Technological readiness and implementation of genomic-driven precision medicine for complex diseases.

The fields of human genetics and genomics have generated considerable knowledge about the mechanistic basis of many diseases. Genomic approaches to diagnosis, prognostication, prevention and treatment - genomic-driven precision medicine (GDPM) - may help optimize medical practice. Here, we provide a comprehensive review of GDPM of complex diseases across major medical specialties. We focus on technological readiness: how rapidly a test can be implemented into health care. Although these areas of medicine are diverse, key similarities exist across almost all areas. Many medical areas have, within their standards of care, at least one GDPM test for a genetic variant of strong effect that aids the identification/diagnosis of a more homogeneous subset within a larger disease group or identifies a subset with different therapeutic requirements. However, for almost all complex diseases, the majority of patients do not carry established single-gene mutations with large effects. Thus, research is underway that seeks to determine the polygenic basis of many complex diseases. Nevertheless, most complex diseases are caused by the interplay of genetic, behavioural and environmental risk factors, which will likely necessitate models for prediction and diagnosis that incorporate genetic and non-genetic data.

Using proximity extension proteomics assay to discover novel biomarkers associated with circulating leptin levels in patients with type 2 diabetes.

We aimed to discover novel associations between leptin and circulating proteins which could link leptin to the development of cardiovascular disease in patients with type 2 diabetes (T2DM). In a discovery phase, we investigated associations between 88 plasma proteins, assessed with a proximity extension assay, and plasma leptin in a cohort of middle-aged patients with T2DM. Associations passing the significance threshold of a False discovery rate of 5% (corresponding to p < 0.0017) were replicated in patients with T2DM in an independent cohort. We also investigated if proteins mediated the longitudinal association between plasma leptin and the incidence of major cardiovascular events (MACE). One protein, adipocyte fatty acid binding protein (A-FABP), was significantly associated with leptin in both the discovery phase [95% CI (0.06, 0.17) p = 0.00002] and the replication cohort [95% CI (0.12, 0.39) p = 0.0003]. Multiplicative interaction analyses in the two cohorts suggest a stronger association between A-FABP and leptin in men than in women. In longitudinal analyses, the association between leptin and MACE was slightly attenuated after adding A-FABP to the multivariate model. Our analysis identified a consistent association between leptin and A-FABP in two independent cohorts of patients with T2DM, particularly in men.Trial registration: ClinicalTrials.gov identifier NCT01049737.

Description and Characterization of a Novel Human Mast Cell Line for Scientific Study.

BACKGROUND: Laboratory of allergic diseases 2 (LAD2) human mast cells were developed over 15 years ago and have been distributed worldwide for studying mast cell proliferation, receptor expression, mediator release/inhibition, and signaling. LAD2 cells were derived from CD34+ cells following marrow aspiration of a patient with aggressive mastocytosis with no identified mutations in KIT. Another aspiration gave rise to a second cell line which has recently been re-established (LADR). We queried whether LADR had unique properties for the preclinical study of human mast cell biology. METHODS: LADR and LAD2 cells were cultured under identical conditions. Experiments examined proliferation, beta-hexosaminidase (ß-hex) release, surface receptor and granular protease expression, infectivity with HIV, and gene expression. RESULTS: LADR cells were larger and more granulated as seen with Wright-Giemsa staining and flow cytometry, with cell numbers doubling in 4 weeks, in contrast to LAD2 cells, which doubled every 2 weeks. Both LADR and LAD2 cells released granular contents following aggregation of FcεRI. LADR cells showed log-fold increases in FcεRI/CD117 and expressed CD13, CD33, CD34, CD63, CD117, CD123, CD133, CD184, CD193, and CD195, while LAD2 cells expressed CD33, CD34, CD63, CD117, CD133, CD193 but not CD13, CD123, CD184, or CD195. LADR tryptase expression was one-log-fold increased. LADR cell and LAD2 cell chymase expression were similar. Both cell lines could be infected with T-tropic, M-tropic, and dual tropic HIV. Following monomeric human IgE stimulation, LADR cells showed greater surface receptor and mRNA expression for CD184 and CD195. Expression arrays revealed differences in gene upregulation, especially for the suppressor of cytokine signaling (SOCS) family of genes with their role in JAK2/STAT3 signaling and cellular myelocytomatosis oncogene (c-MYC) in cell growth and regulation. CONCLUSIONS: LADR cells are thus unique in that they exhibit a slower proliferation rate, are more advanced in development, have increased FcεRI/CD117 and tryptase expression, have a different profile of gene expression, and show earlier infectivity with HIV-BAL, LAV, and TYBE when compared to LAD2 cells. This new cell line is thus a valuable addition to the few FcεRI+ human mast cell lines previously described and available for scientific inquiry.

