BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Exposure to sodium channel-inhibiting drugs and cancer survival: protocol for a cohort study using the QResearch primary care database

For peer review only Journal: BMJ Open

Manuscript ID: bmjopen-2014-006604

Article Type: Protocol

Date Submitted by the Author: 11-Sep-2014

Complete List of Authors: Fairhurst, Caroline; University of York, Health Sciences Watt, Ian; University of York, Health Sciences Martin, Fabiola; Hull York Medical School, Bland, Martin; University of York, Health Sciences Brackenbury, William; University of York, Biology

Primary Subject Oncology Heading:

Secondary Subject Heading: Epidemiology, Pharmacology and therapeutics

PRIMARY CARE, Epilepsy < NEUROLOGY, Breast tumours < ONCOLOGY, Keywords: Epidemiology < ONCOLOGY, STATISTICS & RESEARCH METHODS, Clinical trials < THERAPEUTICS http://bmjopen.bmj.com/

on September 29, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Exposure to sodium channel-inhibiting drugs and cancer survival: 4 5 protocol for a cohort study using the QResearch primary care database 6 7 8 9 Caroline Fairhurst1, Ian Watt1,2, Fabiola Martin2,3, Martin Bland1, and William 10 11 J. Brackenbury3* 12 13 1Department of Health Sciences, University of York, York, UK 14 15 2HullFor York Medical peer School, York, review UK only 16 17 3Department of Biology, University of York, York, UK 18 19 20 21 *Corresponding author: 22 23 William J. Brackenbury 24 25 Department of Biology, University of York, Heslington, York, YO10 5DD, UK 26 27 E-mail: [email protected] 28 29 Telephone: +44 (0) 1904 328 284 30 31

32 33 Keywords: Anticonvulsants / Breast neoplasms / Colonic neoplasms / Prostatic

34 http://bmjopen.bmj.com/ 35 36 neoplasms / Sodium channels 37 38 39 40 Word count, excluding title page, abstract, references, figures and tables: 2327. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 1 BMJ Open Page 2 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 ABSTRACT 4 5 6 7 Introduction: Metastasis from solid tumours is associated with significant 8 9 morbidity and mortality, and is the leading cause of cancer-related deaths. Voltage- 10 11 gated sodium channels (VGSCs) are drug targets for the treatment of epilepsy. 12 13 VGSCs are also present in cancer cells, where they regulate metastatic cell 14 15 behaviours,For including peer cellular movement review and invasion. Treating only cancer cells with the 16 17 VGSC-inhibiting anticonvulsant phenytoin reduces cellular invasion and migration. 18 19 Together, these suggest that VGSCs may be useful targets for inhibiting metastasis. 20 21 The purpose of this study is to test the hypothesis that use of VGSC-inhibiting drugs 22 23 will reduce metastasis, and therefore increase survival time in cancer patients. 24 25 26 27 Methods and analysis: A cohort study based on primary care data from the 28 29 QResearch database will include patients with one of three common tumours: breast, 30 31 bowel, and prostate. The primary outcome will be overall survival from date of cancer 32 33 diagnosis. Cox proportional hazards regression will be used to compare the survival

34 http://bmjopen.bmj.com/ 35 36 of cancer patients taking VGSC-inhibiting drugs (including anticonvulsants and Class 37 38 I antiarrhythmic agents) with cancer patients not exposed to these drugs, adjusting 39 40 for age and sex. Exposure to VGSC-inhibiting drugs will be defined as having at 41 42 least one prescription for these drugs prior to cancer diagnosis. High and low on September 29, 2021 by guest. Protected copyright. 43 44 exposure groups will be identified based on length of use. A number of sensitivity and 45 46 secondary analyses will be conducted. 47 48 49 50 Ethics and dissemination: The protocol has been independently peer- 51 52 reviewed and approved by the QResearch Scientific Board. The project has also 53 54 been approved by the University of York Ethical Review Process. The results will be 55 56 presented at international conferences and published in an open access peer- 57 58 reviewed journal, in accordance with the STROBE criteria. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 2 Page 3 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 STRENGTHS AND LIMITATIONS OF THIS STUDY 6 7 8 9 • Primary care research data 10 11 • Large sample size and statistical power 12 13 • Planned sensitivity analyses 14 15 • Prescription-basedFor peer study review only 16 17 • No direct information on metastasis, estimation is via overall survival 18 19 20 • Some variables of interest may be missing and/or poor quality in GP data 21 22 23 24 INTRODUCTION 25 26 27 28 Bowel, breast and colon cancer are common cancers, which if diagnosed late have 29 30 often already spread to secondary sites (metastasised). Metastasis is associated with 31 32 significant morbidity and mortality. Metastasis is the leading cause of cancer-related 33

34 deaths[1] because a metastatic cancer is rarely amenable to cure, and interventions http://bmjopen.bmj.com/ 35 36 are largely limited to palliation.[2] Therefore, there is an urgent need to identify and/or 37 38 develop new metastasis prevention strategies. 39 40 41 42

The classical role of voltage-gated sodium channels (VGSCs) is to transmit action on September 29, 2021 by guest. Protected copyright. 43 44 potentials in electrically excitable cells, e.g. neurons and cardiomyocytes.[3] VGSCs 45 46 also regulate neuronal growth and migration.[4-7] Related to these functions, VGSCs 47 48 are clinical targets for a range of disorders, including epilepsy, cardiac arrhythmias, 49 50 neuropathic pain and depression.[8] The mode of action of a number of commonly 51 52 prescribed antiepileptic drugs (anticonvulsants), including phenytoin, lamotrigine, 53 54 carbamazepine and valproate, is to inhibit VGSCs.[9] Similarly, the principal mode of 55 56 action of Class I antiarrhythmic drugs is to inhibit VGSCs.[10] 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 3 BMJ Open Page 4 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 More recently, VGSCs have been identified in cells from a number of major cancers, 6 7 including carcinomas of the breast, prostate and colon.[11 12] In these cells, VGSCs 8 9 promote in vitro cellular behaviours that are associated with metastasis, including 10 11 migration and invasion.[13-18] Overexpression of the VGSC β1 subunit in breast 12 13 cancer cells increases metastasis in mice.[19-21] The VGSC-inhibiting 14 15 anticonvulsantFor phenytoin peer significantly review reduces migration and only invasion of metastatic 16 17 breast and prostate cancer cells in vitro[22 23] and has been shown to reduce 18 19 metastasis in a mouse model of breast cancer (M. Nelson, M. Yang, A. A. Dowle, J. 20 21 R. Thomas and W. J. Brackenbury, manuscript under review, 2014). Together, these 22 23 data suggest that VGSCs may be useful targets for anti-metastatic therapy, and that 24 25 VGSC-inhibiting drugs may improve survival from certain cancers.[11 24] Although 26 27 28 the effect of several anticonvulsants on risk of developing various cancers has been 29 30 studied before (reviewed in[25]), the relationship between VGSC-inhibiting drugs and 31 32 survival of cancer patients has not been investigated. 33

34 http://bmjopen.bmj.com/ 35 36 The purpose of this study is to test the hypothesis that use of VGSC-inhibiting drugs 37 38 will predict increased time to metastasis and thus improved survival time in cancer 39 40 patients. The objectives are to investigate: 41 42 • The relationship between use of all VGSC-inhibiting (anticonvulsant and on September 29, 2021 by guest. Protected copyright. 43 44 Class I antiarrhythmic) drugs and overall survival of cancer patients. We will 45 46 focus on carcinomas of the breast, colon and prostate because they are the 47 48 most common and VGSC expression has been extensively studied in these 49 50 tumours.[11 13-17 26-30] 51 52 • The relationship between use of all VGSC-inhibiting drugs and cancer specific 53 54 survival. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 4 Page 5 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 • The relationship between individual VGSC-inhibiting drugs and overall 4 5 survival. 6 7 8 9 There are no systematic reviews exploring this area and we are addressing this gap 10 11 by conducting a review concurrent to this study [PROSPERO registration number 12 13 CRD42014013574]. 14 15 For peer review only 16 17 METHODS AND ANALYSIS 18 19 20 21 Data source and sample selection 22 23 24 25 This study will use general practice data accessed from QResearch 26 27 (http://www.qresearch.org), a large consolidated database derived from the 28 29 anonymised health records of over 13 million patients from 753 general practices 30 31 32 (representing around 7% of UK practices). QResearch data are collected from the 33 EMIS GP computer system and have been validated using other sources and shown

34 http://bmjopen.bmj.com/ 35 36 to yield similar results to other databases, e.g. the Clinical Practice Research 37 38 Datalink (CPRD).[31 32] QResearch has been used previously to study associations 39 40 between cancer and prescription information.[31] 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 An open cohort of 100,000 patients (aged 30 years or older) with a diagnosis of 45 46 breast, colorectal or prostate cancer will be identified who were registered with a 47 st st 48 QResearch practice during the study period between 1 January 1998 and 31 49 50 December 2013. This will include all those cancer patients in the database who have 51 52 a prescription of one of the index drugs recorded before their date of cancer 53 54 diagnosis (Table 1).[33] The remaining patients will be randomly selected controls. 55 56 Time from date of diagnosis to death will be investigated and data will be right- 57 58 censored in patients who are still alive at the end of the study period. Cancer 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 5 BMJ Open Page 6 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 diagnoses will be based on Read code information (see online supplementary 4 5 appendix). 6 7 8 9 Table 1. Voltage-gated Na+ channel-inhibiting drugs 10 11 Drug/derivative Alternative names Classification British National 12 13 Formulary (BNF) 14 15 For peer review onlysection[33] 16 17 Carbamazapine, Arbil, Carbagen Anticonvulsant 4.2.3, 4.7.3, 4.8.1 18 19 eslicarbazepine, SR, Epimaz, 20 21 oxcarbazepine Inovelon, Tegretol, 22 23 Teril, Timonil, 24 25 Trileptal, Zebinix 26 27 28 Disopyramide Dirythmin, Isomide, Class Ia 2.3.2 29 30 Rythmodan antiarrhythmic 31 32 Flecainide Tambocor Class Ic 2.3.2 33 antiarrhythmic 34 http://bmjopen.bmj.com/ 35 36 Lacosamide Vimpat Anticonvulsant 4.8.1 37 38 Lamotrigine Lamictal Anticonvulsant 4.8.1 39 40 Lidocaine Lignocaine, Class Ib 2.3.2, 15.2 41 42 Xylocard antiarrhythmic on September 29, 2021 by guest. Protected copyright. 43 44 Mexiletine Mexitil Class Ib 2.3.2 45 46 antiarrhythmic 47 48 Moracizine Ethmozine Class Ic - 49 50 antiarrhythmic 51 52 Phenytoin, Epanutin, Pentran Anticonvulsant, 4.7.3, 4.8.1, 4.8.2 53 54 fosphenytoin Class Ib 55 56 antiarrhythmic 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 6 Page 7 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Procainamide Pronestyl Class Ia 2.3.2 4 5 antiarrhythmic 6 7 Propafenone Arythmol Class Ic 2.3.2 8 9 antiarrhythmic 10 11 Quinidine Kiditard Class Ia - 12 13 antiarrhythmic 14 15 RanolazineFor peer Ranexa reviewAntianginal only 2.6.3 16 17 Riluzole Rilutek Treatment for 4.9.3 18 19 amyotrophic lateral 20 21 sclerosis 22 23 24 Tocainide Tonocard Class Ib - 25 26 antiarrhythmic 27 28 Topiramate Topamax Anticonvulsant 4.7.4, 4.8.1 29 30 Valproic acid, Convulex, Anticonvulsant 4.2.3, 4.7.4, 4.8.1 31 32 sodium valproate Depakote, Epilim, 33

34 Epival, Episenta, http://bmjopen.bmj.com/ 35 36 Orlept 37 38 39 40 Exclusions 41 42

on September 29, 2021 by guest. Protected copyright. 43 44 Temporary residents and patients registered with QResearch within 12 months of 45 46 data extraction will be excluded. Cases without diagnosis of one of the three index 47 48 cancers (breast, colorectal, or prostate cancer) will be excluded. Patients with 49 50 anomalous, incorrect or infeasible dates will be excluded, e.g., dates of cancer 51 52 diagnoses recorded before birth or after death. We shall assume that dates of birth 53 54 and death are correct. Any patient with a date of diagnosis that indicates they were 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 7 BMJ Open Page 8 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 younger than 25 at the time of diagnosis will be excluded as it is unlikely a person of 4 5 that age would get one of these three cancers. 6 7 8 9 Exposure 10 11 12 13 A participant will be considered as exposed if they have had at least one prescription 14 15 for one ofFor the index drugs.peer Assuming review continuous treatment onlyuse between 16 17 prescriptions, we will identify two exposure groups: a low exposure group (less than 6 18 19 months worth of prescriptions) and a moderate to high exposure group (six months or 20 21 more prescriptions). The exposed groups, separately and in combination, will be 22 23 compared with the control group (cases without any prescription for one of the index 24 25 drugs). Patients with one prescription for a drug that would have been used as a local 26 27 anaesthetic, e.g. lidocaine, will be excluded. 28 29

30 31 Outcome measures 32 33

34 http://bmjopen.bmj.com/ 35 36 Metastasis is estimated to be responsible for 90% of deaths from solid tumours.[34] 37 38 However, metastasis itself is not reliably recorded in GP data and so the primary 39 40 outcome measure will be overall survival following cancer diagnosis as a proxy for 41 42 metastasis. Secondary outcome measures will be cancer-specific survival for each on September 29, 2021 by guest. Protected copyright. 43 44 index type of cancer and overall survival across each drug, numbers permitting. 45 46 47 48 Confounding factors 49 50 51 52 Data on the following confounders will be requested: age, gender, alcohol 53 54 consumption, smoking status, body mass index (BMI) and ethnicity. Data on alcohol, 55 56 smoking and BMI are routinely collected and as such a single patient may have 57 58 multiple recorded observations for these variables assessed over time. We will 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 8 Page 9 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 consider the observations measured at the closest date before the date of cancer 4 5 diagnosis, based on appropriate Read codes (detailed in online supplementary 6 7 appendix). The patients will be categorised as follows: 8 9 • Alcohol consumption[35] categorised as: Non/trivial drinker (<1unit/day), Light 10 11 drinker (1-2units/day), Moderate-very heavy drinker(3+units/day) and Not 12 13 recorded/known. 14 15 • SmokingFor status[36] peer categorised review as: Ex-smoker, Smoker, only Non-smoker, and 16 17 Not recorded/known. 18 19 • BMI,[37] categorised as Underweight (<18.5), Normal range [18.5-25), 20 21 Overweight [25, 30), Obese (30+) and Not recorded/known. 22 23 24 • Ethnicity[38] categorised according to the groupings used in the 2011 UK 25 26 census: White; Mixed/Multiple ethnic groups; Asian/Asian British; 27 28 Black/African/Caribbean/Black British; Other Ethnic group. We shall also 29 30 include a ‘Not recorded/known’ category. 31 32 33 Sample size calculation 34 http://bmjopen.bmj.com/ 35 36 37 38 Up to 100,000 eligible cases will be used, which is the maximum sample size that will 39 40 be released by QResearch. At breast cancer diagnosis, approximately 6% of patients 41 42 present with metastatic lesions, with bone being the most common site.[39] Of on September 29, 2021 by guest. Protected copyright. 43 44 patients presenting without bone metastasis at diagnosis, 3.6% subsequently 45 46 develop metastases.[40 41] The majority (90%) of metastases will lead to death.[34] 47 48 Pharmacological blockade of VGSCs inhibits invasion of breast, colorectal and 49 50 prostate cancer cells in vitro by 25-50%.[13 15 22 23] Therefore, assuming 3.6% of 51 52 cancer diagnoses lead to a metastasis and most of these to death, with standard 53 54 significance level alpha = 5% and power = 90%, we would require 4248 in the 55 56 exposed group to detect a fall of 25% in the metastasis (or death) rate and 928 to 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 9 BMJ Open Page 10 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 detect a fall of 50%. This is based on 20 comparison patients per exposed patient, 4 5 but this ratio is not critical. If we include 6% with a metastasis present at initial 6 7 diagnosis, these numbers fall to 1503 and 330. 8 9 10 11 The prevalence of epilepsy is estimated to be 1%.[42] Together, the most commonly 12 13 used VGSC-inhibiting anticonvulsants, phenytoin, lamotrigine, carbamazepine and 14 15 valproate,For account forpeer >82% of all antiepileptic review drug use.[43] only By contrast, Class I 16 17 antiarrhythmic drug use has been considerably less common: <5% in patients with 18 19 cardiac arrhythmia.[44] Thus, using these data as a guide, we might reasonably 20 21 anticipate that around 0.8% of cancer patients would be using one of these VGSC- 22 23 inhibiting drugs. To meet our largest target sample size, 4248, we would therefore be 24 25 looking for a sample that contained 530,000 people with a diagnosis of one of the 26 27 target cancers. To meet the lower target of 928, we would require 116,000 28 29 diagnoses. Given that we are studying deaths rather than metastases per se, we will 30 31 be unable to distinguish between metastases present at diagnosis and detected 32 33 subsequently. Therefore, if we include 6% assumed to have a metastasis present at

34 http://bmjopen.bmj.com/ 35 36 initial diagnosis, we would require 187,875 and 41,250 diagnoses to detect falls in 37 38 metastasis of 25% and 50% respectively. 39 40 41 42 According to Cancer Research UK,[45] the lifetime risk in 2010 for the four major on September 29, 2021 by guest. Protected copyright. 43 44 cancer sites was almost 13% (female breast), 6% (female lung), 8% (male lung), 6% 45 46 (female bowel including anus), 7% (male bowel including anus) and 13% (prostate). 47 48 Hence for our chosen sites, we expect approximately 21% of women and 20% of 49 50 men to experience a positive diagnosis at some time. We will not have lifetime data 51 52 for many in the database, but we might anticipate that 10% of a database sample 53 54 would have a history of one of these sites. Thus the Qresearch database of 13 55 56 million people is large enough to achieve our largest sample target. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 10 Page 11 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Statistical analysis 4 5 6 7 Analysis will be conducted in Stata v13, using two-sided significance at the 5% level. 8 9 For each Cox model, only the patients with complete data for each of the covariates 10 11 controlled for in the model will be included in the analysis. 12 13 14 15 DescriptiveFor summaries peer review only 16 17 18 19 The characteristics of the comparison groups will be described using summary 20 21 statistics. Categorical data will be presented as frequency and percentage, and 22 23 continuous variables will be summarised using descriptive statistics (mean, standard 24 25 deviation, median, 1st and 3rd quartiles, minimum and maximum). The flow of 26 27 patients in the QResearch database will be presented in a diagram. 28 29

