Precision under pressure: tandem approach to HIT diagnosis Free

In this issue of Blood, Steinauer et al prospectively validate and compare an approach combining pretest probability with tandem rapid immunoassays (IAs) to improve the accuracy and speed of heparin-induced thrombocytopenia (HIT) diagnosis in a real-life setting.¹ The authors report that sequential combinations of rapid IA led to high sensitivity and specificity (≥96%), 0 false-negatives, and very few indeterminant results. This approach outperformed single or simultaneous IA algorithms in one of the largest, prospective cohorts of suspected HIT cases.

HIT is an immune-mediated adverse drug reaction to heparin, leading to a severe prothrombotic state due to circulating antibodies directed against platelet factor 4 (PF4) bound to heparin.² Approximately 0.2% to 3% of patients with immune response to PF4/heparin complexes develop severe clinical complications of thrombocytopenia and thrombosis.³ Prompt recognition and diagnosis, avoidance of heparin-containing products, and switching to an alternate anticoagulant are essential to prevent life- and limb-threatening arterial and venous thromboses.⁴ Both heparin use and thrombocytopenia are common in hospitalized patients; thus, the suspicion for HIT is raised frequently in the inpatient setting. Due to inherent limitations in assay availability and performance characteristics across institutions, patients with suspected HIT are often exposed to diagnostic delays and overtreatment, which have direct clinical and cost implications when deciding to switch to a nonheparin anticoagulant.

Diagnosis of HIT is complex, requiring both pretest probability assessment (ie, 4-T score) combined with a search for anti-PF4/heparin antibodies.^5,6 Although several laboratory assays exist, the optimal approach to diagnosis, balancing speed, availability, cost, and accuracy, has remained a pertinent question in the field. In patients with suspected HIT, the sequential approach to diagnosis first includes the use of a screening IA to detect the presence of anti-PF4/heparin antibodies, followed by a functional assay to determine if those antibodies activate platelets in a heparin-dependent manner. The classic screening IA, the enzyme-linked immunosorbent assay (ELISA), is widely available and highly sensitive, but specificity is compromised by the occurrence of asymptomatic seroconversion leading to overdiagnosis.³ Functional tests, such as the serotonin-release assay (SRA), are considered the gold standard for the diagnosis of HIT, yet a key limitation is a lack of widespread availability leading to long turnaround times and delays in diagnosis. The window between a preliminary positive ELISA to confirmatory SRA exposes a meaningful proportion of patients with thrombocytopenia to therapeutic nonheparin anticoagulants that are not only costly but may be unnecessary and potentially harmful. To bridge this gap, several rapid IAs have been developed in recent years, including chemiluminescent IA (CLIA) and latex immuneturbidimetric assay (LIA), and provide access to rapid results within 10 to 30 minutes with higher specificity.⁷ In 2018, American Society of Hematology guidelines for HIT identified the incorporation of rapid IA into diagnostic algorithms as a key research priority in the field.⁶ 

In response, several diagnostic approaches using rapid IA screening have been developed, including the Lausanne algorithm, which uses a Bayesian combination of pretest probability and sequential rapid IA^7,8; the Hamilton algorithm, based on simultaneous rapid IA without considering the clinical context⁹; and the TORADI-HIT algorithm, a machine-learning model incorporating single rapid IA with clinical and laboratory features.¹⁰ In this present study, Steinauer et al prospectively validated the current Lausanne algorithm, combining 4-T score followed by sequential CLIA and LIA in a cohort of nearly 1200 adult cases of suspected HIT in Switzerland and compared its performance to the existing Hamilton and TORADI-HIT algorithms. Remarkably, this validation study of the Lausanne algorithm confirmed high sensitivity and specificity ≥96% for HIT, with no false-negative predictions and a low rate of undetermined cases (<3%). This sequential Bayesian approach also outperformed the Hamilton and TORADI-HIT algorithms, which both suffered from higher rates of false-negative results, risking missed diagnoses with potentially fatal consequences in the clinical setting.

Potential limitations of this tandem IA approach include the cost and limited reagent stability, requiring continuous quality control. The importance of continuous monitoring of reagent stability cannot be overlooked, as the authors noted that they were able to perform fewer than 10 analyses for each LIA reagent kit before control values were no longer acceptable, considerably less than the >20 analyses per kit provided by the manufacturer. This potentially limits widespread adoption and implementation of this approach due to cost and quality variables, especially in nonspecialized laboratories with lower numbers of analyses and longer reagent storage times. However, due to limited stability of LIA agents, the authors suggest that a CLIA first, LIA second approach should be preferred, noting that the first IA result of the algorithm reached a conclusion in most cases, only requiring the second rapid IA in <20% of cases.

Taken together, the results of Steinauer et al reflect a potential paradigm shift of HIT diagnosis, validating an algorithm in a real-life setting that encompasses high prediction and rapid results in <1 hour, while limiting false-negative results. Although external, prospective validation is required to confirm reproducibility and generalizability, this approach could offer both rapid and accurate results with direct implications to improve patient care. Precision under pressure can thus be achieved to improve evidence-based clinical decision-making in patients with suspected HIT.

Conflict-of-interest disclosure: The authors declare no competing financial interests.