Normal radiological lymph node appearance in the thorax.

Modern treatment of esophageal cancer is multimodal and highly dependent on a detailed diagnostic assessment of clinical stage, which includes nodal stage. Clinical appraisal of nodal stage is highly dependent on knowledge of normal radiological appearance, information of which is scarce. We aimed to describe lymph node appearance on computed tomography (CT) investigations in a randomly selected cohort of healthy subjects. In a sample of the Swedish Cardiopulmonary bioimage study, which investigates a sample of the Swedish population aged 50-64 years, the CT scans of 426 subjects were studied in detail concerning intrathoracic node stations relevant in clinical staging of esophageal cancer. With stratification for sex, the short axis of visible lymph nodes was measured and the distribution of lymph node sizes was calculated as well as proportion of patients with visible nodes above 5 and 10 millimeters for each station. Probability of having any lymph node station above 5 and 10 millimeters was calculated with a logistic regression model adjusted for age and sex. In the 214 men (aged: 57.3 ± 4.1 years) and 212 women (aged: 57.8 ± 4.4 years) included in this study, a total of 309 (72.5%) had a lymph node with a short axis of 5 mm or above was seen in at least one of the node stations investigated. When using 10 mm as a cutoff, nodes were visible in 29 (6.81%) of the subjects. Men had higher odds of having any lymph node with short axis 5 mm or above (OR 3.03 95% CI 1.89-4.85, P < 0.001) as well as 10 mm or above (OR 2.31 95% CI 1.02-5.23, P = 0.044) compared to women. Higher age was not associated with propensity for lymph nodes above 5 or 10 millimeters in this sample. We conclude that, in a randomly selected cohort of patients between 50 and 64 years, almost 10% of the men and 4% of the women had lymph nodes above 10 millimeters, most frequently in the subcarinal station (station 107). More than half of the patients had nodes above 5 millimeters on CT and men were much more prone to have this finding. The probability of finding lymph nodes in specific stations relevant of esophageal cancer is now described.

Systematic review with meta-analysis: the prevalence of coeliac disease in patients with osteoporosis.

BACKGROUND: Earlier studies have produced highly varying risk estimates for the prevalence of coeliac disease (CD) in osteoporosis. AIMS: To investigate the prevalence of CD among individuals with osteoporosis. METHODS: We conducted a systematic review of articles published in PubMed, Medline or EMBASE through May 2017 to identify studies looking at prevalence of CD in patients with osteoporosis. Search terms included "coeliac disease" combined with "fractures", "bone disease", "bone density", "densitometry", "osteoporos*", "osteomal*", "osteodys" or "dexa" or "dxa" or "skelet". Non-English papers with English-language abstracts were included. We used fixed-effects inverse variance-weighted models, and tested heterogeneity through subgroup analysis as well as through meta-regression. RESULTS: We identified eight relevant studies, comprising data from 3188 individuals with osteoporosis. Of these, 59 individuals (1.9%) had CD. A weighted pooled analysis demonstrated biopsy-confirmed CD in 1.6% (95% CI = 1.1%-2.0%) of individuals with osteoporosis. The heterogeneity was moderate (I2  = 40.1%), and influenced by the underlying CD prevalence in the general population. After adding four studies (n = 814) with CD defined as positive tissue transglutaminase or endomysial antibodies, the pooled prevalence was comparable (1.6%; 95% CI = 1.2%-2.0%). CONCLUSIONS: About 1 in 62 individuals with osteoporosis, or 1.6%, have biopsy-verified CD. This prevalence is comparable to that in the general population. These findings argue against routinely screening patients with osteoporosis for CD, which is contrary to current guideline recommendations. Additional studies are needed to determine the true utility of such screening programs.

Effectiveness of Drugs in Routine Care: A Model for Sequential Monitoring of New Medicines Using Dronedarone as Example.

Although there is no doubt about the scientific value of randomized controlled clinical trials, they are usually conducted in selected populations different from those treated in clinical practice. Therefore, it is important to optimize real-time postmarketing evaluation of the effectiveness, safety, and cost of new drugs. Using electronic health records and administrative health databases from a well-defined region with universal access to healthcare, we have built a framework for real-time sequential monitoring of the effectiveness of newly marketed drugs in routine care. We chose the antiarrhythmic agent dronedarone as the study drug and flecainide as the comparator drug for illustration of the model. We demonstrate that this model produces consistent results with increasing precision over time as data accumulates in the clinical systems. We believe that use of this model at the introduction of new drugs can provide complementary evidence, especially in settings of adaptive licensing of new drugs.