30 31 Primary analysis 32 33

34 http://bmjopen.bmj.com/ 35 36 The primary analysis will compare the combined exposure group to the control group. 37 38 For each group, the distribution of time from diagnosis of cancer to death will be 39 40 described using Kaplan-Meier survival estimates. Kaplan-Meier survival curves will 41 42 be presented for the two groups. The statistical equivalence of the two curves will be on September 29, 2021 by guest. Protected copyright. 43 44 tested using the log-rank test. Right censoring will occur if the patient is still alive at 45 st 46 the end of the study period (31 December 2013). Median time to death, with a 95% 47 48 confidence interval (CI) will be presented. If the estimated survivor function is greater 49 50 than 0.5 throughout the study it will not be possible to estimate the median 51 52 survival time and other percentiles survival values (i.e. 90%, 80%, 75%, as 53 54 appropriate) will be presented. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 11 BMJ Open Page 12 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 We will compare the survival of exposed cases with control cases from time of 4 5 diagnosis of one of the three index cancers using a Cox Proportional Hazards 6 7 regression model. The endpoint will be all cause mortality. We will adjust the Cox 8 9 model for type of cancer (breast, bowel or prostate), gender and age at diagnosis. 10 11 Age will be included with both a linear and quadratic term (age + age2). We will 12 13 assume that all included patients are receiving the most appropriate standard 14 15 treatmentFor for their disease, peer so we will review not adjust for cancer-treating only drug intake. 16 17 Hazard ratios will be presented with p-values and 95% confidence intervals. 18 19 20 21 Cox regression assumes that the proportional hazards model applies. To assess 22 23 this, we shall plot –log(-log(S(t))) against log(time), where S(t) is the survivor function 24 25 at time t. The curves for the two groups should be parallel. We will also consider a 26 27 chi-squared test of the Schoenfeld residuals to assess the null hypothesis of no 28 29 relationship between the hazards in each group. If the assumptions are not met, we 30 31 shall try to investigate why this is. 32 33

34 http://bmjopen.bmj.com/ 35 36 Sensitivity analysis 37 38 39 40 We will repeat the primary analysis, but adjust the Cox model, in turn, for 41 42 confounding variables: ethnicity, body mass index, smoking and alcohol on September 29, 2021 by guest. Protected copyright. 43 44 consumption. 45 46 47 48 Secondary analyses 49 50 51 52 Each of the following secondary end points will be analysed like the primary outcome 53 54 (unless indicated) with identical censoring strategy: 55 56 • If cancer type proves to be a significant predictor in the primary model then 57 58 we will consider cancer specific survival 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 12 Page 13 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 • Survival of Low exposure group compared with control group 4 5 • Survival of High exposure group compared with control group 6 7 • Survival of combined exposure group and control group with outcome of time 8 9 to death from first diagnosis of any cancer, since some patients may have a 10 11 diagnosis of another cancer before one of breast, bowel or prostate (a 12 13 category for ‘Other’ will be included in the covariate for type of cancer) 14 15 • SurvivalFor of patients peer dependent review on the main drug classonly that they are exposed 16 17 to (numbers permitting). 18 19 20 21 22 ETHICS AND DISSEMINATION 23 24 25 26 This protocol has been independently peer-reviewed by the QResearch Scientific 27 28 Board. It has also been approved by the University of York Ethical Review Process. 29 30 Only the authors will have access to the data during the study, in order to guarantee 31 32 confidentiality of patient information. An article detailing the results of the study will 33

34 be submitted for publication in an international peer-reviewed journal, in accordance http://bmjopen.bmj.com/ 35 36 with the Strengthening the Reporting of Observational Studies in Epidemiology 37 38 (STROBE) criteria.[46] The full protocol and statistical analysis will be available from 39 40 the authors after publication of the results. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 AUTHORS’ CONTRIBUTIONS 45 46 47 48 WB had the original idea for this study. CF and WB wrote the draft of the manuscript. 49 50 IW, FM and MB contributed to the development of the idea, the study design, and 51 52 revised the manuscript. All authors approved the final submitted version of the 53 54 manuscript. 55 56

57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 13 BMJ Open Page 14 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 FUNDING STATEMENT 4 5 6 7 This work was supported by the Medical Research Council [Fellowship 8 9 G1000508] and the Wellcome Trust [ref: 097829] through the Centre for Chronic 10 11 Diseases and Disorders (C2D2) at the University of York. 12 13 14 15 COMPETINGFor peer INTERESTS STATEMENTreview only 16 17 18 19 The authors declare that they have no competing interests. 20 21 22 23 REFERENCES 24 25 26 27 1. Rugo HS. The importance of distant metastases in hormone-sensitive breast 28 cancer. Breast 2008;17 Suppl 1:S3-8 doi: 10.1016/S0960-9776(08)70002- 29 Xpublished Online First: Epub Date]|. 30 2. Suva LJ, Griffin RJ, Makhoul I. Mechanisms of bone metastases of breast cancer. 31 Endocr Relat Cancer 2009;16(3):703-13 doi: 10.1677/ERC-09-0012published 32 Online First: Epub Date]|. 33 3. Hille B. Ionic channels of excitable membranes. 2nd ed. Sunderland

34 http://bmjopen.bmj.com/ 35 (Massachusetts): Sinauer Associates Inc., 1992. 36 4. Brackenbury WJ, Calhoun JD, Chen C, et al. Functional reciprocity between Na+ 37 channel Nav1.6 and β1 subunits in the coordinated regulation of excitability 38 and neurite outgrowth. Proc Natl Acad Sci U S A 2010;107(5):2283-88 39 5. Brackenbury WJ, Yuan Y, O'Malley HA, Parent JM, Isom LL. Abnormal neuronal 40 patterning occurs during early postnatal brain development of Scn1b-null 41 mice and precedes hyperexcitability. Proc Natl Acad Sci U S A 42 2013;110(3):1089-94 doi: 10.1073/pnas.1208767110published Online First: on September 29, 2021 by guest. Protected copyright. 43 Epub Date]|. 44 6. Catterall WA. From ionic currents to molecular mechanisms: the structure and 45 function of voltage-gated sodium channels. Neuron 2000;26(1):13-25 + 46 7. Brackenbury WJ, Davis TH, Chen C, et al. Voltage-gated Na channel β1 subunit- 47 mediated neurite outgrowth requires fyn kinase and contributes to central 48 nervous system development in vivo. J Neurosci 2008;28(12):3246-56 49 8. Clare JJ, Tate SN, Nobbs M, Romanos MA. Voltage-gated sodium channels as 50 therapeutic targets. Drug Discov Today 2000;5(11):506-20 51 9. Mantegazza M, Curia G, Biagini G, Ragsdale DS, Avoli M. Voltage-gated sodium 52 channels as therapeutic targets in epilepsy and other neurological disorders. 53 Lancet Neurol 2010;9(4):413-24 doi: 10.1016/S1474-4422(10)70059- 54 4published Online First: Epub Date]|. 55 10. Vaughan Williams EM. The relevance of cellular to clinical electrophysiology in 56 classifying antiarrhythmic actions. J Cardiovasc Pharmacol 1992;20 Suppl 57 2:S1-7 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 14 Page 15 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 11. Brackenbury WJ. Voltage-gated sodium channels and metastatic disease. 4 Channels (Austin) 2012;6(5):352-61 doi: 10.4161/chan.21910published 5 Online First: Epub Date]|. 6 12. Brackenbury WJ, Djamgoz MB, Isom LL. An emerging role for voltage-gated Na+ 7 channels in cellular migration: regulation of central nervous system 8 development and potentiation of invasive cancers. Neuroscientist 9 2008;14(6):571-83 doi: 10.1177/1073858408320293published Online First: 10 Epub Date]|. 11 13. House CD, Vaske CJ, Schwartz A, et al. Voltage-gated Na+ channel SCN5A is a 12 key regulator of a gene transcriptional network that controls colon cancer 13 invasion. Cancer Res 2010;70(17):6957-67 doi: 10.1158/0008-5472.CAN-10- 14 1169published Online First: Epub Date]|. 15 14. GrimesFor JA, Fraser peer SP, Stephens review GJ, et al. Differential expressiononly of voltage- 16 activated Na+ currents in two prostatic tumour cell lines: contribution to 17 invasiveness in vitro. FEBS Lett 1995;369(2-3):290-4 18 15. Fraser SP, Diss JK, Chioni AM, et al. Voltage-gated sodium channel expression 19 and potentiation of human breast cancer metastasis. Clin Cancer Res 20 2005;11(15):5381-89 doi: 10.1158/1078-0432.CCR-05-0327published Online 21 First: Epub Date]|. 22 16. Brackenbury WJ, Chioni AM, Diss JK, Djamgoz MB. The neonatal splice variant 23 of Nav1.5 potentiates in vitro metastatic behaviour of MDA-MB-231 human 24 breast cancer cells. Breast Cancer Res. Treat. 2007;101(2):149-60 25 17. Brackenbury WJ, Djamgoz MB. Activity-dependent regulation of voltage-gated 26 Na+ channel expression in Mat-LyLu rat prostate cancer cell line. J Physiol 27 2006;573(Pt 2):343-56 doi: 10.1113/jphysiol.2006.106906published Online 28 First: Epub Date]|. 29 18. Brackenbury WJ, Djamgoz MB. Nerve growth factor enhances voltage-gated Na+ 30 channel activity and Transwell migration in Mat-LyLu rat prostate cancer cell 31 line. J Cell Physiol 2007;210(3):602-8 doi: 10.1002/jcp.20846published Online 32 First: Epub Date]|. 33 19. Nelson M, Millican-Slater R, Forrest LC, Brackenbury WJ. The sodium channel

34 http://bmjopen.bmj.com/ 35 beta1 subunit mediates outgrowth of neurite-like processes on breast cancer 36 cells and promotes tumour growth and metastasis. Int J Cancer 37 2014;135(10):2338-51 doi: 10.1002/ijc.28890published Online First: Epub 38 Date]|. 39 20. Brackenbury WJ, Isom LL. Na Channel beta Subunits: Overachievers of the Ion 40 Channel Family. Frontiers in pharmacology 2011;2:53 doi: 41 10.3389/fphar.2011.00053published Online First: Epub Date]|. 42 21. Chioni AM, Brackenbury WJ, Calhoun JD, Isom LL, Djamgoz MB. A novel on September 29, 2021 by guest. Protected copyright. 43 adhesion molecule in human breast cancer cells: voltage-gated Na+ channel 44 beta1 subunit. Int J Biochem Cell Biol 2009;41(5):1216-27 doi: 45 10.1016/j.biocel.2008.11.001published Online First: Epub Date]|. 46 22. Fraser SP, Salvador V, Manning EA, et al. Contribution of functional voltage- + 47 gated Na channel expression to cell behaviors involved in the metastatic 48 cascade in rat prostate cancer: I. lateral motility. J Cell Physiol 49 2003;195(3):479-87 50 23. Yang M, Kozminski DJ, Wold LA, et al. Therapeutic potential for phenytoin: 51 targeting Na(v)1.5 sodium channels to reduce migration and invasion in 52 metastatic breast cancer. Breast Cancer Res Treat 2012;134(2):603-15 doi: 53 10.1007/s10549-012-2102-9published Online First: Epub Date]|. 54 24. Brackenbury WJ, Isom LL. Voltage-gated Na+ channels: potential for beta 55 subunits as therapeutic targets. Expert Opin Ther Targets 2008;12(9):1191- 56 203 doi: 10.1517/14728222.12.9.1191published Online First: Epub Date]|. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 15 BMJ Open Page 16 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 25. Singh G, Driever PH, Sander JW. Cancer risk in people with epilepsy: the role of 4 antiepileptic drugs. Brain 2005;128(Pt 1):7-17 doi: 5 10.1093/brain/awh363published Online First: Epub Date]|. 6 26. Brisson L, Driffort V, Benoist L, et al. NaV1.5 Na(+) channels allosterically 7 regulate the NHE-1 exchanger and promote the activity of breast cancer cell 8 invadopodia. J Cell Sci 2013;126(Pt 21):4835-42 doi: 9 10.1242/jcs.123901published Online First: Epub Date]|. 10 27. Ding L, Ellis MJ, Li S, et al. Genome remodelling in a basal-like breast cancer 11 metastasis and xenograft. Nature 2010;464(7291):999-1005 doi: 12 10.1038/nature08989published Online First: Epub Date]|. 13 28. Grimes JA, Djamgoz MB. Electrophysiological characterization of voltage-gated 14 Na(+) current expressed in the highly metastatic Mat-LyLu cell line of rat 15 prostateFor cancer. peer J Cell Physiol review 1998;175(1):50-8 only 16 29. Gillet L, Roger S, Besson P, et al. Voltage-gated Sodium Channel Activity 17 Promotes Cysteine Cathepsin-dependent Invasiveness and Colony Growth of 18 Human Cancer Cells. J Biol Chem 2009;284(13):8680-91 doi: M806891200 19 [pii] 10.1074/jbc.M806891200published Online First: Epub Date]|. 20 30. Ding Y, Brackenbury WJ, Onganer PU, et al. Epidermal growth factor upregulates 21 motility of Mat-LyLu rat prostate cancer cells partially via voltage-gated Na+ 22 channel activity. J Cell Physiol 2008;215(1):77-81 doi: 23 10.1002/jcp.21289published Online First: Epub Date]|. 24 31. Vinogradova Y, Coupland C, Hippisley-Cox J. Exposure to bisphosphonates and 25 risk of gastrointestinal cancers: series of nested case-control studies with 26 QResearch and CPRD data. BMJ (Clinical research ed.) 2013;346:f114 doi: 27 10.1136/bmj.f114published Online First: Epub Date]|. 28 32. Reeves D, Springate DA, Ashcroft DM, et al. Can analyses of electronic patient 29 records be independently and externally validated? The effect of statins on 30 the mortality of patients with ischaemic heart disease: a cohort study with 31 nested case-control analysis. BMJ open 2014;4(4):e004952 doi: 32 10.1136/bmjopen-2014-004952published Online First: Epub Date]|. 33 33. British National Formulary London: British Medical Association and Royal

34 http://bmjopen.bmj.com/ 35 Pharmaceutical Society of Great Britain, 2013. 36 34. Gupta GP, Massague J. Cancer metastasis: building a framework. Cell 37 2006;127(4):679-95 doi: 10.1016/j.cell.2006.11.001published Online First: 38 Epub Date]|. 39 35. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans. 40 Volume 44 Alcohol drinking, 1988. 41 36. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans. 42 Volume 83 Tobacco smoke and involuntary smoking, 2004. on September 29, 2021 by guest. Protected copyright. 43 37. Reeves GK, Pirie K, Beral V, Green J, Spencer E, Bull D. Cancer incidence and 44 mortality in relation to body mass index in the Million Women Study: cohort 45 study. BMJ (Clinical research ed.) 2007;335(7630):1134 doi: 46 10.1136/bmj.39367.495995.AEpublished Online First: Epub Date]|. 47 38. National Cancer Intelligence Network and Cancer Research UK. Cancer 48 Incidence and Survival by Major Ethnic Group, England 2002-2006., 2009. 49 39. Coleman RE, Rubens RD. The clinical course of bone metastases from breast 50 cancer. Br J Cancer 1987;55(1):61-6 51 40. Yong M, Jensen AO, Jacobsen JB, Norgaard M, Fryzek JP, Sorensen HT. 52 Survival in breast cancer patients with bone metastases and skeletal-related 53 events: a population-based cohort study in Denmark (1999-2007). Breast 54 Cancer Res Treat 2011;129(2):495-503 doi: 10.1007/s10549-011-1475- 55 5published Online First: Epub Date]|. 56 41. Jensen AO, Jacobsen JB, Norgaard M, Yong M, Fryzek JP, Sorensen HT. 57 Incidence of bone metastases and skeletal-related events in breast cancer 58 patients: a population-based cohort study in Denmark. BMC Cancer 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 16 Page 17 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 2011;11:29 doi: 10.1186/1471-2407-11-29published Online First: Epub 4 Date]|. 5 42. Joint Epilepsy Council of the UK and Ireland. Epilepsy prevalence, incidence and 6 other statistics. 7 http://www.epilepsyscotland.org.uk/pdf/Joint_Epilepsy_Council_Prevalence_ 8 and_Incidence_September_11_(3).pdf 2011 9 43. Nicholas JM, Ridsdale L, Richardson MP, Ashworth M, Gulliford MC. Trends in 10 antiepileptic drug utilisation in UK primary care 1993-2008: cohort study using 11 the General Practice Research Database. Seizure : the journal of the British 12 Epilepsy Association 2012;21(6):466-70 doi: 13 10.1016/j.seizure.2012.04.014published Online First: Epub Date]|. 14 44. Fang MC, Stafford RS, Ruskin JN, Singer DE. National trends in antiarrhythmic 15 andFor antithrombotic peer medication review use in atrial fibrillation. only Arch Intern Med 16 2004;164(1):55-60 doi: 10.1001/archinte.164.1.55published Online First: 17 Epub Date]|. 18 45. Cancer Research UK. Lifetime risk of cancer. 2014. 19 http://www.cancerresearchuk.org/cancer- 20 21 info/cancerstats/incidence/risk/statistics-on-the-risk-of-developing-cancer - 22 Lifetime5). 23 46. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. 24 Strengthening the Reporting of Observational Studies in Epidemiology 25 (STROBE) statement: guidelines for reporting observational studies. BMJ 26 (Clinical research ed.) 2007;335(7624):806-8 doi: 27 10.1136/bmj.39335.541782.ADpublished Online First: Epub Date]|. 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 17 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 18 of 36

1 2 3 Group Name Read code Description 4 5 Breast cancer B34 Malignant neoplasm of female breast 6 Breast cancer B34-1 Ca female breast 7 Breast cancer B34-98 Ca breast - NOS 8 Breast cancer B34-99 Carcinoma breast 9 Breast cancer B340 Malignant neoplasm of nipple and areola of female breast 10 Breast cancer B340-99 Ca breast - nipple/central 11 Breast cancer B3400 ForMalignant neoplasmpeer of nipple ofreview female breast only 12 Breast cancer B3401 Malignant neoplasm of areola of female breast 13 Breast cancer B340z Malignant neoplasm of nipple or areola of female breast NOS 14 15 Breast cancer B341 Malignant neoplasm of central part of female breast 16 Breast cancer B342 Malignant neoplasm of upper-inner quadrant of female breast Breast cancer B342-99 Ca breast-upper,inner quadrant

