Development, validation, and pilot-testing of a comprehensive patient-reported EORTC custom questionnaire for monitoring real-world tyrosine kinase toxicity in chronic myeloid leukemia

Purpose In chronic myeloid leukemia (CML), treatment choices between tyrosine kinase inhibitors (TKIs) are guided by efficacy and tolerability, yet no instrument adequately captures tolerability from the patient perspective. As a result, TKI choice may not comprehensively take into account the prevalence and severity of what patients’ experience. We therefore developed, validated, and pilot-tested a new custom questionnaire to capture real-world patients’ experiences, and explored pilot data. Methods A mixed-methods study was conducted, following COREQ and STROBE guidelines. Development started with CML-specific items from existing instruments, followed by semi-structured sessions with purposefully sampled patients, and convenience-sampled professionals, focused on comprehensibility, comprehensiveness, relevance, acceptability, workflow, and graphics. Based on qualitative analyses, items were substituted or added from the EORTC Item library. Psychometric quantitative evaluation assessed filter question validity, redundancy, reliability, measurement error, responsiveness, scale construction, and symptom severity differences in pilot data and meta-analyses. Results A new EORTC custom questionnaire was developed, covering 50 symptoms. Seven of those symptoms are covered by scales consisting of multiple items, with a Cronbach’s alpha ≥ 0.73 for each of those scales. Psychometric properties further showed fair-to-excellent reliability and positive agreement between time points ranging from 45 to 93% in test–retest, while responsiveness was indicated for half of symptoms. Twenty-four generic filter questions are available for computer adaptive testing, an option preferred by 87% of patients. For 23 issues, no existing EORTC item was available; these need to be added later. Pilot data showed that the severity of symptoms, as reported by patients, differed significantly between the three most common TKIs in a third of symptoms. These significant differences ranged from 5 to 30%. The highest difference was for swelling of the eyes or face, which was reported 30% more severe in patients on imatinib, compared to patients on nilotinib, when adjusted for age, gender, years since diagnosis, years on current treatment, and TKI-dosage. We could not adjust for comorbidity or comedication. Conclusion We developed and validated an EORTC custom questionnaire and workflow to monitor 50 CML patient-reported toxicity symptoms in clinical care. Sexuality problems are excluded from pre-existing instruments, yet pilot data revealed these are prevalent and severe from patients’ perspective. One-third of evaluated symptoms varied in severity by type of TKI treatment, independent from age, gender, years since diagnosis, years on current treatment, and TKI-dosage. These findings can tentatively inform shared decision-making, when switching TKIs due to intolerability.