Targeted proteomic analysis of habitual coffee consumption.

BACKGROUND: Coffee drinking has been implicated in mortality and a variety of diseases but potential mechanisms underlying these associations are unclear. Large-scale systems epidemiological approaches may offer novel insights to mechanisms underlying associations of coffee with health. OBJECTIVE: We performed an analysis of known and novel protein markers linked to cardiovascular disease and their association with habitual coffee intake in the Prospective Study of the Vasculature in Uppsala Seniors (PIVUS, n = 816) and followed up top proteins in the Uppsala Longitudinal Study of Adult Men (ULSAM, n = 635) and EpiHealth (n = 2418). METHODS: In PIVUS and ULSAM, coffee intake was measured by 7-day dietary records whilst a computer-based food frequency questionnaire was used in EpiHealth. Levels of up to 80 proteins were assessed in plasma by a proximity extension assay. RESULTS: Four protein-coffee associations adjusted for age, sex, smoking and BMI, met statistical significance in PIVUS (FDR < 5%, P < 2.31 × 10-3 ): leptin (LEP), chitinase-3-like protein 1 (CHI3L), tumour necrosis factor (TNF) receptor 6 and TNF-related apoptosis-inducing ligand. The inverse association between coffee intake and LEP replicated in ULSAM (ß, -0.042 SD per cup of coffee, P = 0.028) and EpiHealth (ß, -0.025 SD per time of coffee, P = 0.004). The negative coffee-CHI3L association replicated in EpiHealth (ß, -0.07, P = 1.15 × 10-7 ), but not in ULSAM (ß, -0.034, P = 0.16). CONCLUSIONS: The current study supports an inverse association between coffee intake and plasma LEP and CHI3L1 levels. The coffee-CHI3L1 association is novel and warrants further investigation given links between CHI3L1 and health conditions that are also potentially influenced by coffee.

Body size and risk of atrial fibrillation: a cohort study of 1.1 million young men.

BACKGROUND: Whilst tall stature has been related to lower risk of vascular disease, it has been proposed as a risk factor for atrial fibrillation. Little is known about other anthropometric measures and their joint effects on risk of atrial fibrillation. OBJECTIVES: We aim to investigate associations and potential joint effects of height, weight, body surface area (BSA) and body mass index (BMI) with risk of atrial fibrillation. METHODS: In a cohort covering 1 153 151 18-year-old men participating in the Swedish military conscription (1972-1995), Cox regression was used to investigate associations of height, weight, BSA and BMI with risk of atrial fibrillation. RESULTS: During a median of 26.3 years of follow-up, higher height was associated with higher risk of atrial fibrillation (hazard ratio [HR] 2.80; 95% CI 2.63-2.98; for 5th vs. 1st quintile) and so was larger BSA (HR 3.05; 95% CI 2.82-3.28; for 5th vs. 1st quintile). Higher weight and BMI were to a lesser extent associated with risk of atrial fibrillation (BMI: 1.42; 95% CI 1.33-1.52, for 5th vs. 1st quintile). We found a multiplicative joint effect of height and weight. Adjusting for muscle strength, exercise capacity and diseases related to atrial fibrillation attenuated these measures. CONCLUSIONS: Higher height and weight are strongly associated with higher risk of atrial fibrillation. These associations are multiplicative and independent of each other and are summarized in a strong association of body surface area with risk of atrial fibrillation. The mechanisms remain unknown but may involve increased atrial volume load with larger body size.

Clinical Perspective on Antihypertensive Drug Treatment in Adults With Grade 1 Hypertension and Low-to-Moderate Cardiovascular Risk: An International Expert Consultation.

