Diagnostic Interpretation of Laboratory Results

In patients with a high pretest probability of HIT who have a negative screening test, the test should be repeated and the complementary activation or antigen assay should be performed. The diagnosis of HIT is very unlikely if both activation and antigen assays are negative. In patients with a moderate pretest probability who have one or more positive tests for HIT, the final diagnosis may well rest on the overall clinical picture, rather than on the test result alone. This conclusion results from two clinical realities: (1) sensitive HIT assays frequently detect clinically insignificant anti-PF4-H antibodies in patients who have received heparin for more than 5 days, and (2) thrombocytopenia, whether caused by HIT or not, is common in clinical practice. There is evidence that positive washed platelet activation assays for HIT have greater diagnostic specificity for clinical HIT (Warkentin et al., 2000, 2005a), especially when strong, rapid platelet activation is produced by patient serum. Regardless, these considerations underscore the importance of conceptualiz-

Specificity

FIGURE 7 Sensitivity-specificity tradeoffs for diagnosis of HIT (receiver operating characteristic curve analysis). The arrows indicate various cut-offs between positive and negative test results; e.g., the open circle indicates 90% serotonin release (postcardiac surgery patient) using a washed platelet activation assay (serotonin release assay). The likelihood ratio for HIT for a given positive test result can be estimated from the graph, using the formula: likelihood ratio = sensitivity/ (1 - specificity). Thus, for 90% serotonin release (postcardiac surgery), the estimated likelihood ratio is 0.7/(1 - 0.965) = 20. Source: From Warkentin, 2003b.

Specificity

FIGURE 7 Sensitivity-specificity tradeoffs for diagnosis of HIT (receiver operating characteristic curve analysis). The arrows indicate various cut-offs between positive and negative test results; e.g., the open circle indicates 90% serotonin release (postcardiac surgery patient) using a washed platelet activation assay (serotonin release assay). The likelihood ratio for HIT for a given positive test result can be estimated from the graph, using the formula: likelihood ratio = sensitivity/ (1 - specificity). Thus, for 90% serotonin release (postcardiac surgery), the estimated likelihood ratio is 0.7/(1 - 0.965) = 20. Source: From Warkentin, 2003b.

ing HIT as a clinicopathologic syndrome, in which both clinical information and results of HIT antibody testing are used for diagnosis.

The diagnostic usefulness of certain laboratory tests for HIT is shown in Figure 7 (Warkentin, 2003b). Both the SRA and an in-house EIA that detect only HIT-IgG were very sensitive and specific for clinical HIT in postorthopedic surgery patients. The diagnostic usefulness of these assays was somewhat less in a postcardiac surgery population. For example, among postcardiac surgery patients, the likelihood ratio of a strong-positive SRA result (90% serotonin release; see open circle in Fig. 7) is about 20. The likelihood ratio, which is defined as the extent to which a given test result alters the physician's estimate of the pretest probability of HIT, is defined as sensitivity/(1 — specificity). In this example, the corresponding likelihood ratio is 0.70/(1 — 0.965) = 20. Thus, if the physician had estimated a pretest probability of 50% (odds of 0.5:0.5), then this test result would increase the posttest probability of having HIT to more than 95% (0.5:0.5 X 20:1 = 20:1, or 95.2%). In contrast, the high sensitivity of this assay to detect clinically important HIT antibodies (>95%) means that a negative test result lowers the posttest probability to less than 5%.

The diagnostic impact of such a strong-positive SRA result (90% serotonin release) is even greater in postorthopedic surgery patients, for whom the corresponding likelihood ratio is about 85, i.e., 0.85/(1 — 0.99). As before, a negative test result essentially rules out HIT.

Although the EIA that detects only HIT-IgG antibodies has lower diagnostic specificity than the SRA, it remains a useful assay. The likelihood ratios for a strong positive test result (e.g., optical density of 1.5) range from about 10 to 40 for postcardiac and postorthopedic surgery patients, respectively. Also, its high sensitivity (>98%) means that a negative test generally rules out HIT (Greinacher et al., 2007).

Thus, HIT antibody testing is among the most useful of platelet immunology assays. For comparison, Figure 7 also shows the profile of a "noninformative assay" (see line A). This is the profile for various tests of "platelet-associated IgG" for the diagnosis of autoimmune thrombocytopenia. Certain glycoprotein-specific platelet antibody tests have operating characteristics intermediate between those for HIT and a noninformative assay. For example, the monoclonal antibody immobilization of platelet antigens (MAIPA) assay has only moderate sensitivity but high specificity for diagnosis of autoimmune thrombocytopenia.

Figure 7 shows the operating characteristics for two assays: the SRA and a PF4-H-EIA that detects only IgG antibodies. Recently, Warkentin and coworkers (2005a) used archived plasma samples to compare the operating characteristics of a commercial EIA (PF4-polyvinylsulfonate EIA, GTI) that detects antibodies of all three major immunoglobulin classes (IgG, IgA, IgM) (Fig. 8). This study showed that the additional detection of IgA and IgM class antibodies worsened the test's operating characteristics, by detecting clinically insignificant IgA and IgM antibodies without any offsetting improvement in test sensitivity. These data are consistent with the view that HIT is usually (perhaps invariably) caused by IgG antibodies, i.e., the only antibody class able to activate platelets via their IgG (Fcy) receptors.

A practical implication of Figure 7 is that the magnitude of a positive HIT antibody test provides diagnostically useful information, with a strong positive result associated with a greater likelihood of a patient having clinical HIT than a weak positive result (Warkentin, 2003b; Zwicker et al., 2004). Similarly, if two sensitive and complementary tests for HIT antibodies (washed platelet activation

0 0

Post a comment