Laboratory markers in drug safety research: studies on drug-induced thrombocytopenia

Abstract

It regularly happens that medications are found to cause severe adverse effects that remained undetected during premarketing research. From the perspective of the patient, the public and drug manufacturers there is a need for better tools for earlier detection, quantification and mechanistic unraveling of associations between drug exposure and adverse health outcomes in patient populations. A tool that is considered to be valuable for this purpose is a database with patient-oriented data on medication exposure and laboratory test results collected in clinical care. The objective of this thesis was to investigate the value of laboratory data collected in patient care for drug safety research. We focused on drug-induced thrombocytopenia as an example. First, we investigated the association between thrombocytopenia and exposure to drugs most often reported to cause thrombocytopenia. For this study we used data from the PHARMO Record Linkage System. To identify patients with thrombocytopenia hospital discharge diagnoses were used. An increased risk for thrombocytopenia following exposure to beta-lactam antibiotics was found. The expected increase in risk for thrombocytopenia could not be confirmed for the other medications investigated. This may be the result of limited statistical power, in which incomplete identification of patients with drug-induced thrombocytopenia by using hospital discharge diagnoses could play a role. Following-up this study we compared the use of hospital discharge diagnoses for thrombocytopenia with the use of platelet measurements as strategy for case-finding potential drug-induced thrombocytopenia from health care data. We found that using platelet measurements is a more sensitive approach for this purpose. However a low platelet count was observed to be non-specific for potential drug-induced thrombocytopenia. For this study we used data from the Utrecht Patient Oriented Database (UPOD). UPOD is a recently established database for (pharmaco-)epidemiological research, encompassing automated data on laboratory test results, medication exposure, hospital discharge diagnoses, medical procedures and patient demographics for all patients treated at the UMC Utrecht. By using data from UPOD we performed three studies concerning different aspects of drug-induced thrombocytopenia: the incidence and relative risk of chemotherapy-induced thrombocytopenia, a biomarker for the mechanism of chemotherapy-induced thrombocytopenia and compliance with recommendations for monitoring the platelet count for heparin-induced thrombocytopenia. We respectively found that the cytostatic drugs carboplatin, gemcitabine and paclitaxel are associated with the highest risk for thrombocytopenia in oncology patients treated in clinical practice, platelet size indices can not be used to distinguish bone marrow and immune-related causes of chemotherapy-induced thrombocytopenia and compliance to recommendations to monitor for heparin-induced thrombocytopenia in patients treated with low molecular weight heparin is low. We concluded that linking laboratory and medication data with in a research database is a valuable tool for investigating the safety of drugs in patient populations, including the investigation of the incidence, risk factors and monitoring for adverse drug reactions that can be detected with a biochemical test (such as drug-induced blood disorders) and the identification of potential biomarkers for adverse drug reactions. The knowledge from this research is valuable in maximizing the benefits of effective medications in patients treated in clinical practice.