Systematic assessment of pharmaceutical prescriptions in association with cancer risk: a method to conduct a population-wide medication-wide longitudinal study

Abstract

It is a public health priority to identify the adverse and non-adverse associations between pharmaceutical medications and cancer. We search for and evaluate associations between all prescribed medications and longitudinal cancer risk in participants of the Swedish Cancer Register (N = 9,014,975). We associated 552 different medications with incident cancer risk (any, breast, colon, and prostate) during 5.5 years of follow-up (7/1/2005-12/31/2010) in two types of statistical models, time-to-event and case-crossover. After multiple hypotheses correction and replication, 141 (26%) drugs were associated with any cancer in a time-to-event analysis constraining drug exposure to 1 year before first cancer diagnosis and adjusting for history of medication use. In a case-crossover analysis, 36 drugs (7%) were associated with decreased cancer risk. 12 drugs were found in common in both analyses with concordant direction of association. We found 14, 10, 7% of all drugs associated with colon, prostate, and breast cancers in time-to-event models. We only found 1, 2%, and 0% for these cancers, respectively, in case-crossover analyses. Pharmacoepidemiologic analyses of cancer risk are sensitive to modeling choices and false-positive findings are a threat. Medication-wide analyses using different analytical models may help suggest consistent signals of increased cancer risk.

Figure 1. Overview of method to associate 552 medications with cancer risk.

(A) The data source was the Swedish longitudinal database, consisting of the Cancer Registry and Prescribed Drug Register (“ATC” = Anatomical Therapeutic Classification). (B) We split data into a training and testing dataset by location. (C) We conducted 2 possible analyses for each cancer type (e.g., any, breast, colon, and prostate) with (B), Cox proportional hazards regression (Cox PH), and a Case-Crossover (CC) analyses, only adjusting for sex in the any cancer or colon cancer outcomes, (D) Association testing. (E) Claiming a verified signal (p < 10⁻⁵ in both training and testing datasets) (F) Estimate concordance between each analysis, and (G) Estimate concordance between each analysis and the previous literature.

Figure 2. Manhattan plots (−log10(p-value)) for each drug categorized by Anatomical Therapeutic Chemical anatomical main group) in Cox analyses by cancer type.

Orange color denotes tentative signals. P-values lower than 1 × 10⁻¹⁰⁰ set to 1 × 10⁻¹⁰⁰ for clarity. Upward triangles indicate HR >1 and downward triangles HR <1.

Figure 3. Manhattan plot (−log10(p-value)) for each drug categorized by Anatomical Therapeutic Chemical anatomical main group) in case-crossover analyses.

Orange color denotes tentative signals. P-values lower than 1 × 10⁻¹⁰⁰ set to 1 × 10⁻¹⁰⁰ for clarity. Upward triangles indicate OR >1 and downward triangles OR <1.