Digoxin Use After Diagnosis of Colorectal Cancer and Survival: A Population-Based Cohort Study

The main effect of digoxin, a cardiac glycoside, is on the inhibition of the sodium potassium ATPase pump, but it affects a number of pathways relevant to cancer. For instance, studies have shown increases in breast and uterus cancer probably related to estrogenic effects of digoxin (1). A recent large UK study reported a 40% increase in colorectal cancer risk in digoxin users (2), which the researchers suggest possibly reflects direct effects of the sodium potassium ATPase pump on tumorigenic pathways such as the Src/MAPK (3). In addition, preclinical studies have found that digoxin may reduce chemotherapy efficacy (4). These findings raise questions about the safety of digoxin in colorectal cancer patients. As there has been little research into digoxin use and colorectal cancer progression, we investigated whether colorectal cancer patients using digoxin had increased colorectal cancer–specific mortality.

The data source and methods have been discussed in detail previously (5). In brief, patients with newly diagnosed colorectal cancer from English cancer registries between 1998 and 2009 were identified from the National Cancer Data Repository (NCDR). Colorectal cancer–specific deaths up to January 2012 were identified from the underlying cause of death from Office of National Statistics (ONS) death registration data. The Clinical Practice Research Datalink (CPRD) provided digoxin use from general practitioner (GP) prescribing records. Potential confounders, including stage, grade, and treatment, were determined from NCDR. Smoking, body mass index, deprivation, and comorbidities were determined from GP records.

Patients were followed up from 1 year after colorectal cancer diagnosis until death, end of GP registration, last date of data collection from GP, or end of ONS follow-up. In the main analysis, time-dependent Cox regression models were used to calculate HRs and 95% confidence intervals (CI) for colorectal cancer–specific death for digoxin users compared with nonusers using a time varying covariate (lagged by 6 months). Adjusted analyses were conducted including potential confounders. Further analyses were conducted by number of digoxin prescriptions and number of defined daily doses. Analyses were repeated for all-cause mortality. Analyses were conducted in STATA 13.

The final cohort included 10,357 colorectal cancer patients with mean of 4.8 years of follow-up from diagnosis (maximum = 14 years) containing 2,724 colorectal cancer–specific deaths and 1,263 deaths from other causes. Table 1 shows characteristics by digoxin use. Digoxin use was associated with increased colorectal cancer–specific mortality before adjustment (HR, 1.25; 95% CI, 1.07–1.46), but after adjustment for confounders, the association was attenuated (adjusted HR, 1.10; 95% CI, 0.91–1.34) and no dose response associations were apparent. After adjustment for confounders, there was an increase in all-cause mortality in digoxin users (HR, 1.53; 95% CI, 1.34–1.73). This increase was most marked for cardiovascular deaths (adjusted HR, 2.73; 95% CI, 2.11–3.52), as expected, and there was only a small increase in the risk of death for noncardiovascular causes (adjusted HR, 1.26; 95% CI, 1.08–1.47; Table 2). A simplified analysis for colorectal cancer–specific mortality, based upon digoxin use in the year after diagnosis, also revealed little evidence of association (adjusted HR, 0.98; 95% CI, 0.79–1.22). A further sensitivity analyses revealed little evidence of association between colorectal cancer mortality and digoxin use in the year before diagnosis (adjusted HR, 0.88; 95% CI, 0.73–1.06).

We observed little evidence of increased colorectal cancer–specific mortality in digoxin users, providing some reassurance that digoxin is safe in colorectal cancer patients, despite recent evidence that digoxin users may have increased colorectal cancer risk (2). Our findings do not support a French study that observed reduced overall mortality with digoxin in 75 colorectal cancer patients (6), nor some preclinical studies suggesting that digoxin could have inhibitory effects on colorectal cancer cell growth (7).

This study is the first population-based cohort to investigate digoxin use and colorectal cancer–specific mortality. Other strengths include large size and long duration of follow-up, but we cannot rule out the possibility of type 2 error (a power calculation, using Schoenfeld method, based on observed medication use and cancer-specific deaths, indicated that we had approximately 80% power to detect as significant an HR of 1.25 for digoxin). Although verification of cancer diagnosis and death was robust, misclassification of colorectal cancer cause of death is possible; however, methodological studies suggest that comparative risk estimates are unlikely to be greatly affected where misclassification is unlikely to be differential. Recall bias was eliminated by using routinely collected GP-prescribed drug. Confounding by indication, often a problem in pharmacoepidemiology is unlikely to have influenced our main finding for colorectal cancer–specific mortality, but would explain the increase in all-cause mortality due largely to raised cardiovascular mortality in digoxin users (8). Misclassification of digoxin usage is possible because of noncompliance. As with all observational studies, confounding caused by unrecorded or incomplete potential confounders (e.g. stage) cannot be ruled out. In conclusion, there was little evidence of an increase in colorectal cancer–specific mortality with digoxin use after diagnosis.

No potential conflicts of interest were disclosed.

This study is based partly on data from the CPRD obtained under license from the UK Medicines and Healthcare products Regulatory Agency. However, the interpretation and conclusions contained in this study are those of the authors alone. The funders had no role in the study design; collection, analysis, and interpretation of data; writing of the report; or the decision to submit for publication.

Conception and design: L.J. Murray, C.M. Hughes, C.R. Cardwell

Development of methodology: L.J. Murray, C.R. Cardwell

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.J. Murray, C.R. Cardwell

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.A. Karasneh, L.J. Murray, C.R. Cardwell

Writing, review, and/or revision of the manuscript: R.A. Karasneh, L.J. Murray, C.M. Hughes, C.R. Cardwell

Study supervision: L.J. Murray, C.R. Cardwell

This work was supported by the Health and Social Care, Research and Development, Public Health Agency, Northern Ireland who funded a UK NIHR Career Development Fellowship (to C.R. Cardwell) and access to the CPRD dataset; and Yarmouk University, Jordon, who funded a PhD studentship (to R.A. Karasneh).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.