Proton pump inhibitors are not associated with an increased risk of colorectal cancer

The clinical relevance of proton pump inhibitors use as a risk factor for colorectal cancer in humans is still unclear.


| INTRODUC TI ON
Proton pump inhibitors (PPIs) are widely used in the management of acid peptic disorders and as preventive medication to avoid gastrointestinal bleeding associated with non-steroidal anti-inflammatory drug (NSAID) use. Because of their efficacy, the use of PPIs increased largely in the Netherlands in the last two decades and especially among older patients. 1 Between 2001 and 2015, the prevalence of PPI use increased from 6% to 15%.
Colorectal cancer is one of the most common types of cancer in the Netherlands. The survival rate of colorectal cancer improved in the last decade, but the incidence of colorectal cancer increased by 25% in the past 25 years. 2 This increased incidence may be related to an increasingly sedentary lifestyle, an unhealthy diet and increasing rates of obesity and diabetes. 3,4 Besides these risk factors, a concern has risen regarding the possible link between PPI-induced hypergastrinemia and gastrointestinal cancer, including colorectal cancer.
High gastrin levels are associated with the growth and proliferation of adenoma cells and might increase the risk of colorectal cancer. 5,6 The increased risk of colorectal cancer is mainly seen in experimental and animal model studies, 7,8 but whether an association between PPI use and risk of colorectal cancer exists in humans is less clear.
One study performed in the United States found a possible association between use of PPIs and colorectal cancer. 9 However, this study suffered from a small sample size. Two other studies performed in Taiwan also found an association between the use of PPI and the risk of colorectal cancer, but this pertained to recent use or short duration of PPI use. This seemed less likely to have causally contributed to colorectal cancer due to the long latency period of this disease. 10,11 Other studies on the use of PPI and risk of colorectal cancer found no or a marginal association. [11][12][13][14] The results of previous studies examining the association between the use of PPI and risk of colorectal cancer were based on small numbers of exposed patients, did not account for duration of use or the use of other medication such as NSAIDs, which is an indication for the prescription of PPIs and inversely associated with colorectal cancer. 15 Also only one study differentiated between the left and right hemicolon. 16 In the current study, a comprehensive large cohort with detailed data was used to investigate the risk of colorectal cancer with the use of PPIs taking into account timing and duration of PPI use and concomitant use of NSAIDs. In addition, a further analysis was performed determining whether the association varied across various subgroups with long-term use of PPIs.

| Setting
Data for this population-based case-control study were obtained from the Netherlands Cancer Registry (NCR) linked on a patientlevel to the General Practitioner (GP) Database of the PHARMO Database Network (the NCR-PHARMO GP cohort). This cohort covers a catchment area of approximately 4 million inhabitants (approximately 20%-25% of the Dutch population). The GP Database comprises data from electronic patient records registered by GPs.
The records include information on diagnoses and symptoms, laboratory test results, referrals to specialists and healthcare product/ drug prescriptions. The prescription records include information on type of product, prescription date, strength, dosage regimen, quantity and route of administration. Drug prescriptions are coded according to the WHO Anatomical Therapeutic Chemical (ATC) Classification System. 17 The NCR is a population-based registry which is maintained by the Comprehensive Cancer Centre the Netherlands (IKNL) and comprises information on newly diagnosed cancer patients in the Netherlands. The NCR is notified for new patients with cancer by pathology departments, general hospitals and radiotherapy institutes.
On a daily basis, trained data managers register data from hospital records within all Dutch hospitals using the NCR's registration and coding manual.
Further detailed information on the linkage and formation of NCR-PHARMO GP cohort can be found elsewhere. [18][19][20]

| Case population
From the NCR-PHARMO GP cohort, all subjects who were diagnosed between 2007 and 2014 with primary colorectal cancer (ICD 10-CM code C18-C20) were identified. The date of diagnosis was defined as the index date. In order to determine long-term use of PPI, patients without 6 years of exposure information available (defined as the time between date of registration at a GP practice and date of colorectal cancer diagnosis) were excluded. To reduce the potential for including patients with heritable colorectal cancer syndromes, patients younger than 40 years of age at diagnosis were excluded as well.

| Control population
Each colorectal cancer case was randomly matched to four controls based on gender, birth year, GP practice and period of available primary care data. Matched controls received the same index date as their matched colorectal cancer case. Controls were not allowed to have a diagnosis of cancer before index date (excluding basal cell carcinoma) and could not be matched more than once.