17 http://bmjopen.bmj.com/ 18 Breast cancer B343 Malignant neoplasm of lower-inner quadrant of female breast 19 Breast cancer B343-99 Ca breast-lower,inner quadrant 20 Breast cancer B344 Malignant neoplasm of upper-outer quadrant of female breast 21 Breast cancer B344-99 Ca breast-upper,outer quadrant 22 Breast cancer B345 Malignant neoplasm of lower-outer quadrant of female breast 23 Breast cancer B345-99 Ca breast-lower,outer quadrant 24 25 Breast cancer B346 Malignant neoplasm of axillary tail of female breast 26 Breast cancer B346-99 Ca breast - axillary tail on September 29, 2021 by guest. Protected copyright. 27 Breast cancer B347 Malignant neoplasm, overlapping lesion of breast 28 Breast cancer B34y Malignant neoplasm of other site of female breast 29 Breast cancer B34y0 Malignant neoplasm of ectopic site of female breast 30 Breast cancer B34yz Malignant neoplasm of other site of female breast NOS 31 Breast cancer B34z Malignant neoplasm of female breast NOS 32 Breast cancer B34z-99 Ca breast - NOS 33 Breast cancer B35 Malignant neoplasm of male breast 34 35 Breast cancer B35-99 Ca breast - male 36 Breast cancer B350 Malignant neoplasm of nipple and areola of male breast 37 Breast cancer B3500 Malignant neoplasm of nipple of male breast 38 Breast cancer B3501 Malignant neoplasm of areola of male breast 39 Breast cancer B350z Malignant neoplasm of nipple or areola of male breast NOS 40 Breast cancer B35z Malignant neoplasm of other site of male breast 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 19 of 36 BMJ Open

1 2 3 Breast cancer B35z0 Malignant neoplasm of ectopic site of male breast 4 5 Breast cancer B35zz Malignant neoplasm of male breast NOS 6 Breast cancer B36 Local recurrence of malignant tumour of breast 7 Breast cancer B58y0 Secondary malignant neoplasm of breast 8 Breast cancer B830 Carcinoma in situ of breast 9 Breast cancer B830-99 Carcinoma in situ breast 10 Breast cancer B8300 Lobular carcinoma in situ of breast 11 Breast cancer B8301 ForIntraductal peercarcinoma in situ ofreview breast only 12 Breast cancer B933 Neoplasm of uncertain behaviour of breast 13 Breast cancer BA03 Neoplasm of unspecified nature of breast 14 15 Breast cancer BB94-1 [M]Secretory breast carcinoma 16 Breast cancer Byu6 [X]Malignant neoplasm of breast Breast cancer EMISNQLO13 Local recurrence of malignant tumour of breast

17 http://bmjopen.bmj.com/ 18 Breast cancer HNG0190 [RFC] Breast cancer 19 Breast cancer ZV103 [V]Personal history of malignant neoplasm of breast 20 21 Group Name Read code Description 22 Colorectal cancer B13 Malignant neoplasm of colon 23 Colorectal cancer B130 Malignant neoplasm of hepatic flexure of colon 24 25 Colorectal cancer B130-99 Ca hepatic flexure - colon 26 Colorectal cancer B131 Malignant neoplasm of transverse colon on September 29, 2021 by guest. Protected copyright. 27 Colorectal cancer B131-99 Ca transverse colon 28 Colorectal cancer B132 Malignant neoplasm of descending colon 29 Colorectal cancer B132-99 Ca descending colon 30 Colorectal cancer B133 Malignant neoplasm of sigmoid colon 31 Colorectal cancer B133-99 Ca sigmoid colon 32 Colorectal cancer B134 Malignant neoplasm of caecum 33 Colorectal cancer B134-1 Carcinoma of caecum 34 35 Colorectal cancer B135 Malignant neoplasm of appendix 36 Colorectal cancer B136 Malignant neoplasm of ascending colon 37 Colorectal cancer B136-99 Ca ascending colon 38 Colorectal cancer B137 Malignant neoplasm of splenic flexure of colon 39 Colorectal cancer B137-99 Ca splenic flexure - colon 40 Colorectal cancer B138 Malignant neoplasm, overlapping lesion of colon 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 20 of 36

1 2 3 Colorectal cancer B139 Hereditary nonpolyposis colon cancer 4 5 Colorectal cancer B13y Malignant neoplasm of other specified sites of colon 6 Colorectal cancer B13z Malignant neoplasm of colon NOS 7 Colorectal cancer B13z-1 Colonic cancer 8 Colorectal cancer B14 Malignant neoplasm of rectum, rectosigmoid junction and anus 9 Colorectal cancer B14-99 Ca rectum + Ca anus 10 Colorectal cancer B140 Malignant neoplasm of rectosigmoid junction 11 Colorectal cancer B140-99 ForCa rectosigmoid peer junction review only 12 Colorectal cancer B141 Malignant neoplasm of rectum 13 Colorectal cancer B141-1 Carcinoma of rectum 14 15 Colorectal cancer B141-2 Rectal carcinoma 16 Colorectal cancer B141-99 Adenocarcinoma of rectum Colorectal cancer B1420 Malignant neoplasm of cloacogenic zone

17 http://bmjopen.bmj.com/ 18 Colorectal cancer B143 Malignant neoplasm of anus unspecified 19 Colorectal cancer B14y Malig neop other site rectum, rectosigmoid junction and anus 20 Colorectal cancer B14z Malignant neoplasm rectum,rectosigmoid junction and anus NOS 21 Colorectal cancer B14z-99 Ca rectum + Ca anus NOS 22 Colorectal cancer B1z0-1 Cancer of bowel 23 Colorectal cancer B575 Secondary malignant neoplasm of large intestine and rectum 24 25 Colorectal cancer B5750 Secondary malignant neoplasm of colon 26 Colorectal cancer B5751 Secondary malignant neoplasm of rectum on September 29, 2021 by guest. Protected copyright. 27 Colorectal cancer B575z Secondary malig neop of large intestine or rectum NOS 28 Colorectal cancer B803 Carcinoma in situ of colon 29 Colorectal cancer B8030 Carcinoma in situ of hepatic flexure of colon 30 Colorectal cancer B8031 Carcinoma in situ of transverse colon 31 Colorectal cancer B8032 Carcinoma in situ of descending colon 32 Colorectal cancer B8033 Carcinoma in situ of sigmoid colon 33 Colorectal cancer B8036 Carcinoma in situ of ascending colon 34 35 Colorectal cancer B8037 Carcinoma in situ of splenic flexure of colon 36 Colorectal cancer B8038 High grade dysplasia of colon 37 Colorectal cancer B803z Carcinoma in situ of colon NOS 38 Colorectal cancer B804 Carcinoma in situ of rectum and rectosigmoid junction 39 Colorectal cancer B8040 Carcinoma in situ of rectosigmoid junction 40 Colorectal cancer B8041 Carcinoma in situ of rectum 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 21 of 36 BMJ Open

1 2 3 Colorectal cancer B804z Carcinoma in situ of rectum or rectosigmoid junction NOS 4 5 Colorectal cancer B902 Neop of uncertain behaviour stomach, intestines and rectum 6 Colorectal cancer B9024 Neoplasm of uncertain behaviour of colon 7 Colorectal cancer B9025 Neoplasm of uncertain behaviour of rectum 8 Colorectal cancer B902z Neop of uncertain behaviour stomach, intestine or rectum NOS 9 Colorectal cancer BB5N [M]Adenomatous and adenocarcinomatous polyps of colon 10 Colorectal cancer EGTON2B152 Cause of Death- Ca Rectum 11 Colorectal cancer EMISNQHE41ForHereditary peernonpolyposis colon review cancer only 12 Colorectal cancer HNG0201 [RFC] Cancer of the rectum 13 14 15 Group Name Read code Description 16 Prostate cancer 14270 H/O: prostate cancer Prostate cancer 1J08 Suspected prostate cancer

17 http://bmjopen.bmj.com/ 18 Prostate cancer 4M0 Gleason grading of prostate cancer 19 Prostate cancer 4M00 Gleason prostate grade 2-4 (low) 20 Prostate cancer 4M01 Gleason prostate grade 5-7 (medium) 21 Prostate cancer 4M02 Gleason prostate grade 8-10 (high) 22 Prostate cancer B46 Malignant neoplasm of prostate 23 Prostate cancer B58y5 Secondary malignant neoplasm of prostate 24 25 Prostate cancer B7C20 Adenoma of prostate 26 Prostate cancer B834 Carcinoma in situ of prostate on September 29, 2021 by guest. Protected copyright. 27 Prostate cancer B8340 High grade prostatic intraepithelial neoplasia 28 Prostate cancer B8341 Prostatic intraepithelial neoplasia 29 Prostate cancer B915 Neoplasm of uncertain behaviour of prostate 30 Prostate cancer EMISNQPR140 Prostatic intraepithelial neoplasia 31 Prostate cancer HNG0200 [RFC] Cancer of the prostate 32 Prostate cancer K223 Dysplasia of prostate 33 Prostate cancer PCSDT1PR16 Prostate carcinoma 34 35 Prostate cancer ZV104-5 [V]Personal history of malignant neoplasm of prostate 36 37 Group Name Read code Description 38 Alcohol status 136 Alcohol consumption 39 Alcohol status 1361 Teetotaller 40 Alcohol status 1361-1 Non drinker alcohol 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 22 of 36

1 2 3 Alcohol status 1361-2 Non-drinker alcohol 4 5 Alcohol status 1362 Trivial drinker - <1u/day 6 Alcohol status 1362-1 Drinks rarely 7 Alcohol status 1362-2 Drinks occasionally 8 Alcohol status 1363 Light drinker - 1-2u/day 9 Alcohol status 1364 Moderate drinker - 3-6u/day 10 Alcohol status 1365 Heavy drinker - 7-9u/day 11 Alcohol status 1366 ForVery heavy peer drinker - >9u/day review only 12 Alcohol status 1367 Stopped drinking alcohol 13 Alcohol status 1368 Alcohol consumption unknown 14 15 Alcohol status 1369 Suspect alcohol abuse - denied 16 Alcohol status 136A Ex-trivial drinker (<1u/day) Alcohol status 136a Increasing risk drinking

17 http://bmjopen.bmj.com/ 18 Alcohol status 136B Ex-light drinker - (1-2u/day) 19 Alcohol status 136b Feels should cut down drinking 20 Alcohol status 136C Ex-moderate drinker - (3-6u/d) 21 Alcohol status 136c Higher risk drinking 22 Alcohol status 136D Ex-heavy drinker - (7-9u/day) 23 Alcohol status 136d Lower risk drinking 24 25 Alcohol status 136E Ex-very heavy drinker-(>9u/d) 26 Alcohol status 136e Declines to state current alcohol consumption on September 29, 2021 by guest. Protected copyright. 27 Alcohol status 136F Spirit drinker 28 Alcohol status 136G Beer drinker 29 Alcohol status 136H Drinks beer and spirits 30 Alcohol status 136I Drinks wine 31 Alcohol status 136J Social drinker 32 Alcohol status 136K Alcohol intake above recommended sensible limits 33 Alcohol status 136L Alcohol intake within recommended sensible limits 34 35 Alcohol status 136M Current non drinker 36 Alcohol status 136N Light drinker 37 Alcohol status 136O Moderate drinker 38 Alcohol status 136P Heavy drinker 39 Alcohol status 136Q Very heavy drinker 40 Alcohol status 136R Binge drinker 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 23 of 36 BMJ Open

1 2 3 Alcohol status 136S Hazardous alcohol use 4 5 Alcohol status 136T Harmful alcohol use 6 Alcohol status 136V Alcohol units per week 7 Alcohol status 136X Alcohol units consumed on heaviest drinking day 8 Alcohol status 136Y Drinks in morning to get rid of hangover 9 Alcohol status 136Z Alcohol consumption NOS 10 Alcohol status E23-2 Alcohol problem drinking 11 Alcohol status EGTON418 ForAlcohol intake peer review only 12 Alcohol status EGTONGR6 Grade A none drinker or rare 13 Alcohol status EGTONGR7 Grade B 0-20 units/week 14 15 Alcohol status EGTONGR8 Grade C 21-49 units/week 16 Alcohol status EGTONGR9 Grade D 50 or more units/week Alcohol status EMISNQUN23 Unsuccessful attempts to reduce alcohol consumption

17 http://bmjopen.bmj.com/ 18 Alcohol status EMISQAL1 Alcohol AUDIT score 19 Alcohol status EMISQAL2 Alcohol reduction-maintain abstinence 20 Alcohol status EMISQAU1 AUDIT score <4 no drinking problem 21 Alcohol status EMISQAU2 AUDIT score >5 alcohol dependence syndrome 22 Alcohol status EMISQAU3 AUDIT score >5 alcohol problem drinker 23 Alcohol status EMISQNO6 Not interested in reducing alcohol 24 25 Alcohol status EMISQOT3 Other alcohol related information 26 Alcohol status EMISQRE13 Ready to start reducing alcohol on September 29, 2021 by guest. Protected copyright. 27 Alcohol status EMISQTH4 Thinking about reducing alcohol 28 29 Group Name Read code Description 30 BMI 22K Body Mass Index 31 BMI 22K1 Body Mass Index normal K/M2 32 BMI 22K2 Body Mass Index high K/M2 33 BMI 22K3 Body Mass Index low K/M2 34 35 BMI 22K4 Body mass index index 25-29 - overweight 36 BMI 22K5 Body mass index 30+ - obesity 37 BMI 22K6 Body mass index less than 20 38 BMI 22K7 Body mass index 40+ - severely obese 39 BMI 22K8 Body mass index 20-24 - normal 40 BMI 22K9 Body mass index centile 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 24 of 36

1 2 3 BMI 22K90 Baseline body mass index centile 4 5 BMI 22KA Target body mass index 6 BMI 22KB Baseline body mass index 7 BMI EMISNQBM1 BMI centile 8 BMI EMISNQBO29 Body mass index less than 18.5 9 BMI EMISNQTA9 Target body mass index 10 BMI JHCBO5 Body mass index 18.5-24.9 11 BMI PCNQBO1 ForBody mass peer index review only 12 13 Group Name Read code Description 14 15 Ethnicity 226 O/E - ethnic group 16 Ethnicity 226-1 O/E - ethnic origin Ethnicity 226Z O/E - ethnic group NOS

17 http://bmjopen.bmj.com/ 18 Ethnicity 916E Patient ethnicity unknown 19 Ethnicity 918t Carer from Black and minority ethnic group 20 Ethnicity 9i Ethnic category - 2001 census 21 Ethnicity 9i0 British or mixed British - ethnic category 2001 census 22 Ethnicity 9i00 White British - ethnic category 2001 census 23 Ethnicity 9i1 Irish - ethnic category 2001 census 24 25 Ethnicity 9i10 White Irish - ethnic category 2001 census 26 Ethnicity 9i2 Other White background - ethnic category 2001 census on September 29, 2021 by guest. Protected copyright. 27 Ethnicity 9i20 English - ethnic category 2001 census 28 Ethnicity 9i21 Scottish - ethnic category 2001 census 29 Ethnicity 9i22 Welsh - ethnic category 2001 census 30 Ethnicity 9i23 Cornish - ethnic category 2001 census 31 Ethnicity 9i24 Northern Irish - ethnic category 2001 census 32 Ethnicity 9i25 Ulster Scots - ethnic category 2001 census 33 Ethnicity 9i26 Cypriot (part not stated) - ethnic category 2001 census 34 35 Ethnicity 9i27 Greek - ethnic category 2001 census 36 Ethnicity 9i28 Greek Cypriot - ethnic category 2001 census 37 Ethnicity 9i29 Turkish - ethnic category 2001 census 38 Ethnicity 9i2A Turkish Cypriot - ethnic category 2001 census 39 Ethnicity 9i2B Italian - ethnic category 2001 census 40 Ethnicity 9i2C Irish Traveller - ethnic category 2001 census 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 25 of 36 BMJ Open

1 2 3 Ethnicity 9i2D Traveller - ethnic category 2001 census 4 5 Ethnicity 9i2E Gypsy/Romany - ethnic category 2001 census 6 Ethnicity 9i2F Polish - ethnic category 2001 census 7 Ethnicity 9i2J Kosovan - ethnic category 2001 census 8 Ethnicity 9i2K Albanian - ethnic category 2001 census 9 Ethnicity 9i2L Bosnian - ethnic category 2001 census 10 Ethnicity 9i2M Croatian - ethnic category 2001 census 11 Ethnicity 9i2N ForSerbian - ethnicpeer category 2001 review census only 12 Ethnicity 9i2P Other republics former Yugoslavia - ethnic categ 2001 census 13 Ethnicity 9i2Q Mixed Irish and other White - ethnic category 2001 census 14 15 Ethnicity 9i2S Other mixed White - ethnic category 2001 census 16 Ethnicity 9i2T Other White or White unspecified ethnic category 2001 census Ethnicity 9i3 White and Black Caribbean - ethnic category 2001 census

17 http://bmjopen.bmj.com/ 18 Ethnicity 9i4 White and Black African - ethnic category 2001 census 19 Ethnicity 9i5 White and Asian - ethnic category 2001 census 20 Ethnicity 9i6 Other Mixed background - ethnic category 2001 census 21 Ethnicity 9i60 Black and Asian - ethnic category 2001 census 22 Ethnicity 9i61 Black and Chinese - ethnic category 2001 census 23 Ethnicity 9i62 Black and White - ethnic category 2001 census 24 25 Ethnicity 9i63 Chinese and White - ethnic category 2001 census 26 Ethnicity 9i64 Asian and Chinese - ethnic category 2001 census on September 29, 2021 by guest. Protected copyright. 27 Ethnicity 9i65 Other Mixed or Mixed unspecified ethnic category 2001 census 28 Ethnicity 9i7 Indian or British Indian - ethnic category 2001 census 29 Ethnicity 9i8 Pakistani or British Pakistani - ethnic category 2001 census 30 Ethnicity 9iA Other Asian background - ethnic category 2001 census 31 Ethnicity 9iA1 Punjabi - ethnic category 2001 census 32 Ethnicity 9iA2 Kashmiri - ethnic category 2001 census 33 Ethnicity 9iA3 East African Asian - ethnic category 2001 census 34 35 Ethnicity 9iA4 Sri Lankan - ethnic category 2001 census 36 Ethnicity 9iA5 Tamil - ethnic category 2001 census 37 Ethnicity 9iA6 Sinhalese - ethnic category 2001 census 38 Ethnicity 9iA7 Caribbean Asian - ethnic category 2001 census 39 Ethnicity 9iA8 British Asian - ethnic category 2001 census 40 Ethnicity 9iA9 Mixed Asian - ethnic category 2001 census 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 26 of 36

