Cancer risk among insulin users: comparing analogues with human insulin in the CARING five-country cohort study

Aims/hypothesis The aim of this work was to investigate the relationship between use of certain insulins and risk for cancer, when addressing the limitations and biases involved in previous studies. Methods National Health Registries from Denmark (1996–2010), Finland (1996–2011), Norway (2005–2010) and Sweden (2007–2012) and the UK Clinical Practice Research Datalink database (1987–2013) were used to conduct a cohort study on new insulin users (N = 327,112). By using a common data model and semi-aggregate approach, we pooled individual-level records from five cohorts and applied Poisson regression models. For each of ten cancer sites studied, we estimated the rate ratios (RRs) by duration (≤0.5, 0.5–1, 1–2, 2–3, 3–4, 4–5, 5–6 and >6 years) of cumulative exposure to insulin glargine or insulin detemir relative to that of human insulin. Results A total of 21,390 cancer cases occurred during a mean follow-up of 4.6 years. No trend with cumulative treatment time for insulin glargine relative to human insulin was observed in risk for any of the ten studied cancer types. Of the 136 associations tested in the main analysis, only a few increased and decreased risks were found: among women, a higher risk was observed for colorectal (RR 1.54, 95% CI 1.06, 2.25) and endometrial cancer (RR 1.78, 95% CI 1.07, 2.94) for ≤0.5 years of treatment and for malignant melanoma for 2–3 years (RR 1.92, 95% CI 1.02, 3.61) and 4–5 years (RR 3.55, 95% CI 1.68, 7.47]); among men, a lower risk was observed for pancreatic cancer for 2–3 years (RR 0.34, 95% CI 0.17, 0.66) and for liver cancer for 3–4 years (RR 0.36, 95% CI 0.14, 0.94) and >6 years (RR 0.22, 95% CI 0.05, 0.92). Comparisons of insulin detemir with human insulin also showed no consistent differences. Conclusions/interpretation The present multi-country study found no evidence of consistent differences in risk for ten cancers for insulin glargine or insulin detemir use compared with human insulin, at follow-up exceeding 5 years. Electronic supplementary material The online version of this article (doi:10.1007/s00125-017-4312-5) contains peer-reviewed but unedited supplementary material, which is available to authorised users.


Detailed description of the calculation of cumulative treatment time
Below we describe a time-dependent approach that we used to construct the cumulative exposure as a cumulative time on treatment. Hereafter, by exposure to specific insulin we mean a cumulative time on this insulin treatment. Prescription data form the Nordic registries included the date and amount purchased in defined daily doses (DDDs) [29], but no information on individual dosage (http://www.whocc.no/ddd/definition_and_general_considera/). For the CPRD cohort, we derived DDDs from the dosage information (substance strength and amount) contained by prescription data.
Given a large effect of some individual's characteristics such as age, weight etc., on actual dosing, we assumed that cumulative time on particular insulin is more relevant exposure measure than actual DDDs per se. DDDs were used to calculate cumulative time under exposure, assuming a dose of 1 DDD per day. One DDD equals 40 IU for all insulins of interest (human insulin, insulin glargine and detemir), except human insulin for inhalation (A10AF01).
In exposure calculations we used the R language package Epi (Bendix Carstensen, Martyn Plummer, Esa Läärä, Michael Hills (2016). Epi: A Package for Statistical Analysis in Epidemiology. R package version 2.0. URL http://CRAN.R-project.org/package=Epi). More specifically, we used the gen.exp fuction, which is designed for construction of relevant covariates from purchase records. The information we used to construct the exposure variables included person identification numbers, date of each purchase, amount purchased, assumed daily dose, follow-up purchase (prescription) data. The basic ideas behind this function and illustrative examples are presented in the Tabulation outline (http://www.bendixcarstensen.com/DCRC/Tabulate/tabulate.pdf) by Bendix Carstensen.
For each insulin type of interest, the exposure was defined to begin on the date of first purchase (prescription), after which point an individual was considered exposed for the rest of his/her followup period. We generated exposure covariates for the insulin treatments of interest (human insulin, insulin glargine, insulin detemir) for the entire follow-up of each individual. We transformed each purchase (prescription) record into a period it covered according to the amount of DDDs, when assuming a constant dose rate of 1 DDD per day. For instance, 400 IUs of insulin glargine equals to ten DDDs and, thus, cover the period of ten days. Therefore, the period covered by this purchase (prescription) is a time from the date of purchase (prescription) to the date of purchase plus 10 days.
All purchases (prescriptions) were translated into exposure time in the same way for all persons.
To record any changes in one's medication state during the follow-up period, we split the individual follow-up time into 120-day intervals, and updated the exposure at the beginning of each interval. In the gen.exp function, use of information of dose rate was allowed by setting argument use.dpt=TRUE. In episodes of repeated prescriptions, this resulted in stacked exposure periods. By setting push.max=1 we allowed the start of exposure to particular purchase to be pushed into the future by maximum of 1 day. Time on a particular insulin accumulated until exposure stopped, and cumulative exposure remained unchanged unless individual resumed treatment. In episodes of repeated prescriptions, possible gaps between the periods not covered by adjacent purchases were ignored in the calculation of cumulative treatment time.
Resulting datasets (one for each cohort) consisted of records with the intervals of short length, at the start for which cumulative treatment time was calculated for each of three insulin types of interest.
Some of the individuals were treated with only one insulin type through the entire follow-up period, whereas others used different types of insulin. Thus, there were individuals who remained unexposed to all of the insulins of interest or to some of them, while for others the time at each of three insulin treatments of interest accumulated.
For instance, an individual with an entire follow-up length of four years started with human insulin and used it during the first three years of follow-up period so that purchased amount covered only 2.5 years. After two years he or she started detemir and during one year purchased amount corresponding to 1.08 years of exposure. Because the gaps of six months between the purchases of human insulin do not contribute to the cumulative treatment time, and because of quitting this treatment after three years, exposure to human insulin reached maximum of 2.5 years after three years and remained at this level until the end of follow-up. During the first two years of follow-up, individual remained unexposed to detemir and during the third year of follow-up the exposure exceeded one exposure year by one month so that this one month of the exposure was not accounted for at the end of followup. The exposure to human insulin remained zero throughout the entire follow-up.
Finally, three exposure variables (human insulin, insulin glargine, insulin detemir) were categorized into broader and finer exposure covariates by dividing cumulative treatment time into half-year categories for the first year, followed by one-year categories for longer exposure; the last categories were > 6 years for the broadly, and > 12 years for the finely categorized long-term exposure. In addition, each exposure variable incorporated a category to denote whether one stayed unexposed to the specific insulin represented by the variable.

ESM tables
ESM Table 1. Adjusted (all sites and any cancer: age, calendar time, NIADs, duration of insulin-treated diabetes, country; liver, colorectal, breast, and endometrial cancer: additional adjustment for relevant co-medications) rate ratios (RR) with 95% confidence interval (CI) for site-and sexspecific cancers: pairwise comparisons of glargine, detemir, and human insulin by the cumulative treatment time (years).

ESM Figures
ESM Figure 1. Adjusted (all sites and any cancer: age, calendar time, sex, NIADs, duration of insulin-treated diabetes, country; liver, colorectal, breast, and endometrial cancer: additional adjustment for relevant co-medications) rate ratio (RR) with 95% confidence interval (CI) for sitespecific cancers by the cumulative treatment time (years) on glargine versus human insulin.