside Blood commentary jointly observed a 39-year-old male burn victim in whom acute, severe thrombocytopenia (platelet count nadir, 12 × 10⁹/L) began 1 week after beginning a course of mirtazapine (tetracyclic antidepressant), and in whom no other convincing explanation was found; the thrombocytopenia quickly resolved after stopping mirtazapine. A PubMed search (“mirtazapine” and “thrombocytopenia”) yielded a single report⁹  of mirtazapine-induced DITP, and in which mirtazapine-dependent GPIIb/IIIa-reactive antibodies were identified. The existence of this single published report, in our view, strongly supported the plausibility of a diagnosis of mirtazapine-induced DITP in our patient, even though we could not detect mirtazapine-dependent antibodies in our patient's serum by flow cytometry (in vitro testing for drug-dependent antibodies is relatively specific, but, unfortunately, not always very sensitive⁷ ). However, as only one (“probable”) report of mirtazapine-induced DITP currently exists,⁹  and as no such case has undergone serologic verification by the Milwaukee laboratory,⁶  mirtazapine failed on 2 counts to meet the authors' criteria for inclusion within the list of 24 drugs.¹  Encouragingly, AERS data mining did identify mirtazapine as a possible cause of DITP.¹ 

Between these 2 extremes of putative DITP due to ceftriaxone and mirtazapine—the former almost attaining the top tier and the latter meeting only the novel data-mining category—there undoubtedly exist many other drugs that cause DITP. The strongest candidates are those for which laboratory evidence of drug-dependent antibodies exist,⁶  followed by those for which reasonably convincing clinical evidence has been tabulated.^4,5  By these standards, 106 drugs (beyond those 24 listed) meet this level of culpability.¹  The contribution of Reese et al—including the searchable online data supplement—is a valuable “one-stop” resource for clinicians. Their 3-pronged approach of careful evaluation of clinical observations,^4,5  laboratory detection of drug-dependent antibodies,⁶  and, now, data mining of spontaneously generated AERS reports,¹  will help focus attention on those drugs that really cause DITP.