Hypertension is a leading risk factor for disease burden globally. An unresolved question is whether grade 1 hypertension (140-159/90-99mmHg) with low (cardiovascular mortality <1% at 10 years) to moderate (cardiovascular mortality ≥1% and <5% at 10 years) absolute total cardiovascular risk (CVR) should be treated with antihypertensive agents. A virtual international consultation process was undertaken to summarize the opinions of select experts. After holistic analysis of all epidemiological, clinical, psychosocial, and public health elements, this consultation process reached the following consensus in hypertensive adults aged <80 years: (1) The question of whether drug treatment in grade 1 should be preceded by a period of some weeks or months during which only lifestyle measures are recommended cannot be evidence based, but the consensus opinion is to have a period of lifestyle alone reserved only to patients with grade 1 "isolated" hypertension (grade 1 uncomplicated hypertension with low absolute total CVR, and without other major CVR factors and risk modifiers). (2) The initiation of antihypertensive drug therapy in grade 1 hypertension with moderate absolute total CVR should not be delayed. (3) Men ≥55 years and women ≥60 years with uncomplicated grade 1 hypertension should automatically be classified within the moderate absolute total CVR category, even in the absence of other major CVR factors and risk modifiers. (4) Statins should be considered along with blood-pressure lowering therapy, irrespective of cholesterol levels, in patients with grade 1 hypertensive with moderate CVR.

The Swedish CArdioPulmonary BioImage Study: objectives and design.

Cardiopulmonary diseases are major causes of death worldwide, but currently recommended strategies for diagnosis and prevention may be outdated because of recent changes in risk factor patterns. The Swedish CArdioPulmonarybioImage Study (SCAPIS) combines the use of new imaging technologies, advances in large-scale 'omics' and epidemiological analyses to extensively characterize a Swedish cohort of 30 000 men and women aged between 50 and 64 years. The information obtained will be used to improve risk prediction of cardiopulmonary diseases and optimize the ability to study disease mechanisms. A comprehensive pilot study in 1111 individuals, which was completed in 2012, demonstrated the feasibility and financial and ethical consequences of SCAPIS. Recruitment to the national, multicentre study has recently started.

Higher mortality after myocardial infarction in patients with severe mental illness: a nationwide cohort study.

OBJECTIVES: The aim of this study was to explore the impact of severe mental illness (SMI) on myocardial infarction survival and determine the influence of risk factor burden, myocardial infarction severity and different treatments. DESIGN, SETTING AND PARTICIPANTS: This population-based cohort study, conducted in Sweden during the period 1997-2010, included all patients with a first diagnosis of myocardial infarction in the Swedish nationwide myocardial infarction register SWEDEHEART (n = 209 592). Exposure was defined as a diagnosis of SMI (i.e. bipolar disorder or schizophrenia) in the national patient register prior to infarction. Bias-minimized logistic regression models were identified using directed acyclic graphs and included covariates age, gender, smoking, diabetes, previous cardiovascular disease, myocardial infarction characteristics and treatment. MAIN OUTCOME MEASURES: The outcomes were 30-day and 1-year mortality, obtained through linkage with national population registers. RESULTS: Patients with bipolar disorder (n = 442) and schizophrenia (n = 541) were younger (mean age 68 and 63 years, respectively) than those without SMI (n = 208 609; mean age 71 years). The overall 30-day and 1-year mortality rates were 10% and 18%, respectively. Compared with patients without SMI, patients with SMI had higher 30-day [odds ratio (OR) 1.99, 95% confidence interval (CI) 1.55-2.56] and 1-year mortality (OR 2.11, 95% CI 1.74-2.56) in the fully adjusted model. The highest mortality was observed amongst patients with schizophrenia (30-day mortality: OR 2.58, 95% CI 1.88-3.54; 1-year mortality: OR 2.55, 95% CI 1.98-3.29). CONCLUSION: SMI is associated with a markedly higher mortality after myocardial infarction, also after accounting for contributing factors. It is imperative to identify the reasons for this higher mortality.

Systematic review with meta-analysis: associations between coeliac disease and type 1 diabetes.

BACKGROUND: In the past decade, a number of population-based studies have examined the prevalence of coeliac disease in individuals with type 1 diabetes but prevalences have differed considerably. AIM: To examine the prevalence of coeliac disease in individuals with type 1 diabetes. METHODS: A systematic review of English-language articles published in PubMed Medline between 2000 and May 2014. Search terms included 'celiac disease' or 'coeliac disease' and 'diabetes mellitus'. Studies were selected with at least 100 individuals with type 1 diabetes being screened for coeliac disease where the coeliac diagnosis was later confirmed through small intestinal biopsy. Data synthesis used random-effects inverse variance-weighted models, and metaregression was used to examine heterogeneity in subgroups. RESULTS: A pooled analysis, based on 26,605 patients with type 1 diabetes, found a prevalence of biopsy-confirmed coeliac disease of 6.0% (95% CI = 5.0-6.9%). Heterogeneity was large (I(2) = 93.2%). The prevalence was lower in adults with type 1 diabetes (2.7%), and in mixed populations with both children and adults with type 1 diabetes (4.7%) than in children (6.2%) with type 1 diabetes (P < 0.001). Additional subgroup analyses could not explain the large variation in coeliac disease prevalence between studies. CONCLUSION: More than one in twenty patients with type 1 diabetes have biopsy-verified coeliac disease. This prevalence is high enough to motivate screening for coeliac disease among patients with type 1 diabetes.

