Refine
Document Type
- Article (2)
Language
- English (2) (remove)
Has Fulltext
- yes (2)
Is part of the Bibliography
- no (2)
Keywords
- Lung cancer (2) (remove)
Institute
- Medizin (2) (remove)
Introduction: Cancer patients tend to prefer oral instead of parenteral chemotherapy. To date, there is little evidence on the medication adherence in cancer patients. We investigated medication adherence to tyrosine kinase inhibitors in patients suffering from non-small cell lung cancer. Methods: Tyrosine kinase inhibitor adherence was measured electronically by MEMS® (medication event monitoring system) over at least six months. Adherence rates were calculated in terms of Dosing Compliance, Timing Compliance, Taking Compliance, and Drug Holidays. Patients were dichotomized as adherent when Dosing Compliance and Timing Compliance were ≥80%, Taking Compliance ranged between 90 and 110%, and <1 Drug Holiday was registered. Quality of life was assessed by two questionnaires (EORTC QLQ-C30 version 3.0, EORTC QLQ-LC13) at three time points. Adverse drug events were reported via patient diaries. Results: Out of 32 patients enrolled, data from 23 patients were evaluable. Median Dosing Compliance, Taking Compliance, and Timing Compliance adherence rates of tyrosine kinase inhibitor intake amounted to 100%, 98%, and 99%, respectively; Drug Holidays were observed in three patients. Four patients were dichotomized as non-adherent. Three of them had a twice-daily tyrosine kinase inhibitor regimen. Median quality of life scores amounted to 67 (max. 100) and remained unchanged over the study period. Fatigue and rash were the most frequently reported adverse drug events. Conclusion: Medication adherence of non-small cell lung cancer patients treated with tyrosine kinase inhibitors was extraordinarily high and is likely to support the effectiveness of tyrosine kinase inhibitor treatment and a good quality of life over a long period of time. Adherence facilitating information and education is especially relevant for patients taking tyrosine kinase inhibitors in a twice-daily regimen.
BACKGROUND: Vermeulen et al. 2014 published a meta-regression analysis of three relevant epidemiological US studies (Steenland et al. 1998, Garshick et al. 2012, Silverman et al. 2012) that estimated the association between occupational diesel engine exhaust (DEE) exposure and lung cancer mortality. The DEE exposure was measured as cumulative exposure to estimated respirable elemental carbon in μg/m(3)-years. Vermeulen et al. 2014 found a statistically significant dose-response association and described elevated lung cancer risks even at very low exposures.
METHODS: We performed an extended re-analysis using different modelling approaches (fixed and random effects regression analyses, Greenland/Longnecker method) and explored the impact of varying input data (modified coefficients of Garshick et al. 2012, results from Crump et al. 2015 replacing Silverman et al. 2012, modified analysis of Moehner et al. 2013).
RESULTS: We reproduced the individual and main meta-analytical results of Vermeulen et al. 2014. However, our analysis demonstrated a heterogeneity of the baseline relative risk levels between the three studies. This heterogeneity was reduced after the coefficients of Garshick et al. 2012 were modified while the dose coefficient dropped by an order of magnitude for this study and was far from being significant (P = 0.6). A (non-significant) threshold estimate for the cumulative DEE exposure was found at 150 μg/m(3)-years when extending the meta-analyses of the three studies by hockey-stick regression modelling (including the modified coefficients for Garshick et al. 2012). The data used by Vermeulen and colleagues led to the highest relative risk estimate across all sensitivity analyses performed. The lowest relative risk estimate was found after exclusion of the explorative study by Steenland et al. 1998 in a meta-regression analysis of Garshick et al. 2012 (modified), Silverman et al. 2012 (modified according to Crump et al. 2015) and Möhner et al. 2013. The meta-coefficient was estimated to be about 10-20 % of the main effect estimate in Vermeulen et al. 2014 in this analysis.
CONCLUSIONS: The findings of Vermeulen et al. 2014 should not be used without reservations in any risk assessments. This is particularly true for the low end of the exposure scale.