1 2 3 Ethnicity 9iAA Other Asian or Asian unspecified ethnic category 2001 census 4 5 Ethnicity 9iB Caribbean - ethnic category 2001 census 6 Ethnicity 9iC African - ethnic category 2001 census 7 Ethnicity 9iD Other Black background - ethnic category 2001 census 8 Ethnicity 9iD0 Somali - ethnic category 2001 census 9 Ethnicity 9iD1 Nigerian - ethnic category 2001 census 10 Ethnicity 9iD2 Black British - ethnic category 2001 census 11 Ethnicity 9iD3 ForMixed Black peer - ethnic category 2001review census only 12 Ethnicity 9iD4 Other Black or Black unspecified ethnic category 2001 census 13 Ethnicity 9iE Chinese - ethnic category 2001 census 14 15 Ethnicity 9iF Other - ethnic category 2001 census 16 Ethnicity 9iF0 Vietnamese - ethnic category 2001 census Ethnicity 9iF1 Japanese - ethnic category 2001 census

17 http://bmjopen.bmj.com/ 18 Ethnicity 9iF2 Filipino - ethnic category 2001 census 19 Ethnicity 9iF3 Malaysian - ethnic category 2001 census 20 Ethnicity 9iF4 Buddhist - ethnic category 2001 census 21 Ethnicity 9iF5 Hindu - ethnic category 2001 census 22 Ethnicity 9iF6 Jewish - ethnic category 2001 census 23 Ethnicity 9iF7 Muslim - ethnic category 2001 census 24 25 Ethnicity 9iF8 Sikh - ethnic category 2001 census 26 Ethnicity 9iF9 Arab - ethnic category 2001 census on September 29, 2021 by guest. Protected copyright. 27 Ethnicity 9iFA North African - ethnic category 2001 census 28 Ethnicity 9iFC Israeli - ethnic category 2001 census 29 Ethnicity 9iFD Iranian - ethnic category 2001 census 30 Ethnicity 9iFE Kurdish - ethnic category 2001 census 31 Ethnicity 9iFF Moroccan - ethnic category 2001 census 32 Ethnicity 9iFG Latin American - ethnic category 2001 census 33 Ethnicity 9iFH South and Central American - ethnic category 2001 census 34 35 Ethnicity 9iFK Any other group - ethnic category 2001 census 36 Ethnicity 9iG Ethnic category not stated - 2001 census 37 Ethnicity 9S Ethnic groups (census) 38 Ethnicity 9S12 Other white ethnic group 39 Ethnicity 9S14 Other white British ethnic group 40 Ethnicity 9SA Other ethnic non-mixed (NMO) 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 27 of 36 BMJ Open

1 2 3 Ethnicity 9SA1 Brit. ethnic minor. spec.(NMO) 4 5 Ethnicity 9SA2 Brit. ethnic minor. unsp (NMO) 6 Ethnicity 9SAD Other ethnic NEC (NMO) 7 Ethnicity 9SD Ethnic group not given - patient refused 8 Ethnicity 9SE Ethnic group not recorded 9 Ethnicity 9SG Other black ethnic group 10 Ethnicity 9SH Other Asian ethnic group 11 Ethnicity 9SJ ForOther ethnic peer group review only 12 Ethnicity 9SZ Ethnic groups (census) NOS 13 Ethnicity 9T Ethnicity and other related nationality data 14 15 Ethnicity 9T1 New Zealand ethnic groups 16 Ethnicity 9T11 New Zealand European Ethnicity 9T11-1 Pakeha

17 http://bmjopen.bmj.com/ 18 Ethnicity 9T12 Other European in New Zealand 19 Ethnicity 9T13 New Zealand Maori 20 Ethnicity 9T14 Samoan 21 Ethnicity 9T15 Cook Island Maori 22 Ethnicity 9T16 Tongan 23 Ethnicity 9T17 Niuean 24 25 Ethnicity 9T18 Tokelauan 26 Ethnicity 9T19 Fijian on September 29, 2021 by guest. Protected copyright. 27 Ethnicity 9T1A Other Pacific ethnic group 28 Ethnicity 9T1B South East Asian 29 Ethnicity 9T1C Chinese 30 Ethnicity 9T1D Indian 31 Ethnicity 9T1E Other Asian 32 Ethnicity 9T1Y Other New Zealand ethnic group 33 Ethnicity 9T1Z New Zealand ethnic group NOS 34 35 Ethnicity 9T2 Traveller - gypsy 36 Ethnicity 9T3 Yemeni 37 Ethnicity 9T4 Romanian 38 Ethnicity 9T5 Bulgarian 39 Ethnicity 9T6 Czech 40 Ethnicity 9T7 Slovak 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 28 of 36

1 2 3 Ethnicity 9T8 Portuguese 4 5 Ethnicity 9T9 Nepali 6 Ethnicity EMISNQBR14 Brazilian 7 Ethnicity EMISNQPO8 Portuguese 8 9 Group Name Read code Description 10 Metastases 1D18 Pain from metastases 11 Metastases B153 ForSecondary peer malignant neoplasm review of liver only 12 Metastases B56 Secondary and unspecified malignant neoplasm of lymph nodes 13 Metastases B56-1 Lymph node metastases 14 15 Metastases B560 Secondary and unspec malig neop lymph nodes head/face/neck 16 Metastases B560-99 Secondary nodes - head/neck Metastases B5600 Secondary and unspec malig neop of superficial parotid LN

17 http://bmjopen.bmj.com/ 18 Metastases B5601 Secondary and unspec malignant neoplasm mastoid lymph nodes 19 Metastases B5602 Secondary and unspec malig neop superficial cervical LN 20 Metastases B5603 Secondary and unspec malignant neoplasm occipital lymph node 21 Metastases B5604 Secondary and unspec malig neop deep parotid lymph nodes 22 Metastases B5605 Secondary and unspec malig neop submandibular lymph nodes 23 Metastases B5606 Secondary and unspec malig neop of facial lymph nodes 24 25 Metastases B5607 Secondary and unspec malig neop submental lymph nodes 26 Metastases B5608 Secondary and unspec malig neop anterior cervical LN on September 29, 2021 by guest. Protected copyright. 27 Metastases B5609 Secondary and unspec malig neop deep cervical LN 28 Metastases B560z Secondary unspec malig neop lymph nodes head/face/neck NOS 29 Metastases B561 Secondary and unspec malig neop intrathoracic lymph nodes 30 Metastases B561-99 Secondary nodes -intrathoracic 31 Metastases B5610 Secondary and unspec malig neop internal mammary lymph nodes 32 Metastases B5611 Secondary and unspec malig neop intercostal lymph nodes 33 Metastases B5612 Secondary and unspec malig neop diaphragmatic lymph nodes 34 35 Metastases B5613 Secondary and unspec malig neop ant mediastinal lymph nodes 36 Metastases B5614 Secondary and unspec malig neop post mediastinal lymph nodes 37 Metastases B5615 Secondary and unspec malig neop paratracheal lymph nodes 38 Metastases B5616 Secondary and unspec malig neop superfic tracheobronchial LN 39 Metastases B5617 Secondary and unspec malig neop inferior tracheobronchial LN 40 Metastases B5618 Secondary and unspec malig neop bronchopulmonary lymph nodes 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 29 of 36 BMJ Open

1 2 3 Metastases B5619 Secondary and unspec malig neop pulmonary lymph nodes 4 5 Metastases B561z Secondary and unspec malig neop intrathoracic LN NOS 6 Metastases B562 Secondary and unspec malig neop intra-abdominal lymph nodes 7 Metastases B562-99 Secondary nodes - intra.abdom. 8 Metastases B5620 Secondary and unspec malig neop coeliac lymph nodes 9 Metastases B5621 Secondary and unspec malig neop superficial mesenteric LN 10 Metastases B5622 Secondary and unspec malig neop inferior mesenteric LN 11 Metastases B5623 ForSecondary peer and unspec malig neopreview common iliac lymph nodesonly 12 Metastases B5624 Secondary and unspec malig neop external iliac lymph nodes 13 Metastases B562z Secondary and unspec malig neop intra-abdominal LN NOS 14 15 Metastases B563 Secondary and unspec malig neop axilla and upper limb LN 16 Metastases B563-99 Secondary nodes - axilla/arm Metastases B5630 Secondary and unspec malig neop axillary lymph nodes

17 http://bmjopen.bmj.com/ 18 Metastases B5631 Secondary and unspec malig neop supratrochlear lymph nodes 19 Metastases B5632 Secondary and unspec malig neop infraclavicular lymph nodes 20 Metastases B5633 Secondary and unspec malig neop pectoral lymph nodes 21 Metastases B563z Secondary and unspec malig neop axilla and upper limb LN NOS 22 Metastases B564 Secondary and unspec malig neop inguinal and lower limb LN 23 Metastases B564-99 Secondary nodes - inguinal/leg 24 25 Metastases B5640 Secondary and unspec malig neop superficial inguinal LN 26 Metastases B5641 Secondary and unspec malig neop deep inguinal lymph nodes on September 29, 2021 by guest. Protected copyright. 27 Metastases B5642 Secondary and unspec malig neop popliteal lymph nodes 28 Metastases B564z Secondary and unspec malig neop of inguinal and leg LN NOS 29 Metastases B565 Secondary and unspec malig neop intrapelvic lymph nodes 30 Metastases B565-99 Secondary nodes - intrapelvic 31 Metastases B5650 Secondary and unspec malig neop internal iliac lymph nodes 32 Metastases B5651 Secondary and unspec malig neop inferior epigastric LN 33 Metastases B5652 Secondary and unspec malig neop circumflex iliac LN 34 35 Metastases B5653 Secondary and unspec malig neop sacral lymph nodes 36 Metastases B5654 Secondary and unspec malig neop obturator lymph nodes 37 Metastases B565z Secondary and unspec malig neop intrapelvic LN NOS 38 Metastases B56y Secondary and unspec malig neop lymph nodes multiple sites 39 Metastases B56y-99 Secondary nodes - multiple 40 Metastases B56z Secondary and unspec malig neop lymph nodes NOS 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 30 of 36

1 2 3 Metastases B56z-99 Secondary nodes NOS 4 5 Metastases B57-1 Metastases of respiratory and/or digestive systems 6 Metastases B570 Secondary malignant neoplasm of lung 7 Metastases B571 Secondary malignant neoplasm of mediastinum 8 Metastases B572 Secondary malignant neoplasm of pleura 9 Metastases B573 Secondary malignant neoplasm of other respiratory organs 10 Metastases B574 Secondary malignant neoplasm of small intestine and duodenum 11 Metastases B5740 ForSecondary peer malignant neoplasm review of duodenum only 12 Metastases B5741 Secondary malignant neoplasm of jejunum 13 Metastases B5742 Secondary malignant neoplasm of ileum 14 15 Metastases B574z Secondary malig neop of small intestine or duodenum NOS 16 Metastases B575 Secondary malignant neoplasm of large intestine and rectum Metastases B5750 Secondary malignant neoplasm of colon

17 http://bmjopen.bmj.com/ 18 Metastases B5751 Secondary malignant neoplasm of rectum 19 Metastases B575z Secondary malig neop of large intestine or rectum NOS 20 Metastases B5760 Secondary malignant neoplasm of retroperitoneum 21 Metastases B5761 Secondary malignant neoplasm of peritoneum 22 Metastases B577 Secondary malignant neoplasm of liver 23 Metastases B577-1 Liver metastases 24 25 Metastases B57y Secondary malignant neoplasm of other digestive organ 26 Metastases B58 Secondary malignant neoplasm of other specified sites on September 29, 2021 by guest. Protected copyright. 27 Metastases B580 Secondary malignant neoplasm of kidney 28 Metastases B5823 Secondary malignant neoplasm of skin of trunk 29 Metastases B5826 Secondary malignant neoplasm of skin of breast 30 Metastases B583 Secondary malignant neoplasm of brain and spinal cord 31 Metastases B583-99 Secondary Ca brain/spinal cord 32 Metastases B5830 Secondary malignant neoplasm of brain 33 Metastases B5831 Secondary malignant neoplasm of spinal cord 34 35 Metastases B5832 Cerebral metastasis 36 Metastases B583z Secondary malignant neoplasm of brain or spinal cord NOS 37 Metastases B584 Secondary malignant neoplasm of other part of nervous system 38 Metastases B585 Secondary malignant neoplasm of bone and bone marrow 39 Metastases B585-99 Secondary Ca bone/bone marrow 40 Metastases B5850 Pathological fracture due to metastatic bone disease 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 31 of 36 BMJ Open

1 2 3 Metastases B586 Secondary malignant neoplasm of ovary 4 5 Metastases B587 Secondary malignant neoplasm of adrenal gland 6 Metastases B58y Secondary malignant neoplasm of other specified sites 7 Metastases B58y0 Secondary malignant neoplasm of breast 8 Metastases B58y5 Secondary malignant neoplasm of prostate 9 Metastases B58yz Secondary malignant neoplasm of other specified site NOS 10 Metastases B58z Secondary malignant neoplasm of other specified site NOS 11 Metastases B58z-99 ForSecondary peer Ca NOS review only 12 Metastases B594 Secondary malignant neoplasm of unknown site 13 Metastases ByuC [X]Malignant neoplasm of ill-defined, secondary and unspeci 14 15 Metastases ByuC3 [X]Secondary malignant neoplasm/oth+unspc respiratory organs 16 Metastases ByuC4 [X]Secondary malignant neoplasm/oth+unspcfd digestive organs Metastases ByuC7 [X]Secondary malignant neoplasm of other specified sites

17 http://bmjopen.bmj.com/ 18 Metastases EMISNQME41 Metastasis stage M0 19 Metastases EMISNQME42 Metastasis stage M1 20 Metastases EMISNQME43 Metastasis stage M2 21 Metastases EMISNQME44 Metastasis stage M3 22 Metastases EMISNQME45 Metastasis stage M4 23 Metastases EMISNQME46 Metastasis stage M1a 24 25 Metastases EMISNQME47 Metastasis stage M1b 26 Metastases EMISNQME48 Metastasis stage M1c on September 29, 2021 by guest. Protected copyright. 27 28 Group Name Read code Description 29 Smoking status 137-1 Smoker - amount smoked 30 Smoking status 1371 Never smoked tobacco 31 Smoking status 1371-1 Non-smoker 32 Smoking status 1372 Trivial smoker - < 1 cig/day 33 Smoking status 1372-1 Occasional smoker 34 35 Smoking status 1373 Light smoker - 1-9 cigs/day 36 Smoking status 1374 Moderate smoker - 10-19 cigs/d 37 Smoking status 1375 Heavy smoker - 20-39 cigs/day 38 Smoking status 1376 Very heavy smoker - 40+cigs/d 39 Smoking status 1377 Ex-trivial smoker (<1/day) 40 Smoking status 1378 Ex-light smoker (1-9/day) 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 32 of 36

1 2 3 Smoking status 1379 Ex-moderate smoker (10-19/day) 4 5 Smoking status 137A Ex-heavy smoker (20-39/day) 6 Smoking status 137a Pipe tobacco consumption 7 Smoking status 137B Ex-very heavy smoker (40+/day) 8 Smoking status 137b Ready to stop smoking 9 Smoking status 137C Keeps trying to stop smoking 10 Smoking status 137c Thinking about stopping smoking 11 Smoking status 137D ForAdmitted tobaccopeer cons untrue review? only 12 Smoking status 137d Not interested in stopping smoking 13 Smoking status 137E Tobacco consumption unknown 14 15 Smoking status 137e Smoking restarted 16 Smoking status 137F Ex-smoker - amount unknown Smoking status 137f Reason for restarting smoking

17 http://bmjopen.bmj.com/ 18 Smoking status 137F-99 EX-Smoker NOS 19 Smoking status 137G Trying to give up smoking 20 Smoking status 137g Cigarette pack-years 21 Smoking status 137H Pipe smoker 22 Smoking status 137h Minutes from waking to first tobacco consumption 23 Smoking status 137I Passive smoker 24 25 Smoking status 137i Ex-tobacco chewer 26 Smoking status 137I0 Exposed to tobacco smoke at home on September 29, 2021 by guest. Protected copyright. 27 Smoking status 137J Cigar smoker 28 Smoking status 137j Ex-cigarette smoker 29 Smoking status 137K Stopped smoking 30 Smoking status 137k Refusal to give smoking status 31 Smoking status 137K0 Recently stopped smoking 32 Smoking status 137L Current non-smoker 33 Smoking status 137l Ex roll-up cigarette smoker 34 35 Smoking status 137L-99 Tobacco Consumption Nil 36 Smoking status 137M Rolls own cigarettes 37 Smoking status 137m Failed attempt to stop smoking 38 Smoking status 137N Ex pipe smoker 39 Smoking status 137n Total time smoked 40 Smoking status 137O Ex cigar smoker 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Page 33 of 36 BMJ Open

1 2 3 Smoking status 137o Waterpipe tobacco consumption 4 5 Smoking status 137P Cigarette smoker 6 Smoking status 137P-1 Smoker 7 Smoking status 137Q Smoking started 8 Smoking status 137Q-1 Smoking restarted 9 Smoking status 137R Current smoker 10 Smoking status 137S Ex smoker 11 Smoking status 137T ForDate ceased peer smoking review only 12 Smoking status 137U Not a passive smoker 13 Smoking status 137V Smoking reduced 14 15 Smoking status 137W Chews tobacco 16 Smoking status 137X Cigarette consumption Smoking status 137Y Cigar consumption

17 http://bmjopen.bmj.com/ 18 Smoking status 137Z Tobacco consumption NOS 19 Smoking status 13p Smoking cessation milestones 20 Smoking status 13p0 Negotiated date for cessation of smoking 21 Smoking status 13p1 Smoking status at 4 weeks 22 Smoking status 13p2 Smoking status between 4 and 52 weeks 23 Smoking status 13p3 Smoking status at 52 weeks 24 25 Smoking status 13p4 Smoking free weeks 26 Smoking status 13p7 Smoking status at 12 weeks on September 29, 2021 by guest. Protected copyright. 27 Smoking status 13WF-1 Smoker in the family 28 Smoking status 9ko Current smoker annual review - enhanced services admin 29 Smoking status 9ko-1 Current smoker annual review 30 Smoking status ASDFGNO1 Non Smoker - Nos 31 Smoking status EGTON1024 Rolls own cigarettes 32 Smoking status EGTON1025 Current Smoker NOS 33 Smoking status EGTON1026 Ex-cigar smoker 34 35 Smoking status EGTON1027 Ex- Rolled Tobacco Smoker 36 Smoking status EGTON1028 Ex-smoker NOS 37 Smoking status EGTON320 Smoking Status 38 Smoking status EGTON321 Cigarette smoker 39 Smoking status EGTON322 Ex-Cigarette Smoker 40 Smoking status EGTON324 Ex-pipe smoker 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