Diabetes and CVD risk during angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker treatment in hypertension: a study of 15,990 patients.

Differences in clinical effectiveness between angiotensin-converting enzyme inhibitors (ACEis) and angiotensin receptor blockers (ARBs) in the primary treatment of hypertension are unknown. The aim of this retrospective cohort study was to assess the prevention of type 2 diabetes and cardiovascular disease (CVD) in patients treated with ARBs or ACEis. Patients initiated on enalapril or candesartan treatment in 71 Swedish primary care centers between 1999 and 2007 were included. Medical records data were extracted and linked with nationwide hospital discharge and cause of death registers. The 11,725 patients initiated on enalapril and 4265 on candesartan had similar baseline characteristics. During a mean follow-up of 1.84 years, 36,482 patient-years, the risk of new diabetes onset was lower in the candesartan group (hazard ratio (HR) 0.81, 95% confidence interval (CI) 0.69-0.96, P=0.01) compared with the enalapril group. No difference between the groups was observed in CVD risk (HR 0.99, 95% CI 0.87-1.13, P=0.86). More patients discontinued treatment in the enalapril group (38.1%) vs the candesartan group (27.2%). In a clinical setting, patients initiated on candesartan treatment had a lower risk of new-onset type 2 diabetes and lower rates of drug discontinuation compared with patients initiated on enalapril. No differences in CVD risk were observed.

Duodenal switch versus Roux-en-Y gastric bypass for morbid obesity: systematic review and meta-analysis of weight results, diabetes resolution and early complications in single-centre comparisons.

Long-term weight loss after Roux-en-Y gastric bypass (RYGB) in super-obese patients has not been ideal. Biliopancreatic diversion with duodenal switch (DS) is argued to be better; however, additional side effects are feared. The aim of the present study was to determine differences in results after DS and RYGB in publications from single-centre comparisons. A systematic review of studies containing DS and RYGB performed at the same centre was performed. Outcome data were weight results, resolution of comorbid conditions, perioperative results and complications. Main outcome was difference in weight loss after DS and RYGB. Secondary outcomes were difference in resolution of comorbidities, perioperative results and complications. The final analysis included 16 studies with in total 874 DS and 1,149 RYGB operations. When comparing weight results at the longest follow-up of each study, DS yielded 6.2 (95% confidence interval 5.0-7.5) body mass index units additional weight loss compared with RYGB, P < 0.001. Operative time and length of stay were significantly longer after DS, as well as the risk for post-operative leaks, P < 0.05. DS is more effective than RYGB as a weight-reducing procedure. However, this comes at the price of more early complications and might also yield slightly higher perioperative mortality.

Risk of symptomatic gallstones and cholecystectomy after a very-low-calorie diet or low-calorie diet in a commercial weight loss program: 1-year matched cohort study.

BACKGROUND: Concern exists regarding gallstones as an adverse event of very-low-calorie diets (VLCDs; <800 kcal per day). OBJECTIVE: To assess the risk of symptomatic gallstones requiring hospital care and/or cholecystectomy in a commercial weight loss program using VLCD or low-calorie diet (LCD). DESIGN: A 1-year matched cohort study of consecutively enrolled adults in a commercial weight loss program conducted at 28 Swedish centers between 2006 and 2009. A 3-month weight loss phase of VLCD (500 kcal per day) or LCD (1200-1500 kcal per day) was followed by a 9-month weight maintenance phase. Matching (1:1) was performed by age, sex, body mass index, waist circumference and gallstone history (n=3320:3320). Gallstone and cholecystectomy data were retrieved from the Swedish National Patient Register. RESULTS: One-year weight loss was greater in the VLCD than in the LCD group (-11.1 versus -8.1 kg; adjusted difference, -2.8 kg, 95% CI -3.1 to -2.4; P<0.001). During 6361 person-years, 48 and 14 gallstones requiring hospital care occurred in the VLCD and LCD groups, respectively, (152 versus 44/10 000 person-years; hazard ratio, 3.4, 95% CI 1.8-6.3; P<0.001; number-needed-to-harm, 92, 95% CI 63-168; P<0.001). Of the 62 gallstone events, 38 (61%) resulted in cholecystectomy (29 versus 9; hazard ratio, 3.2, 95% CI 1.5-6.8; P=0.003; number-needed-to-harm, 151, 95% CI 94-377; P<0.001). Adjusting for 3-month weight loss attenuated the hazard ratios, but the risk remained higher with VLCD than LCD for gallstones (2.5, 95% CI 1.3-5.1; P=0.009) and became borderline for cholecystectomy (2.2, 95% CI 0.9-5.2; P=0.08). CONCLUSION: The risk of symptomatic gallstones requiring hospitalization or cholecystectomy, albeit low, was 3-fold greater with VLCD than LCD during the 1-year commercial weight loss program.