| Exposure definition
Data on PPI exposure was derived from the PHARMO GP Database, which contains detailed information on prescribed medication by GPs. The prescription records include detailed information on type of product, prescription date, strength, dosage regimen, quantity and route of administration. Drug prescriptions are coded according to the WHO Anatomical Therapeutic Chemical (ATC) Classification System. Drugs starting with the fifth-digit ATC code A02BC were classified as PPIs. In addition, PPI combination drugs were included as well (ATC code A02BD04 and M01AE52).
Duration of PPI use was calculated by constructing treatment episodes of PPIs based on quantity prescribed and prescribing information. Ever use of PPIs was defined as using a PPI in the 6 years before index date. Current use of PPIs was defined as using a PPI in the 90 days before index date. Recent use of PPIs was defined as using a PPI in the 91-365 days before index date, but not in the 90 days before index date. Past use of PPIs was defined as using a PPI more than 365 days before index date. Use of NSAIDs was determined in the same manner in order to assess concomitant use with PPIs.

| Confounding factors
The following covariates were evaluated as potential confounders: use of aspirins, nonsteroidal inflammatory drugs, histamine 2 receptor antagonists, statins and antidiabetics in the 6 years before index date. Age, sex and calendar time were inherently adjusted using the matching procedure.
Other risk factors, such as cigarette smoking history, body mass index, family history of cancer and helicobacter pylori status are important but unfortunately less uniformly reported by general practitioners.

| Statistical analyses
Conditional logistic regression was used to estimate the odds ratio (OR) and two-sided 95% confidence interval (CI) for colorectal cancer associated with the use of PPI, adjusted for preselected confounders.

| RE SULTS
A total of 6087 colorectal cancer cases could be matched to at least four cancer-free controls. Of those, 1987 colorectal cancer cases and 7912 cancer-free controls had at least 6 years of history available and were included in the analyses. The population of colorectal cancer cases with at least 6 years of history available did not differ from those with less than 6 years of history available in terms of age, gender and tumour characteristics(data not shown). The mean (±SD) age of colorectal cancer cases and cancer-free controls was 69.1 (±9.7) years and 58% was male (Table 1). About one-third (31%) of the colorectal cancer cases were diagnosed with stage III colorectal cancer. A tumour in the distal colon occurred in 34% of the colorectal cancer cases, 32% had a tumour in the proximal colon and 31% in the rectum.
There were 178 colorectal cancer cases (17%) and 824 cancer-free controls (20%) who had used PPIs for 4 years or longer. No increased risk in colorectal cancer was seen among those using PPIs for 4 years or longer (OR 0.92 (95% CI 0.76-1.11)). A significant increased risk was seen between current use of PPI in combination with NSAID (OR 1.40 (95% CI 1.15-1.71)) ( Table 3). Adjusting for potential confounders only marginally change this association (OR 1.57 (95% CI 1.27-1.93)). However, this association was attenuated for recent and past use of PPI in combination with NSAID.
The use of PPI for longer than 4 years and the risk of colorectal cancer was further explored by stratifying the risk by several subgroups ( Figure 1). Risk variations for long-term use were seen when stratifying by tumour stage; the OR for stage IV colorectal cancer was close to unity (OR 0.74 (95% CI 0.47-1.17) while the OR for stage I colorectal cancer was 1.28 (95% CI 0.87-1.87)) ( Figure 1).Also, the association differed when stratifying by colorectal subsite, including

| D ISCUSS I ON
In this large population-based case-control study, no increased risk of colorectal cancer was seen with ever use of PPIs. There was also no association seen when stratifying by duration of PPI use. When Strengths of the study include the large number of individuals exposed to treatment with PPIs. To our knowledge, this is the largest study to date on this topic. Furthermore, colorectal cancer cases were identified from the Netherlands Cancer Registry, in which data has shown to be of high quality due to thorough training of the registrars. 23 The completeness is estimated to be at least 95%.
Moreover, the use of the GP Database of the PHARMO Database Network ensured complete and high-quality assessment of prescription drug use. In the Netherlands, every inhabitant is registered with a GP which allows comprehensive follow-up of patients in the primary care setting. 24 This study also had some limitations. We lacked information on smoking status and body mass index, which are established risk factors for colorectal cancer. 25,26 Also, we used prescription data as a surrogate for actual use. This could lead to misclassification of some persons as users who were actually nonusers. However, it is not expected that this occur in long-term users as it is unlikely that there are patients who repeatedly fill prescriptions but not use them.
In summary, no overall increased risk of colorectal cancer was seen with the use of PPIs. Current use of PPIs was significantly associated with colorectal cancer, but this is probably the result of reverse causality. Also no increased risk was observed for long-term use of PPIs, but risk variations were seen by tumour stage and subsite. Further research needs to be conducted studying prolonged use of PPIs given the long latency period of cancer. Thereby, colorectal cancer located at different anatomical subsites should be investigated separately.

ACK N OWLED G EM ENTS
The authors thank all the healthcare providers contributing informa- interpretation of the study results, and critically reviewed and revised the article. All authors approved the final version of the article, including the authorship list.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1002/ygh2.409.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated during and/or analysed during the current study are not publicly available due to privacy reasons.