BMJ Open Page 34 of 36

1 2 3 Smoking status EGTON326 Current smoker 4 5 Smoking status EGTON327 Past smoker 6 Smoking status EGTON328 Date stopped smoking 7 Smoking status EGTONGR11 Grade B light smoker (1-10/day) 8 Smoking status EGTONGR12 Grade C moderate smoker (11-20/day) 9 Smoking status EGTONGR13 Grade D heavy smoker (>20 Day) 10 Smoking status EGTONSM2 Smoking Age 11 Smoking status EGTONSM3ForSmoking Agepeer Started review only 12 Smoking status EGTONSM4 Smoking Age Ceased 13 Smoking status EGTONSM6 Smoking clinic 14 15 Smoking status EMISNQEX4 Ex-tobacco chewer 16 Smoking status EMISNQPR396 Primary carer current smoker Smoking status EMISNQSM14 Smoking increased

17 http://bmjopen.bmj.com/ 18 Smoking status EMISNQSN1 Snuff use 19 Smoking status EMISNQTO12 Total time smoked 20 Smoking status EMISNQWA11 Waterpipe tobacco consumption 21 Smoking status EMISOTS1 Other smoking information 22 Smoking status EMISQCO3 Carbon monoxide validation confirms smoker 23 Smoking status EMISQDA1 Date of last cigarette 24 25 Smoking status EMISQGR1 Gradual smoking reduction 26 Smoking status EMISSMRE1 Smoker (Read codes) on September 29, 2021 by guest. Protected copyright. 27 Smoking status PCSDT1DE9 Dependent smoker 28 Smoking status PCSDT1HE1 Heavy smoker review due 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 Page 35 of 36 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from 4 5 6 7 8 9 10 11 12 13 14 15 16 17 For peer review only 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 http://bmjopen.bmj.com/ 39 40 41 42 43 44 45 46 47 on September 29, 2021 by guest. Protected copyright. 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml - only review peer For

BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

60

59

58

57

56

55

54

53

52

51

50

49

48

47

46

45

44

43

42

41

40

39

38

37

36

35

34

33

32

31

30

29

28 27

http://bmjopen.bmj.com/

26

25

24

23

22

21

20 19

on September 29, 2021 by guest. Protected copyright. 18

17 For peer review only review peer For

16

15

14

13

12

11

10

9

8

7

6

5

4

3

2

1

Page 36 of 36 of 36 Page Open BMJ BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

Exposure to sodium channel-inhibiting drugs and cancer survival: protocol for a cohort study using the QResearch primary care database

For peer review only Journal: BMJ Open

Manuscript ID: bmjopen-2014-006604.R1

Article Type: Protocol

Date Submitted by the Author: 24-Sep-2014

Complete List of Authors: Fairhurst, Caroline; University of York, Health Sciences Watt, Ian; University of York, Health Sciences Martin, Fabiola; Hull York Medical School, Bland, Martin; University of York, Health Sciences Brackenbury, William; University of York, Biology

Primary Subject Oncology Heading:

Secondary Subject Heading: Epidemiology, Pharmacology and therapeutics

PRIMARY CARE, Epilepsy < NEUROLOGY, Breast tumours < ONCOLOGY, Keywords: Epidemiology < ONCOLOGY, STATISTICS & RESEARCH METHODS, Clinical trials < THERAPEUTICS http://bmjopen.bmj.com/

on September 29, 2021 by guest. Protected copyright.

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml Page 1 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Exposure to sodium channel-inhibiting drugs and cancer survival: 4 5 protocol for a cohort study using the QResearch primary care database 6 7 8 9 Caroline Fairhurst1, Ian Watt1,2, Fabiola Martin2,3, Martin Bland1, and William 10 11 J. Brackenbury3* 12 13 1Department of Health Sciences, University of York, York, UK 14 15 2HullFor York Medical peer School, York, review UK only 16 17 3Department of Biology, University of York, York, UK 18 19 20 21 *Corresponding author: 22 23 William J. Brackenbury 24 25 Department of Biology, University of York, Heslington, York, YO10 5DD, UK 26 27 E-mail: [email protected] 28 29 Telephone: +44 (0) 1904 328 284 30 31

32 33 Keywords: Anticonvulsants / Breast neoplasms / Colonic neoplasms / Prostatic

34 http://bmjopen.bmj.com/ 35 36 neoplasms / Sodium channels 37 38 39 40 Word count, excluding title page, abstract, references, figures and tables: 2292. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 1 BMJ Open Page 2 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 ABSTRACT 4 5 6 7 Introduction: Metastasis from solid tumours is associated with significant 8 9 morbidity and mortality, and is the leading cause of cancer-related deaths. Voltage- 10 11 gated sodium channels (VGSCs) are drug targets for the treatment of epilepsy. 12 13 VGSCs are also present in cancer cells, where they regulate metastatic cell 14 15 behaviours,For including peer cellular movement review and invasion. Treating only cancer cells with the 16 17 VGSC-inhibiting anticonvulsant phenytoin reduces cellular invasion and migration. 18 19 Together, these suggest that VGSCs may be useful targets for inhibiting metastasis. 20 21 The purpose of this study is to test the hypothesis that use of VGSC-inhibiting drugs 22 23 will reduce metastasis, and therefore increase survival time in cancer patients. 24 25 26 27 Methods and analysis: A cohort study based on primary care data from the 28 29 QResearch database will include patients with one of three common tumours: breast, 30 31 bowel, and prostate. The primary outcome will be overall survival from date of cancer 32 33 diagnosis. Cox proportional hazards regression will be used to compare the survival

34 http://bmjopen.bmj.com/ 35 36 of cancer patients taking VGSC-inhibiting drugs (including anticonvulsants and Class 37 38 I antiarrhythmic agents) with cancer patients not exposed to these drugs, adjusting 39 40 for age and sex. Exposure to VGSC-inhibiting drugs will be defined as having at 41 42 least one prescription for these drugs prior to cancer diagnosis. High and low on September 29, 2021 by guest. Protected copyright. 43 44 exposure groups will be identified based on length of use. A number of sensitivity and 45 46 secondary analyses will be conducted. 47 48 49 50 Ethics and dissemination: The protocol has been independently peer- 51 52 reviewed and approved by the QResearch Scientific Board. The project has also 53 54 been approved by the University of York Ethical Review Process. The results will be 55 56 presented at international conferences and published in an open access peer- 57 58 reviewed journal, in accordance with the STROBE criteria. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 2 Page 3 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 STRENGTHS AND LIMITATIONS OF THIS STUDY 6 7 8 9 • Primary care research data 10 11 • Large sample size and statistical power 12 13 • Planned sensitivity analyses 14 15 • Prescription-basedFor peer study review only 16 17 • No direct information on metastasis, estimation is via overall survival 18 19 20 • Some variables of interest may be missing and/or poor quality in GP data 21 22 23 24 INTRODUCTION 25 26 27 28 Bowel, breast and colon cancer are common cancers, which if diagnosed late have 29 30 often already spread to secondary sites (metastasised). Metastasis is associated with 31 32 significant morbidity and mortality. Metastasis is the leading cause of cancer-related 33

34 deaths[1] because a metastatic cancer is rarely amenable to cure, and interventions http://bmjopen.bmj.com/ 35 36 are largely limited to palliation.[2] Therefore, there is an urgent need to identify and/or 37 38 develop new metastasis prevention strategies. 39 40 41 42

The classical role of voltage-gated sodium channels (VGSCs) is to transmit action on September 29, 2021 by guest. Protected copyright. 43 44 potentials in electrically excitable cells, e.g. neurons and cardiomyocytes.[3] VGSCs 45 46 also regulate neuronal growth and migration.[4-7] Related to these functions, VGSCs 47 48 are clinical targets for a range of disorders, including epilepsy, cardiac arrhythmias, 49 50 neuropathic pain and depression.[8] The mode of action of a number of commonly 51 52 prescribed antiepileptic drugs (anticonvulsants), including phenytoin, lamotrigine, 53 54 carbamazepine and valproate, is to inhibit VGSCs.[9] Similarly, the principal mode of 55 56 action of Class I antiarrhythmic drugs is to inhibit VGSCs.[10] 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 3 BMJ Open Page 4 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 More recently, VGSCs have been identified in cells from a number of major cancers, 6 7 including carcinomas of the breast, prostate and colon.[11 12] In these cells, VGSCs 8 9 promote in vitro cellular behaviours that are associated with metastasis, including 10 11 migration and invasion.[13-18] Overexpression of the VGSC β1 subunit in breast 12 13 cancer cells increases metastasis in mice.[19-21] The VGSC-inhibiting 14 15 anticonvulsantFor phenytoin peer significantly review reduces migration and only invasion of metastatic 16 17 breast and prostate cancer cells in vitro.[22 23] Together, these data suggest that 18 19 VGSCs may be useful targets for anti-metastatic therapy, and that VGSC-inhibiting 20 21 drugs may improve survival from certain cancers.[11 24] Although the effect of 22 23 several anticonvulsants on risk of developing various cancers has been studied 24 25 before (reviewed in[25]), the relationship between VGSC-inhibiting drugs and survival 26 27 28 of cancer patients has not been investigated. 29 30 31 32 The purpose of this study is to test the hypothesis that use of VGSC-inhibiting drugs 33 will predict increased time to metastasis and thus improved survival time in cancer 34 http://bmjopen.bmj.com/ 35 36 patients. The objectives are to investigate: 37 38 • The relationship between use of all VGSC-inhibiting (anticonvulsant and 39 40 Class I antiarrhythmic) drugs and overall survival of cancer patients. We will 41 42 focus on carcinomas of the breast, colon and prostate because they are the on September 29, 2021 by guest. Protected copyright. 43 44 most common and VGSC expression has been extensively studied in these 45 46 tumours.[11 13-17 26-30] 47 48 • The relationship between use of all VGSC-inhibiting drugs and cancer specific 49 50 survival. 51 52 • The relationship between individual VGSC-inhibiting drugs and overall 53 54 survival. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 4 Page 5 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 There are no systematic reviews exploring this area and we are addressing this gap 4 5 by conducting a review concurrent to this study [PROSPERO registration number 6 7 CRD42014013574]. 8 9 10 11 METHODS AND ANALYSIS 12 13 14 15 Data sourceFor and sample peer selection review only 16 17 18 19 This study will use general practice data accessed from QResearch 20 21 (http://www.qresearch.org), a large consolidated database derived from the 22 23 anonymised health records of over 13 million patients from 753 general practices 24 25 (representing around 7% of UK practices). QResearch data are collected from the 26 27 EMIS GP computer system and have been validated using other sources and shown 28 29 to yield similar results to other databases, e.g. the Clinical Practice Research 30 31 Datalink (CPRD).[31 32] QResearch has been used previously to study associations 32 33 between cancer and prescription information.[31]

34 http://bmjopen.bmj.com/ 35 36 37 38 An open cohort of 100,000 patients (aged 30 years or older) with a diagnosis of 39 40 breast, colorectal or prostate cancer will be identified who were registered with a

41 st st 42 QResearch practice during the study period between 1 January 1998 and 31 on September 29, 2021 by guest. Protected copyright. 43 44 December 2013. This will include all those cancer patients in the database who have 45 46 a prescription of one of the index drugs recorded before their date of cancer 47 48 diagnosis (Table 1).[33] The remaining patients will be randomly selected controls. 49 50 Time from date of diagnosis to death will be investigated and data will be right- 51 52 censored in patients who are still alive at the end of the study period. Cancer 53 54 diagnoses will be based on Read code information (available online at 55 56 clinicalcodes.org/medcodes/article/17/). 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 5 BMJ Open Page 6 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Table 1. Voltage-gated Na+ channel-inhibiting drugs 4 5 Drug/derivative Alternative names Classification British National 6 7 Formulary (BNF) 8 9 section[33] 10 11 Carbamazapine, Arbil, Carbagen Anticonvulsant 4.2.3, 4.7.3, 4.8.1 12 13 eslicarbazepine, SR, Epimaz, 14 15 oxcarbazepineFor peerInovelon, Tegretol, review only 16 17 Teril, Timonil, 18 19 Trileptal, Zebinix 20 21 Disopyramide Dirythmin, Isomide, Class Ia 2.3.2 22 23 Rythmodan antiarrhythmic 24 25 26 Flecainide Tambocor Class Ic 2.3.2 27 28 antiarrhythmic 29 30 Lacosamide Vimpat Anticonvulsant 4.8.1 31 32 Lamotrigine Lamictal Anticonvulsant 4.8.1 33

34 Lidocaine Lignocaine, Class Ib 2.3.2, 15.2 http://bmjopen.bmj.com/ 35 36 Xylocard antiarrhythmic 37 38 Mexiletine Mexitil Class Ib 2.3.2 39 40 antiarrhythmic 41 42 Moracizine Ethmozine Class Ic - on September 29, 2021 by guest. Protected copyright. 43 44 antiarrhythmic 45 46 Phenytoin, Epanutin, Pentran Anticonvulsant, 4.7.3, 4.8.1, 4.8.2 47 48 fosphenytoin Class Ib 49 50 antiarrhythmic 51 52 Procainamide Pronestyl Class Ia 2.3.2 53 54 antiarrhythmic 55 56 Propafenone Arythmol Class Ic 2.3.2 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 6 Page 7 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 antiarrhythmic 4 5 Quinidine Kiditard Class Ia - 6 7 antiarrhythmic 8 9 Ranolazine Ranexa Antianginal 2.6.3 10 11 Riluzole Rilutek Treatment for 4.9.3 12 13 amyotrophic lateral 14 15 For peer reviewsclerosis only 16 17 Tocainide Tonocard Class Ib - 18 19 antiarrhythmic 20 21 22 Topiramate Topamax Anticonvulsant 4.7.4, 4.8.1 23 24 Valproic acid, Convulex, Anticonvulsant 4.2.3, 4.7.4, 4.8.1 25 26 sodium valproate Depakote, Epilim, 27 28 Epival, Episenta, 29 30 Orlept 31 32 33

34 Exclusions http://bmjopen.bmj.com/ 35 36 37 38 Temporary residents and patients registered with QResearch within 12 months of 39 40 data extraction will be excluded. Cases without diagnosis of one of the three index 41 42 cancers (breast, colorectal, or prostate cancer) will be excluded. Patients with on September 29, 2021 by guest. Protected copyright. 43 44 anomalous, incorrect or infeasible dates will be excluded, e.g., dates of cancer 45 46 diagnoses recorded before birth or after death. We shall assume that dates of birth 47 48 and death are correct. Any patient with a date of diagnosis that indicates they were 49 50 younger than 25 at the time of diagnosis will be excluded as it is unlikely a person of 51 52 that age would get one of these three cancers. 53 54 55 56 Exposure 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 7 BMJ Open Page 8 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 A participant will be considered as exposed if they have had at least one prescription 6 7 for one of the index drugs. Assuming continuous treatment use between 8 9 prescriptions, we will identify two exposure groups: a low exposure group (less than 6 10 11 months worth of prescriptions) and a moderate to high exposure group (six months or 12 13 more prescriptions). The exposed groups, separately and in combination, will be 14 15 comparedFor with the controlpeer group (cases review without any prescription only for one of the index 16 17 drugs). Patients with one prescription for a drug that would have been used as a local 18 19 anaesthetic, e.g. lidocaine, will be excluded. 20 21 22 23 Outcome measures 24 25 26 27 Metastasis is estimated to be responsible for 90% of deaths from solid tumours.[34] 28 29 However, metastasis itself is not reliably recorded in GP data and so the primary 30 31 outcome measure will be overall survival following cancer diagnosis as a proxy for 32 33 metastasis. Secondary outcome measures will be cancer-specific survival for each

34 http://bmjopen.bmj.com/ 35 36 index type of cancer and overall survival across each drug, numbers permitting. 37 38 39 40 Confounding factors 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 Data on the following confounders will be requested: age, gender, alcohol 45 46 consumption, smoking status, body mass index (BMI) and ethnicity. Data on alcohol, 47 48 smoking and BMI are routinely collected and as such a single patient may have 49 50 multiple recorded observations for these variables assessed over time. We will 51 52 consider the observations measured at the closest date before the date of cancer 53 54 diagnosis, based on appropriate Read codes (available online at 55 56 clinicalcodes.org/medcodes/article/17/). The patients will be categorised as follows: 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 8 Page 9 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 • Alcohol consumption[35] categorised as: Non/trivial drinker (<1unit/day), Light 4 5 drinker (1-2units/day), Moderate-very heavy drinker(3+units/day) and Not 6 7 recorded/known. 8 9 • Smoking status[36] categorised as: Ex-smoker, Smoker, Non-smoker, and 10 11 Not recorded/known. 12 13 • BMI,[37] categorised as Underweight (<18.5), Normal range [18.5-25), 14 15 OverweightFor [25,peer 30), Obese (30+)review and Not recorded/known. only 16 17 • Ethnicity[38] categorised according to the groupings used in the 2011 UK 18 19 census: White; Mixed/Multiple ethnic groups; Asian/Asian British; 20 21 22 Black/African/Caribbean/Black British; Other Ethnic group. We shall also 23 24 include a ‘Not recorded/known’ category. 25 26 27 28 Sample size calculation 29 30 31 32 Up to 100,000 eligible cases will be used, which is the maximum sample size that will 33 be released by QResearch. At breast cancer diagnosis, approximately 6% of patients 34 http://bmjopen.bmj.com/ 35 36 present with metastatic lesions, with bone being the most common site.[39] Of 37 38 patients presenting without bone metastasis at diagnosis, 3.6% subsequently 39 40 develop metastases.[40 41] The majority (90%) of metastases will lead to death.[34] 41 42 Pharmacological blockade of VGSCs inhibits invasion of breast, colorectal and on September 29, 2021 by guest. Protected copyright. 43 44 prostate cancer cells in vitro by 25-50%.[13 15 22 23] Therefore, assuming 3.6% of 45 46 cancer diagnoses lead to a metastasis and most of these to death, with standard 47 48 significance level alpha = 5% and power = 90%, we would require 4248 in the 49 50 exposed group to detect a fall of 25% in the metastasis (or death) rate and 928 to 51 52 detect a fall of 50%. This is based on 20 comparison patients per exposed patient, 53 54 but this ratio is not critical. If we include 6% with a metastasis present at initial 55 56 diagnosis, these numbers fall to 1503 and 330. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 9 BMJ Open Page 10 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 The prevalence of epilepsy is estimated to be 1%.[42] Together, the most commonly 6 7 used VGSC-inhibiting anticonvulsants, phenytoin, lamotrigine, carbamazepine and 8 9 valproate, account for >82% of all antiepileptic drug use.[43] By contrast, Class I 10 11 antiarrhythmic drug use has been considerably less common: <5% in patients with 12 13 cardiac arrhythmia.[44] Thus, using these data as a guide, we might reasonably 14 15 anticipateFor that around peer 0.8% of cancer review patients would be using only one of these VGSC- 16 17 inhibiting drugs. To meet our largest target sample size, 4248, we would therefore be 18 19 looking for a sample that contained 530,000 people with a diagnosis of one of the 20 21 target cancers. To meet the lower target of 928, we would require 116,000 22 23 diagnoses. Given that we are studying deaths rather than metastases per se, we will 24 25 be unable to distinguish between metastases present at diagnosis and detected 26 27 subsequently. Therefore, if we include 6% assumed to have a metastasis present at 28 29 initial diagnosis, we would require 187,875 and 41,250 diagnoses to detect falls in 30 31 metastasis of 25% and 50% respectively. 32 33