Changes in body mass index following newly diagnosed type 2 diabetes and risk of cardiovascular mortality: a cohort study of 8486 primary-care patients.

AIMS: Elevated body mass index (BMI) is associated with an increased risk of type 2 diabetes and cardiovascular disease (CVD). This study explored the association between BMI changes in the first 18 months of newly diagnosed type 2 diabetes and the risk of long-term CVD mortality. METHODS: A total of 8486 patients with newly diagnosed type 2 diabetes and no previous history of CVD or cancer were identified from 84 primary-care centres in Sweden. During the first year after diagnosis, patients were grouped according to BMI change: 'Increase', or ≥+1 BMI unit; 'unchanged', or between +1 and-1 BMI unit; and 'decrease', or ≤-1 BMI unit. Associations between BMI change and CVD mortality, defined as death from stroke, myocardial infarction or sudden death, were estimated using adjusted Cox proportional hazards models (NCT 01121315). RESULTS: Baseline mean age was 60.0 years and mean BMI was 30.2kg/m(2). Patients were followed for up to 9 years (median: 4.6 years). During the first 18 months, 53.4% had no change in their BMI, while 32.2% decreased and 14.4% increased. Compared with patients with unchanged BMI, those with an increased BMI had higher risks of CVD mortality (hazard ratio: 1.63, 95% CI: 1.11-2.39) and all-cause mortality (1.33, 1.01-1.76). BMI decreases had no association with these risks compared with unchanged BMI: 1.06 (0.76-1.48) and 1.06 (0.85-1.33), respectively. CONCLUSION: Increased BMI within the first 18 months of type 2 diabetes diagnosis was associated with an increased long-term risk of CVD mortality. However, BMI decrease did not lower the long-term risk of mortality.

Low-dose acetylsalicylic acid and gastrointestinal ulcers or bleeding--a cohort study of the effects of proton pump inhibitor use patterns.

OBJECTIVE: The aim of this study was to investigate the associations between proton pump inhibitor (PPI) usage patterns and risk of severe gastrointestinal events in patients treated with low-dose acetylsalicylic acid (LDA). DESIGN AND SETTING: A nationwide cohort study in Sweden. PATIENTS: All Swedish residents ≥ 40 years of age, without cancer and receiving LDA treatment (≥ 80% adherence for 365 days between 2005 and 2009) were identified in the Swedish Prescription Register. Continuous PPI use was defined as > 60 of 90 days covered by daily PPI doses and further divided into high (≥ 80%) or moderate (< 80) adherence. All other PPI use was defined as intermittent use. MAIN OUTCOME MEASURES: The risk of a combined end-point of gastrointestinal ulcer or bleeding was analysed using Cox proportional hazard models. We also investigated risk of > 45 days of LDA treatment interruption. RESULTS: During a median follow-up of 2.5 years, 7880 of 648,807 (1.2%) LDA-treated patients experienced gastrointestinal events. In multivariable-adjusted models, both intermittent-PPI and no-PPI use were associated with increased risk of gastrointestinal ulcers or bleeding compared with continuous PPI use with a high level of adherence [hazard ratio (HR) 1.83 (95% CI 1.66-2.02) and 1.14 (95% CI 1.05-1.23), respectively]. Amongst continuous PPI users, moderate adherence also increased the risk of gastrointestinal ulcers or bleeding [HR 1.22 (95% CI 1.07-1.40)]. The risk of LDA treatment interruption was higher with intermittent PPI use [HR 1.16 (95% CI 1.14-1.19)] than continuous PPI use with high adherence. CONCLUSIONS: In this large cohort of LDA users, intermittent PPI use was associated with higher risk of gastrointestinal ulcers or bleeding and interrupted LDA treatment, compared with continuous PPI use.