34 http://bmjopen.bmj.com/ 35 36 According to Cancer Research UK,[45] the lifetime risk in 2010 for the four major 37 38 cancer sites was almost 13% (female breast), 6% (female lung), 8% (male lung), 6% 39 40 (female bowel including anus), 7% (male bowel including anus) and 13% (prostate). 41 42 Hence for our chosen sites, we expect approximately 21% of women and 20% of on September 29, 2021 by guest. Protected copyright. 43 44 men to experience a positive diagnosis at some time. We will not have lifetime data 45 46 for many in the database, but we might anticipate that 10% of a database sample 47 48 would have a history of one of these sites. Thus the Qresearch database of 13 49 50 million people is large enough to achieve our largest sample target. 51 52 53 54 Statistical analysis 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 10 Page 11 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Analysis will be conducted in Stata v13, using two-sided significance at the 5% level. 4 5 For each Cox model, only the patients with complete data for each of the covariates 6 7 controlled for in the model will be included in the analysis. 8 9 10 11 Descriptive summaries 12 13 14 15 The characteristicsFor ofpeer the comparison review groups will be described only using summary 16 17 statistics. Categorical data will be presented as frequency and percentage, and 18 19 continuous variables will be summarised using descriptive statistics (mean, standard 20 21 deviation, median, 1st and 3rd quartiles, minimum and maximum). The flow of 22 23 patients in the QResearch database will be presented in a diagram. 24 25 26 27 Primary analysis 28 29

30 31 The primary analysis will compare the combined exposure group to the control group. 32 33 For each group, the distribution of time from diagnosis of cancer to death will be

34 http://bmjopen.bmj.com/ 35 36 described using Kaplan-Meier survival estimates. Kaplan-Meier survival curves will 37 38 be presented for the two groups. The statistical equivalence of the two curves will be 39 40 tested using the log-rank test. Right censoring will occur if the patient is still alive at

41 st 42 the end of the study period (31 December 2013). Median time to death, with a 95% on September 29, 2021 by guest. Protected copyright. 43 44 confidence interval (CI) will be presented. If the estimated survivor function is greater 45 46 than 0.5 throughout the study it will not be possible to estimate the median 47 48 survival time and other percentiles survival values (i.e. 90%, 80%, 75%, as 49 50 appropriate) will be presented. 51 52 53 54 We will compare the survival of exposed cases with control cases from time of 55 56 diagnosis of one of the three index cancers using a Cox Proportional Hazards 57 58 regression model. The endpoint will be all cause mortality. We will adjust the Cox 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 11 BMJ Open Page 12 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 model for type of cancer (breast, bowel or prostate), gender and age at diagnosis. 4 5 Age will be included with both a linear and quadratic term (age + age2). We will 6 7 assume that all included patients are receiving the most appropriate standard 8 9 treatment for their disease, so we will not adjust for cancer-treating drug intake. 10 11 Hazard ratios will be presented with p-values and 95% confidence intervals. 12 13 14 15 Cox regressionFor assumes peer that the proportional review hazards model only applies. To assess 16 17 this, we shall plot –log(-log(S(t))) against log(time), where S(t) is the survivor function 18 19 at time t. The curves for the two groups should be parallel. We will also consider a 20 21 chi-squared test of the Schoenfeld residuals to assess the null hypothesis of no 22 23 relationship between the hazards in each group. If the assumptions are not met, we 24 25 shall try to investigate why this is. 26 27 28 29 Sensitivity analysis 30 31

32 33 We will repeat the primary analysis, but adjust the Cox model, in turn, for

34 http://bmjopen.bmj.com/ 35 36 confounding variables: ethnicity, body mass index, smoking and alcohol 37 38 consumption. 39 40 41 42 Secondary analyses on September 29, 2021 by guest. Protected copyright. 43 44 45 46 Each of the following secondary end points will be analysed like the primary outcome 47 48 (unless indicated) with identical censoring strategy: 49 50 • If cancer type proves to be a significant predictor in the primary model then 51 52 we will consider cancer specific survival 53 54 • Survival of Low exposure group compared with control group 55 56 • Survival of High exposure group compared with control group 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 12 Page 13 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 • Survival of combined exposure group and control group with outcome of time 4 5 to death from first diagnosis of any cancer, since some patients may have a 6 7 diagnosis of another cancer before one of breast, bowel or prostate (a 8 9 category for ‘Other’ will be included in the covariate for type of cancer) 10 11 • Survival of patients dependent on the main drug class that they are exposed 12 13 to (numbers permitting). 14 15 For peer review only 16 17 18 ETHICS AND DISSEMINATION 19 20 21 22 This protocol has been independently peer-reviewed by the QResearch Scientific 23 24 Board. It has also been approved by the University of York Ethical Review Process. 25 26 Only the authors will have access to the data during the study, in order to guarantee 27 28 confidentiality of patient information. An article detailing the results of the study will 29 30 be submitted for publication in an international peer-reviewed journal, in accordance 31 32 with the Strengthening the Reporting of Observational Studies in Epidemiology 33

34 (STROBE) criteria.[46] The full statistical analysis will be available from the authors http://bmjopen.bmj.com/ 35 36 after publication of the results. 37 38 39 40 AUTHORS’ CONTRIBUTIONS 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 WB had the original idea for this study. CF and WB wrote the draft of the manuscript. 45 46 IW, FM and MB contributed to the development of the idea, the study design, and 47 48 revised the manuscript. All authors approved the final submitted version of the 49 50 manuscript. 51 52 53 54 FUNDING STATEMENT 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 13 BMJ Open Page 14 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 This work was supported by the Medical Research Council [Fellowship 4 5 G1000508] and the Wellcome Trust [ref: 097829] through the Centre for Chronic 6 7 Diseases and Disorders (C2D2) at the University of York. 8 9 10 11 COMPETING INTERESTS STATEMENT 12 13 14 15 The authorsFor declare thatpeer they have noreview competing interests. only 16 17 18 19 REFERENCES 20 21 22 23 1. Rugo HS. The importance of distant metastases in hormone-sensitive breast 24 cancer. Breast 2008;17 Suppl 1:S3-8 doi: 10.1016/S0960-9776(08)70002- 25 Xpublished Online First: Epub Date]|. 26 2. Suva LJ, Griffin RJ, Makhoul I. Mechanisms of bone metastases of breast cancer. 27 Endocr Relat Cancer 2009;16(3):703-13 doi: 10.1677/ERC-09-0012published 28 Online First: Epub Date]|. 29 3. Hille B. Ionic channels of excitable membranes. 2nd ed. Sunderland 30 (Massachusetts): Sinauer Associates Inc., 1992. 31 4. Brackenbury WJ, Calhoun JD, Chen C, et al. Functional reciprocity between Na+ 32 channel Nav1.6 and β1 subunits in the coordinated regulation of excitability 33 and neurite outgrowth. Proc Natl Acad Sci U S A 2010;107(5):2283-88

34 http://bmjopen.bmj.com/ 35 5. Brackenbury WJ, Yuan Y, O'Malley HA, Parent JM, Isom LL. Abnormal neuronal 36 patterning occurs during early postnatal brain development of Scn1b-null 37 mice and precedes hyperexcitability. Proc Natl Acad Sci U S A 38 2013;110(3):1089-94 doi: 10.1073/pnas.1208767110published Online First: 39 Epub Date]|. 40 6. Catterall WA. From ionic currents to molecular mechanisms: the structure and function of voltage-gated sodium channels. Neuron 2000;26(1):13-25 41 + 42 7. Brackenbury WJ, Davis TH, Chen C, et al. Voltage-gated Na channel β1 subunit- on September 29, 2021 by guest. Protected copyright. 43 mediated neurite outgrowth requires fyn kinase and contributes to central 44 nervous system development in vivo. J Neurosci 2008;28(12):3246-56 45 8. Clare JJ, Tate SN, Nobbs M, Romanos MA. Voltage-gated sodium channels as 46 therapeutic targets. Drug Discov Today 2000;5(11):506-20 47 9. Mantegazza M, Curia G, Biagini G, Ragsdale DS, Avoli M. Voltage-gated sodium 48 channels as therapeutic targets in epilepsy and other neurological disorders. 49 Lancet Neurol 2010;9(4):413-24 doi: 10.1016/S1474-4422(10)70059- 50 4published Online First: Epub Date]|. 51 10. Vaughan Williams EM. The relevance of cellular to clinical electrophysiology in 52 classifying antiarrhythmic actions. J Cardiovasc Pharmacol 1992;20 Suppl 53 2:S1-7 54 11. Brackenbury WJ. Voltage-gated sodium channels and metastatic disease. 55 Channels (Austin) 2012;6(5):352-61 doi: 10.4161/chan.21910published 56 Online First: Epub Date]|. 57 12. Brackenbury WJ, Djamgoz MB, Isom LL. An emerging role for voltage-gated Na+ 58 channels in cellular migration: regulation of central nervous system 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 14 Page 15 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 development and potentiation of invasive cancers. Neuroscientist 4 2008;14(6):571-83 doi: 10.1177/1073858408320293published Online First: 5 Epub Date]|. 6 13. House CD, Vaske CJ, Schwartz A, et al. Voltage-gated Na+ channel SCN5A is a 7 key regulator of a gene transcriptional network that controls colon cancer 8 invasion. Cancer Res 2010;70(17):6957-67 doi: 10.1158/0008-5472.CAN-10- 9 1169published Online First: Epub Date]|. 10 14. Grimes JA, Fraser SP, Stephens GJ, et al. Differential expression of voltage- 11 activated Na+ currents in two prostatic tumour cell lines: contribution to 12 invasiveness in vitro. FEBS Lett 1995;369(2-3):290-4 13 15. Fraser SP, Diss JK, Chioni AM, et al. Voltage-gated sodium channel expression 14 and potentiation of human breast cancer metastasis. Clin Cancer Res 15 2005;For11(15):5381-89 peer doi: 10.1158/1078-0432.CCR-05-0327published review only Online 16 First: Epub Date]|. 17 16. Brackenbury WJ, Chioni AM, Diss JK, Djamgoz MB. The neonatal splice variant 18 of Nav1.5 potentiates in vitro metastatic behaviour of MDA-MB-231 human 19 breast cancer cells. Breast Cancer Res. Treat. 2007;101(2):149-60 20 17. Brackenbury WJ, Djamgoz MB. Activity-dependent regulation of voltage-gated 21 Na+ channel expression in Mat-LyLu rat prostate cancer cell line. J Physiol 22 2006;573(Pt 2):343-56 doi: 10.1113/jphysiol.2006.106906published Online 23 First: Epub Date]|. 24 18. Brackenbury WJ, Djamgoz MB. Nerve growth factor enhances voltage-gated Na+ 25 channel activity and Transwell migration in Mat-LyLu rat prostate cancer cell 26 line. J Cell Physiol 2007;210(3):602-8 doi: 10.1002/jcp.20846published Online 27 First: Epub Date]|. 28 19. Nelson M, Millican-Slater R, Forrest LC, Brackenbury WJ. The sodium channel 29 beta1 subunit mediates outgrowth of neurite-like processes on breast cancer 30 cells and promotes tumour growth and metastasis. Int J Cancer 31 2014;135(10):2338-51 doi: 10.1002/ijc.28890published Online First: Epub 32 Date]|. 33 20. Brackenbury WJ, Isom LL. Na Channel beta Subunits: Overachievers of the Ion

34 http://bmjopen.bmj.com/ 35 Channel Family. Frontiers in pharmacology 2011;2:53 doi: 36 10.3389/fphar.2011.00053published Online First: Epub Date]|. 37 21. Chioni AM, Brackenbury WJ, Calhoun JD, Isom LL, Djamgoz MB. A novel 38 adhesion molecule in human breast cancer cells: voltage-gated Na+ channel 39 beta1 subunit. Int J Biochem Cell Biol 2009;41(5):1216-27 doi: 40 10.1016/j.biocel.2008.11.001published Online First: Epub Date]|. 22. Fraser SP, Salvador V, Manning EA, et al. Contribution of functional voltage- 41 + 42 gated Na channel expression to cell behaviors involved in the metastatic on September 29, 2021 by guest. Protected copyright. 43 cascade in rat prostate cancer: I. lateral motility. J Cell Physiol 44 2003;195(3):479-87 45 23. Yang M, Kozminski DJ, Wold LA, et al. Therapeutic potential for phenytoin: 46 targeting Na(v)1.5 sodium channels to reduce migration and invasion in 47 metastatic breast cancer. Breast Cancer Res Treat 2012;134(2):603-15 doi: 48 10.1007/s10549-012-2102-9published Online First: Epub Date]|. 49 24. Brackenbury WJ, Isom LL. Voltage-gated Na+ channels: potential for beta 50 subunits as therapeutic targets. Expert Opin Ther Targets 2008;12(9):1191- 51 203 doi: 10.1517/14728222.12.9.1191published Online First: Epub Date]|. 52 25. Singh G, Driever PH, Sander JW. Cancer risk in people with epilepsy: the role of 53 antiepileptic drugs. Brain 2005;128(Pt 1):7-17 doi: 54 10.1093/brain/awh363published Online First: Epub Date]|. 55 26. Brisson L, Driffort V, Benoist L, et al. NaV1.5 Na(+) channels allosterically 56 regulate the NHE-1 exchanger and promote the activity of breast cancer cell 57 invadopodia. J Cell Sci 2013;126(Pt 21):4835-42 doi: 58 10.1242/jcs.123901published Online First: Epub Date]|. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 15 BMJ Open Page 16 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 27. Ding L, Ellis MJ, Li S, et al. Genome remodelling in a basal-like breast cancer 4 metastasis and xenograft. Nature 2010;464(7291):999-1005 doi: 5 10.1038/nature08989published Online First: Epub Date]|. 6 28. Grimes JA, Djamgoz MB. Electrophysiological characterization of voltage-gated 7 Na(+) current expressed in the highly metastatic Mat-LyLu cell line of rat 8 prostate cancer. J Cell Physiol 1998;175(1):50-8 9 29. Gillet L, Roger S, Besson P, et al. Voltage-gated Sodium Channel Activity 10 Promotes Cysteine Cathepsin-dependent Invasiveness and Colony Growth of 11 Human Cancer Cells. J Biol Chem 2009;284(13):8680-91 doi: M806891200 12 [pii] 10.1074/jbc.M806891200published Online First: Epub Date]|. 13 30. Ding Y, Brackenbury WJ, Onganer PU, et al. Epidermal growth factor upregulates 14 motility of Mat-LyLu rat prostate cancer cells partially via voltage-gated Na+ 15 channelFor activity. peer J Cell Physiol review 2008;215(1):77-81 doi:only 16 10.1002/jcp.21289published Online First: Epub Date]|. 17 31. Vinogradova Y, Coupland C, Hippisley-Cox J. Exposure to bisphosphonates and 18 risk of gastrointestinal cancers: series of nested case-control studies with 19 QResearch and CPRD data. BMJ (Clinical research ed.) 2013;346:f114 doi: 20 10.1136/bmj.f114published Online First: Epub Date]|. 21 32. Reeves D, Springate DA, Ashcroft DM, et al. Can analyses of electronic patient 22 records be independently and externally validated? The effect of statins on 23 the mortality of patients with ischaemic heart disease: a cohort study with 24 nested case-control analysis. BMJ open 2014;4(4):e004952 doi: 25 10.1136/bmjopen-2014-004952published Online First: Epub Date]|. 26 33. British National Formulary London: British Medical Association and Royal 27 Pharmaceutical Society of Great Britain, 2013. 28 34. Gupta GP, Massague J. Cancer metastasis: building a framework. Cell 29 2006;127(4):679-95 doi: 10.1016/j.cell.2006.11.001published Online First: 30 Epub Date]|. 31 35. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans. 32 Volume 44 Alcohol drinking, 1988. 33 36. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans.

34 http://bmjopen.bmj.com/ 35 Volume 83 Tobacco smoke and involuntary smoking, 2004. 36 37. Reeves GK, Pirie K, Beral V, Green J, Spencer E, Bull D. Cancer incidence and 37 mortality in relation to body mass index in the Million Women Study: cohort 38 study. BMJ (Clinical research ed.) 2007;335(7630):1134 doi: 39 10.1136/bmj.39367.495995.AEpublished Online First: Epub Date]|. 40 38. National Cancer Intelligence Network and Cancer Research UK. Cancer 41 Incidence and Survival by Major Ethnic Group, England 2002-2006., 2009. 42 39. Coleman RE, Rubens RD. The clinical course of bone metastases from breast on September 29, 2021 by guest. Protected copyright. 43 cancer. Br J Cancer 1987;55(1):61-6 44 40. Yong M, Jensen AO, Jacobsen JB, Norgaard M, Fryzek JP, Sorensen HT. 45 Survival in breast cancer patients with bone metastases and skeletal-related 46 events: a population-based cohort study in Denmark (1999-2007). Breast 47 Cancer Res Treat 2011;129(2):495-503 doi: 10.1007/s10549-011-1475- 48 5published Online First: Epub Date]|. 49 41. Jensen AO, Jacobsen JB, Norgaard M, Yong M, Fryzek JP, Sorensen HT. 50 Incidence of bone metastases and skeletal-related events in breast cancer 51 patients: a population-based cohort study in Denmark. BMC Cancer 52 2011;11:29 doi: 10.1186/1471-2407-11-29published Online First: Epub 53 Date]|. 54 42. Joint Epilepsy Council of the UK and Ireland. Epilepsy prevalence, incidence and 55 other statistics. 56 http://www.epilepsyscotland.org.uk/pdf/Joint_Epilepsy_Council_Prevalence_ 57 and_Incidence_September_11_(3).pdf 2011 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 16 Page 17 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 43. Nicholas JM, Ridsdale L, Richardson MP, Ashworth M, Gulliford MC. Trends in 4 antiepileptic drug utilisation in UK primary care 1993-2008: cohort study using 5 the General Practice Research Database. Seizure : the journal of the British 6 Epilepsy Association 2012;21(6):466-70 doi: 7 10.1016/j.seizure.2012.04.014published Online First: Epub Date]|. 8 44. Fang MC, Stafford RS, Ruskin JN, Singer DE. National trends in antiarrhythmic 9 and antithrombotic medication use in atrial fibrillation. Arch Intern Med 10 2004;164(1):55-60 doi: 10.1001/archinte.164.1.55published Online First: 11 Epub Date]|. 12 45. Cancer Research UK. Lifetime risk of cancer. 2014. 13 http://www.cancerresearchuk.org/cancer- 14 info/cancerstats/incidence/risk/statistics-on-the-risk-of-developing-cancer - 15 Lifetime5)For. peer review only 16 46. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. 17 Strengthening the Reporting of Observational Studies in Epidemiology 18 (STROBE) statement: guidelines for reporting observational studies. BMJ 19 20 (Clinical research ed.) 2007;335(7624):806-8 doi: 21 10.1136/bmj.39335.541782.ADpublished Online First: Epub Date]|. 22 23 24 25 26 27 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 17 BMJ Open Page 18 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Exposure to sodium channel-inhibiting drugs and cancer survival: 4 5 protocol for a cohort study using the QResearch primary care database 6 7 8 9 Caroline Fairhurst1, Ian Watt1,2, Fabiola Martin2,3, Martin Bland1, and William 10 11 J. Brackenbury3* 12 13 1Department of Health Sciences, University of York, York, UK 14 15 2HullFor York Medical peer School, York, review UK only 16 17 3Department of Biology, University of York, York, UK 18 19 20 21 *Corresponding author: 22 23 William J. Brackenbury 24 25 Department of Biology, University of York, Heslington, York, YO10 5DD, UK 26 27 Email: [email protected] 28 29 Telephone: +44 (0) 1904 328 284 30 31

32 33 Keywords: Anticonvulsants / Breast neoplasms / Colonic neoplasms / Prostatic

34 http://bmjopen.bmj.com/ 35 36 neoplasms / Sodium channels 37 38 39 40 Word count, excluding title page, abstract, references, figures and tables: 2292. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 1 Page 19 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 ABSTRACT 4 5 6 7 Introduction: Metastasis from solid tumours is associated with significant 8 9 morbidity and mortality, and is the leading cause of cancerrelated deaths. Voltage 10 11 gated sodium channels (VGSCs) are drug targets for the treatment of epilepsy. 12 13 VGSCs are also present in cancer cells, where they regulate metastatic cell 14 15 behaviours,For including peer cellular movement review and invasion. Treating only cancer cells with the 16 17 VGSCinhibiting anticonvulsant phenytoin reduces cellular invasion and migration. 18 19 Together, these suggest that VGSCs may be useful targets for inhibiting metastasis. 20 21 The purpose of this study is to test the hypothesis that use of VGSCinhibiting drugs 22 23 will reduce metastasis, and therefore increase survival time in cancer patients. 24 25 26 27 Methods and analysis: A cohort study based on primary care data from the 28 29 QResearch database will include patients with one of three common tumours: breast, 30 31 bowel, and prostate. The primary outcome will be overall survival from date of cancer 32 33 diagnosis. Cox proportional hazards regression will be used to compare the survival

34 http://bmjopen.bmj.com/ 35 36 of cancer patients taking VGSCinhibiting drugs (including anticonvulsants and Class 37 38 I antiarrhythmic agents) with cancer patients not exposed to these drugs, adjusting 39 40 for age and sex. Exposure to VGSCinhibiting drugs will be defined as having at 41 42 least one prescription for these drugs prior to cancer diagnosis. High and low on September 29, 2021 by guest. Protected copyright. 43 44 exposure groups will be identified based on length of use. A number of sensitivity and 45 46 secondary analyses will be conducted. 47 48 49 50 Ethics and dissemination: The protocol has been independently peer 51 52 reviewed and approved by the QResearch Scientific Board. The project has also 53 54 been approved by the University of York Ethical Review Process. The results will be 55 56 presented at international conferences and published in an open access peer 57 58 reviewed journal, in accordance with the STROBE criteria. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 2 BMJ Open Page 20 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 STRENGTHS AND LIMITATIONS OF THIS STUDY 6 7 8 9 • Primary care research data 10 11 • Large sample size and statistical power 12 13 • Planned sensitivity analyses 14 15 • PrescriptionbasedFor peer study review only 16 17 • No direct information on metastasis, estimation is via overall survival 18 19 20 • Some variables of interest may be missing and/or poor quality in GP data 21 22 23 24 INTRODUCTION 25 26 27 28 Bowel, breast and colon cancer are common cancers, which if diagnosed late have 29 30 often already spread to secondary sites (metastasised). Metastasis is associated with 31 32 significant morbidity and mortality. Metastasis is the leading cause of cancerrelated 33

34 deaths[1] because a metastatic cancer is rarely amenable to cure, and interventions http://bmjopen.bmj.com/ 35 36 are largely limited to palliation.[2] Therefore, there is an urgent need to identify and/or 37 38 develop new metastasis prevention strategies. 39 40 41 42

The classical role of voltagegated sodium channels (VGSCs) is to transmit action on September 29, 2021 by guest. Protected copyright. 43 44 potentials in electrically excitable cells, e.g. neurons and cardiomyocytes.[3] VGSCs 45 46 also regulate neuronal growth and migration.[47] Related to these functions, VGSCs 47 48 are clinical targets for a range of disorders, including epilepsy, cardiac arrhythmias, 49 50 neuropathic pain and depression.[8] The mode of action of a number of commonly 51 52 prescribed antiepileptic drugs (anticonvulsants), including phenytoin, lamotrigine, 53 54 carbamazepine and valproate, is to inhibit VGSCs.[9] Similarly, the principal mode of 55 56 action of Class I antiarrhythmic drugs is to inhibit VGSCs.[10] 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 3 Page 21 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 More recently, VGSCs have been identified in cells from a number of major cancers, 6 7 including carcinomas of the breast, prostate and colon.[11 12] In these cells, VGSCs 8 9 promote in vitro cellular behaviours that are associated with metastasis, including 10 11 migration and invasion.[1318] Overexpression of the VGSC β1 subunit in breast 12 13 cancer cells increases metastasis in mice.[1921] The VGSCinhibiting 14 15 anticonvulsantFor phenytoin peer significantly review reduces migration and only invasion of metastatic 16 17 breast and prostate cancer cells in vitro.[22 23] and has been shown to reduce 18 19 metastasis in a mouse model of breast cancer (M. Nelson, M. Yang, A. A. Dowle, J. 20 21 R. Thomas and W. J. Brackenbury, manuscript under review, 2014). Together, these 22 23 data suggest that VGSCs may be useful targets for antimetastatic therapy, and that 24 25 VGSCinhibiting drugs may improve survival from certain cancers.[11 24] Although 26 27 28 the effect of several anticonvulsants on risk of developing various cancers has been 29 30 studied before (reviewed in[25]), the relationship between VGSCinhibiting drugs and 31 32 survival of cancer patients has not been investigated. 33

34 http://bmjopen.bmj.com/ 35 36 The purpose of this study is to test the hypothesis that use of VGSCinhibiting drugs 37 38 will predict increased time to metastasis and thus improved survival time in cancer 39 40 patients. The objectives are to investigate: 41 42 • The relationship between use of all VGSCinhibiting (anticonvulsant and on September 29, 2021 by guest. Protected copyright. 43 44 Class I antiarrhythmic) drugs and overall survival of cancer patients. We will 45 46 focus on carcinomas of the breast, colon and prostate because they are the 47 48 most common and VGSC expression has been extensively studied in these 49 50 tumours.[11 1317 2630] 51 52 • The relationship between use of all VGSCinhibiting drugs and cancer specific 53 54 survival. 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 4 BMJ Open Page 22 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 • The relationship between individual VGSCinhibiting drugs and overall 4 5 survival. 6 7 8 9 There are no systematic reviews exploring this area and we are addressing this gap 10 11 by conducting a review concurrent to this study [PROSPERO registration number 12 13 CRD42014013574]. 14 15 For peer review only 16 17 METHODS AND ANALYSIS 18 19 20 21 Data source and sample selection 22 23 24 25 This study will use general practice data accessed from QResearch 26 27 (http://www.qresearch.org), a large consolidated database derived from the 28 29 anonymised health records of over 13 million patients from 753 general practices 30 31 32 (representing around 7% of UK practices). QResearch data are collected from the 33 EMIS GP computer system and have been validated using other sources and shown

34 http://bmjopen.bmj.com/ 35 36 to yield similar results to other databases, e.g. the Clinical Practice Research 37 38 Datalink (CPRD).[31 32] QResearch has been used previously to study associations 39 40 between cancer and prescription information.[31] 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 An open cohort of 100,000 patients (aged 30 years or older) with a diagnosis of 45 46 breast, colorectal or prostate cancer will be identified who were registered with a 47 st st 48 QResearch practice during the study period between 1 January 1998 and 31 49 50 December 2013. This will include all those cancer patients in the database who have 51 52 a prescription of one of the index drugs recorded before their date of cancer 53 54 diagnosis (Table 1).[33] The remaining patients will be randomly selected controls. 55 56 Time from date of diagnosis to death will be investigated and data will be right 57 58 censored in patients who are still alive at the end of the study period. Cancer 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 5 Page 23 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 diagnoses will be based on Read code information (see online supplementary 4 5 appendixavailable online at clinicalcodes.org/medcodes/article/17/). 6 7 8 9 Table 1. Voltagegated Na+ channelinhibiting drugs 10 11 Drug/derivative Alternative names Classification British National 12 13 Formulary (BNF) 14 15 For peer review onlysection[33] 16 17 Carbamazapine, Arbil, Carbagen Anticonvulsant 4.2.3, 4.7.3, 4.8.1 18 19 eslicarbazepine, SR, Epimaz, 20 21 oxcarbazepine Inovelon, Tegretol, 22 23 Teril, Timonil, 24 25 Trileptal, Zebinix 26 27 28 Disopyramide Dirythmin, Isomide, Class Ia 2.3.2 29 30 Rythmodan antiarrhythmic 31 32 Flecainide Tambocor Class Ic 2.3.2 33 antiarrhythmic 34 http://bmjopen.bmj.com/ 35 36 Lacosamide Vimpat Anticonvulsant 4.8.1 37 38 Lamotrigine Lamictal Anticonvulsant 4.8.1 39 40 Lidocaine Lignocaine, Class Ib 2.3.2, 15.2 41 42 Xylocard antiarrhythmic on September 29, 2021 by guest. Protected copyright. 43 44 Mexiletine Mexitil Class Ib 2.3.2 45 46 antiarrhythmic 47 48 Moracizine Ethmozine Class Ic 49 50 antiarrhythmic 51 52 Phenytoin, Epanutin, Pentran Anticonvulsant, 4.7.3, 4.8.1, 4.8.2 53 54 fosphenytoin Class Ib 55 56 antiarrhythmic 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 6 BMJ Open Page 24 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 Procainamide Pronestyl Class Ia 2.3.2 4 5 antiarrhythmic 6 7 Propafenone Arythmol Class Ic 2.3.2 8 9 antiarrhythmic 10 11 Quinidine Kiditard Class Ia 12 13 antiarrhythmic 14 15 RanolazineFor peer Ranexa reviewAntianginal only 2.6.3 16 17 Riluzole Rilutek Treatment for 4.9.3 18 19 amyotrophic lateral 20 21 sclerosis 22 23 24 Tocainide Tonocard Class Ib 25 26 antiarrhythmic 27 28 Topiramate Topamax Anticonvulsant 4.7.4, 4.8.1 29 30 Valproic acid, Convulex, Anticonvulsant 4.2.3, 4.7.4, 4.8.1 31 32 sodium valproate Depakote, Epilim, 33

34 Epival, Episenta, http://bmjopen.bmj.com/ 35 36 Orlept 37 38 39 40 Exclusions 41 42

on September 29, 2021 by guest. Protected copyright. 43 44 Temporary residents and patients registered with QResearch within 12 months of 45 46 data extraction will be excluded. Cases without diagnosis of one of the three index 47 48 cancers (breast, colorectal, or prostate cancer) will be excluded. Patients with 49 50 anomalous, incorrect or infeasible dates will be excluded, e.g., dates of cancer 51 52 diagnoses recorded before birth or after death. We shall assume that dates of birth 53 54 and death are correct. Any patient with a date of diagnosis that indicates they were 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 7 Page 25 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 younger than 25 at the time of diagnosis will be excluded as it is unlikely a person of 4 5 that age would get one of these three cancers. 6 7 8 9 Exposure 10 11 12 13 A participant will be considered as exposed if they have had at least one prescription 14 15 for one ofFor the index drugs.peer Assuming review continuous treatment onlyuse between 16 17 prescriptions, we will identify two exposure groups: a low exposure group (less than 6 18 19 months worth of prescriptions) and a moderate to high exposure group (six months or 20 21 more prescriptions). The exposed groups, separately and in combination, will be 22 23 compared with the control group (cases without any prescription for one of the index 24 25 drugs). Patients with one prescription for a drug that would have been used as a local 26 27 anaesthetic, e.g. lidocaine, will be excluded. 28 29

30 31 Outcome measures 32 33

34 http://bmjopen.bmj.com/ 35 36 Metastasis is estimated to be responsible for 90% of deaths from solid tumours.[34] 37 38 However, metastasis itself is not reliably recorded in GP data and so the primary 39 40 outcome measure will be overall survival following cancer diagnosis as a proxy for 41 42 metastasis. Secondary outcome measures will be cancerspecific survival for each on September 29, 2021 by guest. Protected copyright. 43 44 index type of cancer and overall survival across each drug, numbers permitting. 45 46 47 48 Confounding factors 49 50 51 52 Data on the following confounders will be requested: age, gender, alcohol 53 54 consumption, smoking status, body mass index (BMI) and ethnicity. Data on alcohol, 55 56 smoking and BMI are routinely collected and as such a single patient may have 57 58 multiple recorded observations for these variables assessed over time. We will 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 8 BMJ Open Page 26 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 consider the observations measured at the closest date before the date of cancer 4 5 diagnosis, based on appropriate Read codes (available online at 6 7 clinicalcodes.org/medcodes/article/17/detailed in online supplementary appendix). 8 9 The patients will be categorised as follows: 10 11 • Alcohol consumption[35] categorised as: Non/trivial drinker (<1unit/day), Light 12 13 drinker (12units/day), Moderatevery heavy drinker(3+units/day) and Not 14 15 recorded/known.For peer review only 16 17 • Smoking status[36] categorised as: Exsmoker, Smoker, Nonsmoker, and 18 19 Not recorded/known. 20 21 • BMI,[37] categorised as Underweight (<18.5), Normal range [18.525), 22 23 Overweight [25, 30), Obese (30+) and Not recorded/known. 24 25 26 • Ethnicity[38] categorised according to the groupings used in the 2011 UK 27 28 census: White; Mixed/Multiple ethnic groups; Asian/Asian British; 29 30 Black/African/Caribbean/Black British; Other Ethnic group. We shall also 31 32 include a ‘Not recorded/known’ category. 33

34 http://bmjopen.bmj.com/ 35 36 Sample size calculation 37 38 39 40 Up to 100,000 eligible cases will be used, which is the maximum sample size that will 41 42 be released by QResearch. At breast cancer diagnosis, approximately 6% of patients on September 29, 2021 by guest. Protected copyright. 43 44 present with metastatic lesions, with bone being the most common site.[39] Of 45 46 patients presenting without bone metastasis at diagnosis, 3.6% subsequently 47 48 develop metastases.[40 41] The majority (90%) of metastases will lead to death.[34] 49 50 Pharmacological blockade of VGSCs inhibits invasion of breast, colorectal and 51 52 prostate cancer cells in vitro by 2550%.[13 15 22 23] Therefore, assuming 3.6% of 53 54 cancer diagnoses lead to a metastasis and most of these to death, with standard 55 56 significance level alpha = 5% and power = 90%, we would require 4248 in the 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 9 Page 27 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 exposed group to detect a fall of 25% in the metastasis (or death) rate and 928 to 4 5 detect a fall of 50%. This is based on 20 comparison patients per exposed patient, 6 7 but this ratio is not critical. If we include 6% with a metastasis present at initial 8 9 diagnosis, these numbers fall to 1503 and 330. 10 11 12 13 The prevalence of epilepsy is estimated to be 1%.[42] Together, the most commonly 14 15 used VGSCinhibitingFor peer anticonvulsants, review phenytoin, lamotrigine, only carbamazepine and 16 17 valproate, account for >82% of all antiepileptic drug use.[43] By contrast, Class I 18 19 antiarrhythmic drug use has been considerably less common: <5% in patients with 20 21 cardiac arrhythmia.[44] Thus, using these data as a guide, we might reasonably 22 23 anticipate that around 0.8% of cancer patients would be using one of these VGSC 24 25 inhibiting drugs. To meet our largest target sample size, 4248, we would therefore be 26 27 looking for a sample that contained 530,000 people with a diagnosis of one of the 28 29 target cancers. To meet the lower target of 928, we would require 116,000 30 31 diagnoses. Given that we are studying deaths rather than metastases per se, we will 32 33 be unable to distinguish between metastases present at diagnosis and detected

34 http://bmjopen.bmj.com/ 35 36 subsequently. Therefore, if we include 6% assumed to have a metastasis present at 37 38 initial diagnosis, we would require 187,875 and 41,250 diagnoses to detect falls in 39 40 metastasis of 25% and 50% respectively. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 According to Cancer Research UK,[45] the lifetime risk in 2010 for the four major 45 46 cancer sites was almost 13% (female breast), 6% (female lung), 8% (male lung), 6% 47 48 (female bowel including anus), 7% (male bowel including anus) and 13% (prostate). 49 50 Hence for our chosen sites, we expect approximately 21% of women and 20% of 51 52 men to experience a positive diagnosis at some time. We will not have lifetime data 53 54 for many in the database, but we might anticipate that 10% of a database sample 55 56 would have a history of one of these sites. Thus the Qresearch database of 13 57 58 million people is large enough to achieve our largest sample target. 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 10 BMJ Open Page 28 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 4 5 Statistical analysis 6 7 8 9 Analysis will be conducted in Stata v13, using twosided significance at the 5% level. 10 11 For each Cox model, only the patients with complete data for each of the covariates 12 13 controlled for in the model will be included in the analysis. 14 15 For peer review only 16 17 Descriptive summaries 18 19 20 21 The characteristics of the comparison groups will be described using summary 22 23 statistics. Categorical data will be presented as frequency and percentage, and 24 25 continuous variables will be summarised using descriptive statistics (mean, standard 26 27 deviation, median, 1st and 3rd quartiles, minimum and maximum). The flow of 28 29 patients in the QResearch database will be presented in a diagram. 30 31

32 33 Primary analysis

34 http://bmjopen.bmj.com/ 35 36 37 38 The primary analysis will compare the combined exposure group to the control group. 39 40 For each group, the distribution of time from diagnosis of cancer to death will be 41 42 described using KaplanMeier survival estimates. KaplanMeier survival curves will on September 29, 2021 by guest. Protected copyright. 43 44 be presented for the two groups. The statistical equivalence of the two curves will be 45 46 tested using the logrank test. Right censoring will occur if the patient is still alive at 47 st 48 the end of the study period (31 December 2013). Median time to death, with a 95% 49 50 confidence interval (CI) will be presented. If the estimated survivor function is greater 51 52 than 0.5 throughout the study it will not be possible to estimate the median 53 54 survival time and other percentiles survival values (i.e. 90%, 80%, 75%, as 55 56 appropriate) will be presented. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 11 Page 29 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 We will compare the survival of exposed cases with control cases from time of 4 5 diagnosis of one of the three index cancers using a Cox Proportional Hazards 6 7 regression model. The endpoint will be all cause mortality. We will adjust the Cox 8 9 model for type of cancer (breast, bowel or prostate), gender and age at diagnosis. 10 11 Age will be included with both a linear and quadratic term (age + age2). We will 12 13 assume that all included patients are receiving the most appropriate standard 14 15 treatmentFor for their disease, peer so we will review not adjust for cancertreating only drug intake. 16 17 Hazard ratios will be presented with pvalues and 95% confidence intervals. 18 19 20 21 Cox regression assumes that the proportional hazards model applies. To assess 22 23 this, we shall plot –log(log(S(t))) against log(time), where S(t) is the survivor function 24 25 at time t. The curves for the two groups should be parallel. We will also consider a 26 27 chisquared test of the Schoenfeld residuals to assess the null hypothesis of no 28 29 relationship between the hazards in each group. If the assumptions are not met, we 30 31 shall try to investigate why this is. 32 33

34 http://bmjopen.bmj.com/ 35 36 Sensitivity analysis 37 38 39 40 We will repeat the primary analysis, but adjust the Cox model, in turn, for 41 42 confounding variables: ethnicity, body mass index, smoking and alcohol on September 29, 2021 by guest. Protected copyright. 43 44 consumption. 45 46 47 48 Secondary analyses 49 50 51 52 Each of the following secondary end points will be analysed like the primary outcome 53 54 (unless indicated) with identical censoring strategy: 55 56 • If cancer type proves to be a significant predictor in the primary model then 57 58 we will consider cancer specific survival 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 12 BMJ Open Page 30 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 • Survival of Low exposure group compared with control group 4 5 • Survival of High exposure group compared with control group 6 7 • Survival of combined exposure group and control group with outcome of time 8 9 to death from first diagnosis of any cancer, since some patients may have a 10 11 diagnosis of another cancer before one of breast, bowel or prostate (a 12 13 category for ‘Other’ will be included in the covariate for type of cancer) 14 15 • SurvivalFor of patients peer dependent review on the main drug classonly that they are exposed 16 17 to (numbers permitting). 18 19 20 21 22 ETHICS AND DISSEMINATION 23 24 25 26 This protocol has been independently peerreviewed by the QResearch Scientific 27 28 Board. It has also been approved by the University of York Ethical Review Process. 29 30 Only the authors will have access to the data during the study, in order to guarantee 31 32 confidentiality of patient information. An article detailing the results of the study will 33

34 be submitted for publication in an international peerreviewed journal, in accordance http://bmjopen.bmj.com/ 35 36 with the Strengthening the Reporting of Observational Studies in Epidemiology 37 38 (STROBE) criteria.[46] The full protocol and statistical analysis will be available from 39 40 the authors after publication of the results. 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 AUTHORS’ CONTRIBUTIONS 45 46 47 48 WB had the original idea for this study. CF and WB wrote the draft of the manuscript. 49 50 IW, FM and MB contributed to the development of the idea, the study design, and 51 52 revised the manuscript. All authors approved the final submitted version of the 53 54 manuscript. 55 56

57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 13 Page 31 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 FUNDING STATEMENT 4 5 6 7 This work was supported by the Medical Research Council [Fellowship 8 9 G1000508] and the Wellcome Trust [ref: 097829] through the Centre for Chronic 10 11 Diseases and Disorders (C2D2) at the University of York. 12 13 14 15 COMPETINGFor peer INTERESTS STATEMENTreview only 16 17 18 19 The authors declare that they have no competing interests. 20 21 22 23 REFERENCES 24 25 26 27 1. Rugo HS. The importance of distant metastases in hormonesensitive breast 28 cancer. Breast 2008;17 Suppl 1:S38 doi: 10.1016/S09609776(08)70002 29 Xpublished Online First: Epub Date]|. 30 2. Suva LJ, Griffin RJ, Makhoul I. Mechanisms of bone metastases of breast cancer. 31 Endocr Relat Cancer 2009;16(3):70313 doi: 10.1677/ERC090012published 32 Online First: Epub Date]|. 33 3. Hille B. Ionic channels of excitable membranes. 2nd ed. Sunderland

34 http://bmjopen.bmj.com/ 35 (Massachusetts): Sinauer Associates Inc., 1992. 36 4. Brackenbury WJ, Calhoun JD, Chen C, et al. Functional reciprocity between Na+ 37 channel Nav1.6 and β1 subunits in the coordinated regulation of excitability 38 and neurite outgrowth. Proc Natl Acad Sci U S A 2010;107(5):228388 39 5. Brackenbury WJ, Yuan Y, O'Malley HA, Parent JM, Isom LL. Abnormal neuronal 40 patterning occurs during early postnatal brain development of Scn1bnull 41 mice and precedes hyperexcitability. Proc Natl Acad Sci U S A 42 2013;110(3):108994 doi: 10.1073/pnas.1208767110published Online First: on September 29, 2021 by guest. Protected copyright. 43 Epub Date]|. 44 6. Catterall WA. From ionic currents to molecular mechanisms: the structure and 45 function of voltagegated sodium channels. Neuron 2000;26(1):1325 + 46 7. Brackenbury WJ, Davis TH, Chen C, et al. Voltagegated Na channel β1 subunit 47 mediated neurite outgrowth requires fyn kinase and contributes to central 48 nervous system development in vivo. J Neurosci 2008;28(12):324656 49 8. Clare JJ, Tate SN, Nobbs M, Romanos MA. Voltagegated sodium channels as 50 therapeutic targets. Drug Discov Today 2000;5(11):50620 51 9. Mantegazza M, Curia G, Biagini G, Ragsdale DS, Avoli M. Voltagegated sodium 52 channels as therapeutic targets in epilepsy and other neurological disorders. 53 Lancet Neurol 2010;9(4):41324 doi: 10.1016/S14744422(10)70059 54 4published Online First: Epub Date]|. 55 10. Vaughan Williams EM. The relevance of cellular to clinical electrophysiology in 56 classifying antiarrhythmic actions. J Cardiovasc Pharmacol 1992;20 Suppl 57 2:S17 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 14 BMJ Open Page 32 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 11. Brackenbury WJ. Voltagegated sodium channels and metastatic disease. 4 Channels (Austin) 2012;6(5):35261 doi: 10.4161/chan.21910published 5 Online First: Epub Date]|. 6 12. Brackenbury WJ, Djamgoz MB, Isom LL. An emerging role for voltagegated Na+ 7 channels in cellular migration: regulation of central nervous system 8 development and potentiation of invasive cancers. Neuroscientist 9 2008;14(6):57183 doi: 10.1177/1073858408320293published Online First: 10 Epub Date]|. 11 13. House CD, Vaske CJ, Schwartz A, et al. Voltagegated Na+ channel SCN5A is a 12 key regulator of a gene transcriptional network that controls colon cancer 13 invasion. Cancer Res 2010;70(17):695767 doi: 10.1158/00085472.CAN10 14 1169published Online First: Epub Date]|. 15 14. GrimesFor JA, Fraser peer SP, Stephens review GJ, et al. Differential expressiononly of voltage 16 activated Na+ currents in two prostatic tumour cell lines: contribution to 17 invasiveness in vitro. FEBS Lett 1995;369(23):2904 18 15. Fraser SP, Diss JK, Chioni AM, et al. Voltagegated sodium channel expression 19 and potentiation of human breast cancer metastasis. Clin Cancer Res 20 2005;11(15):538189 doi: 10.1158/10780432.CCR050327published Online 21 First: Epub Date]|. 22 16. Brackenbury WJ, Chioni AM, Diss JK, Djamgoz MB. The neonatal splice variant 23 of Nav1.5 potentiates in vitro metastatic behaviour of MDAMB231 human 24 breast cancer cells. Breast Cancer Res. Treat. 2007;101(2):14960 25 17. Brackenbury WJ, Djamgoz MB. Activitydependent regulation of voltagegated 26 Na+ channel expression in MatLyLu rat prostate cancer cell line. J Physiol 27 2006;573(Pt 2):34356 doi: 10.1113/jphysiol.2006.106906published Online 28 First: Epub Date]|. 29 18. Brackenbury WJ, Djamgoz MB. Nerve growth factor enhances voltagegated Na+ 30 channel activity and Transwell migration in MatLyLu rat prostate cancer cell 31 line. J Cell Physiol 2007;210(3):6028 doi: 10.1002/jcp.20846published Online 32 First: Epub Date]|. 33 19. Nelson M, MillicanSlater R, Forrest LC, Brackenbury WJ. The sodium channel

34 http://bmjopen.bmj.com/ 35 beta1 subunit mediates outgrowth of neuritelike processes on breast cancer 36 cells and promotes tumour growth and metastasis. Int J Cancer 37 2014;135(10):233851 doi: 10.1002/ijc.28890published Online First: Epub 38 Date]|. 39 20. Brackenbury WJ, Isom LL. Na Channel beta Subunits: Overachievers of the Ion 40 Channel Family. Frontiers in pharmacology 2011;2:53 doi: 41 10.3389/fphar.2011.00053published Online First: Epub Date]|. 42 21. Chioni AM, Brackenbury WJ, Calhoun JD, Isom LL, Djamgoz MB. A novel on September 29, 2021 by guest. Protected copyright. 43 adhesion molecule in human breast cancer cells: voltagegated Na+ channel 44 beta1 subunit. Int J Biochem Cell Biol 2009;41(5):121627 doi: 45 10.1016/j.biocel.2008.11.001published Online First: Epub Date]|. 46 22. Fraser SP, Salvador V, Manning EA, et al. Contribution of functional voltage + 47 gated Na channel expression to cell behaviors involved in the metastatic 48 cascade in rat prostate cancer: I. lateral motility. J Cell Physiol 49 2003;195(3):47987 50 23. Yang M, Kozminski DJ, Wold LA, et al. Therapeutic potential for phenytoin: 51 targeting Na(v)1.5 sodium channels to reduce migration and invasion in 52 metastatic breast cancer. Breast Cancer Res Treat 2012;134(2):60315 doi: 53 10.1007/s1054901221029published Online First: Epub Date]|. 54 24. Brackenbury WJ, Isom LL. Voltagegated Na+ channels: potential for beta 55 subunits as therapeutic targets. Expert Opin Ther Targets 2008;12(9):1191 56 203 doi: 10.1517/14728222.12.9.1191published Online First: Epub Date]|. 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 15 Page 33 of 36 BMJ Open BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 25. Singh G, Driever PH, Sander JW. Cancer risk in people with epilepsy: the role of 4 antiepileptic drugs. Brain 2005;128(Pt 1):717 doi: 5 10.1093/brain/awh363published Online First: Epub Date]|. 6 26. Brisson L, Driffort V, Benoist L, et al. NaV1.5 Na(+) channels allosterically 7 regulate the NHE1 exchanger and promote the activity of breast cancer cell 8 invadopodia. J Cell Sci 2013;126(Pt 21):483542 doi: 9 10.1242/jcs.123901published Online First: Epub Date]|. 10 27. Ding L, Ellis MJ, Li S, et al. Genome remodelling in a basallike breast cancer 11 metastasis and xenograft. Nature 2010;464(7291):9991005 doi: 12 10.1038/nature08989published Online First: Epub Date]|. 13 28. Grimes JA, Djamgoz MB. Electrophysiological characterization of voltagegated 14 Na(+) current expressed in the highly metastatic MatLyLu cell line of rat 15 prostateFor cancer. peer J Cell Physiol review 1998;175(1):508 only 16 29. Gillet L, Roger S, Besson P, et al. Voltagegated Sodium Channel Activity 17 Promotes Cysteine Cathepsindependent Invasiveness and Colony Growth of 18 Human Cancer Cells. J Biol Chem 2009;284(13):868091 doi: M806891200 19 [pii] 10.1074/jbc.M806891200published Online First: Epub Date]|. 20 30. Ding Y, Brackenbury WJ, Onganer PU, et al. Epidermal growth factor upregulates 21 motility of MatLyLu rat prostate cancer cells partially via voltagegated Na+ 22 channel activity. J Cell Physiol 2008;215(1):7781 doi: 23 10.1002/jcp.21289published Online First: Epub Date]|. 24 31. Vinogradova Y, Coupland C, HippisleyCox J. Exposure to bisphosphonates and 25 risk of gastrointestinal cancers: series of nested casecontrol studies with 26 QResearch and CPRD data. BMJ (Clinical research ed.) 2013;346:f114 doi: 27 10.1136/bmj.f114published Online First: Epub Date]|. 28 32. Reeves D, Springate DA, Ashcroft DM, et al. Can analyses of electronic patient 29 records be independently and externally validated? The effect of statins on 30 the mortality of patients with ischaemic heart disease: a cohort study with 31 nested casecontrol analysis. BMJ open 2014;4(4):e004952 doi: 32 10.1136/bmjopen2014004952published Online First: Epub Date]|. 33 33. British National Formulary London: British Medical Association and Royal

34 http://bmjopen.bmj.com/ 35 Pharmaceutical Society of Great Britain, 2013. 36 34. Gupta GP, Massague J. Cancer metastasis: building a framework. Cell 37 2006;127(4):67995 doi: 10.1016/j.cell.2006.11.001published Online First: 38 Epub Date]|. 39 35. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans. 40 Volume 44 Alcohol drinking, 1988. 41 36. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans. 42 Volume 83 Tobacco smoke and involuntary smoking, 2004. on September 29, 2021 by guest. Protected copyright. 43 37. Reeves GK, Pirie K, Beral V, Green J, Spencer E, Bull D. Cancer incidence and 44 mortality in relation to body mass index in the Million Women Study: cohort 45 study. BMJ (Clinical research ed.) 2007;335(7630):1134 doi: 46 10.1136/bmj.39367.495995.AEpublished Online First: Epub Date]|. 47 38. National Cancer Intelligence Network and Cancer Research UK. Cancer 48 Incidence and Survival by Major Ethnic Group, England 20022006., 2009. 49 39. Coleman RE, Rubens RD. The clinical course of bone metastases from breast 50 cancer. Br J Cancer 1987;55(1):616 51 40. Yong M, Jensen AO, Jacobsen JB, Norgaard M, Fryzek JP, Sorensen HT. 52 Survival in breast cancer patients with bone metastases and skeletalrelated 53 events: a populationbased cohort study in Denmark (19992007). Breast 54 Cancer Res Treat 2011;129(2):495503 doi: 10.1007/s105490111475 55 5published Online First: Epub Date]|. 56 41. Jensen AO, Jacobsen JB, Norgaard M, Yong M, Fryzek JP, Sorensen HT. 57 Incidence of bone metastases and skeletalrelated events in breast cancer 58 patients: a populationbased cohort study in Denmark. BMC Cancer 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 16 BMJ Open Page 34 of 36 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

1 2 3 2011;11:29 doi: 10.1186/147124071129published Online First: Epub 4 Date]|. 5 42. Joint Epilepsy Council of the UK and Ireland. Epilepsy prevalence, incidence and 6 other statistics. 7 http://www.epilepsyscotland.org.uk/pdf/Joint_Epilepsy_Council_Prevalence_ 8 and_Incidence_September_11_(3).pdf 2011 9 43. Nicholas JM, Ridsdale L, Richardson MP, Ashworth M, Gulliford MC. Trends in 10 antiepileptic drug utilisation in UK primary care 19932008: cohort study using 11 the General Practice Research Database. Seizure : the journal of the British 12 Epilepsy Association 2012;21(6):46670 doi: 13 10.1016/j.seizure.2012.04.014published Online First: Epub Date]|. 14 44. Fang MC, Stafford RS, Ruskin JN, Singer DE. National trends in antiarrhythmic 15 andFor antithrombotic peer medication review use in atrial fibrillation. only Arch Intern Med 16 2004;164(1):5560 doi: 10.1001/archinte.164.1.55published Online First: 17 Epub Date]|. 18 45. Cancer Research UK. Lifetime risk of cancer. 2014. 19 http://www.cancerresearchuk.org/cancer- 20 21 info/cancerstats/incidence/risk/statistics-on-the-risk-of-developing-cancer - 22 Lifetime5). 23 46. von Elm E, Altman DG, Egger M, Pocock SJ, Gotzsche PC, Vandenbroucke JP. 24 Strengthening the Reporting of Observational Studies in Epidemiology 25 (STROBE) statement: guidelines for reporting observational studies. BMJ 26 (Clinical research ed.) 2007;335(7624):8068 doi: 27 10.1136/bmj.39335.541782.ADpublished Online First: Epub Date]|. 28 29 30 31 32 33

34 http://bmjopen.bmj.com/ 35 36 37 38 39 40 41 42 on September 29, 2021 by guest. Protected copyright. 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml 17 Page 35 of 36 BMJ Open

1 2

3 BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from 4 5 6 7 8 9 10 11 12 13 14 15 16 17 For peer review only 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 http://bmjopen.bmj.com/ 39 40 41 42 43 44 45 46 47 on September 29, 2021 by guest. Protected copyright. 48 49 50 51 52 53 54 55 56 57 58 59 60

For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml For peer review only - http://bmjopen.bmj.com/site/about/guidelines.xhtml - only review peer For

BMJ Open: first published as 10.1136/bmjopen-2014-006604 on 14 November 2014. Downloaded from

60

59

58

57

56

55

54

53

52

51

50

49

48

47

46

45

44

43

42

41

40

39

38

37

36

35

34

33

32

31

30

29

28 27

http://bmjopen.bmj.com/

26

25

24

23

22

21

20 19

on September 29, 2021 by guest. Protected copyright. 18

17 For peer review only review peer For

16

15

14

13

12

11

10

9

8

7

6

5

4

3

2

1

Page 36 of 36 of 36 Page Open BMJ