Predicting Incidental Gallbladder Cancer

This is a cross-sectional study of prospectively identified (identified in advance) patients undergoing routine gallbladder surgery in the United Kingdom (UK). It will be undertaken in two phases and has an adaptive design. This means that the results of the first phase could affect the design of the second phase.

The first phase includes a parallel, interview based study which will explore the risk of incidental Gallbladder Cancer (iGBC) and establish whether Selective Histological Analysis (where gallbladders are chosen for post-operative tests based on their individual risk of iGBC) is acceptable to a representative cohort.

The primary objective of this study is to develop and validate a diagnostic scoring system for iGBC in UK patients undergoing routine gallbladder surgery. There are two similar scores in existence, however they have been conducted in a way which means that their data may be skewed, because they were conducted in countries which already use selective histological analysis and have very different pathways to cholecystectomy when compared to the UK. The investigators therefore need to build a new, UK dataset, and the most reliable way to do this is to intentionally collect all of the required data from a new set of patients. Often this can be done using patients who have already been treated - called retrospective (looking backwards) data collection. However, to be a high-quality dataset requires each patient to consistently have a complete set of data, and the investigators know that retrospective studies often lead to missing data points. Additionally, intraoperative findings are always recorded on an operation note, but not always in a consistent way. By collecting data on patients who will be treated during a future study period (prospective data collection), the investigators can ask for these findings to be recorded in a way which is standardised, so they can be reliably compared between surgeons and hospitals. The investigators have therefore elected to undertake a prospective study of patients undergoing routine gallbladder surgery.

To identify risk factors for a disease, a sample size which would be expected to give rise to ten "events", or diagnoses, of interest is typically felt to be required. As iGBC is thought to be present in about one in every 500 gallbladder operations, this gives a sample size of 5,000 patients to identify risk factors. However, to develop a model of the factors which cause the disease, from which a diagnostic scoring system can be produced, the expected number of diagnostic parameters needed, which may be included in the scoring system, has to be considered. A diagnostic parameter is a variable, for example the biological sex of a patient, and each variable is counted at least once. When a variable has more than two potential options, for example a patient's ethnicity, each additional option is counted as an extra parameter. The previous diagnostic scores have used 3-4 diagnostic parameters, however they have lacked the nuance required of the test. Several of the known and suspected risk factors have more than one potential option, and after discussion, it was felt that allowing for 20 diagnostic parameters was realistic. With their statistician, the investigators have calculated a minimum number of fully completed patient records of just under 21,000. It is unrealistic to expect that every record will be fully completed, and in line with convention the investigators have planned for 80% of records to be fully completed. They provisionally, therefore, expect to require 26,245 patient records to achieve statistical power and will provisionally target 30,000 to allow expansion as the result of necessary variation from expected findings which may become evident after the feasibility phase.

Attempting to conduct a study of up to 30,000 patients without proof of concept could result in a lot of effort without success. Therefore, the investigators have designed a two-stage study. The first stage will test the feasibility of the whole study and will consist of a pilot, or "dry run" of the proposed methodology. This dry run will be conducted across 8 surgical units, from within 7 organisations, and over 6 months will collect data from as many as 500 patients. This will allow the investigators to assess whether they can recruit enough patients, quickly enough from the number of participating sites which are involved to be able to conduct the full study within the proposed timeframe. It will also establish whether the data which is being collected is sufficiently complete (i.e. high quality) to be useful if collected from all the patients needed to develop the diagnostic score. Finally, the investigators will be able to confirm whether the population they are testing has rates of incidental cancer and other changes which are in line with published and previously observed rates. Alongside this pilot study, the investigators will conduct an interview based study to establish how patients and relevant professionals think about the risk of incidental cancer in routine cholecystectomy, and what the participants perceive the barriers to and acceptability of selective histological analysis is.

After the information from the feasibility phase has been collected and analysed, the outcomes will be assessed and any necessary adaptations made to the study. The study will only continue beyond this "feasibility" phase if the outcomes show that it can successfully collect the required data, in the required time frame. This assessment will be made based on how complete the data returned is, how quickly patients were recruited to the study, whether there have been difficulties collecting information on important risk factors (and whether those difficulties can be resolved) and, importantly, whether the investigators are able to obtain enough funding to collect the data.

Patients in either stage of the main study will be identified on or before the day of surgery. All of the information which will be collected about them will normally be gathered in the course of routine care and can be used even when fully de-identified. Patients will not undergo any changes to their normal care as the study will simply observe details about them, their operation, their tests, and the result of their histopathology test. As the study requires so many patients, it is felt that consenting them will be both highly impracticable and prohibitively expensive. The data will therefore be collected without specific patient consent, in line with current guidance, but measures will be taken to inform the patients that the study is ongoing. As the patients do not need to do anything for the study, the burden on them, or risk to them will be minimal. A trainee collaborative of surgical doctors will collect the data before and after the operation and will enter it into a specially designed data collection tool on a secure platform called REDCap. The REDCap data collection tool allows de-identified information to be transferred safely between the participating hospital and the central research team at University Hospitals Plymouth NHS Trust. Data entered will regularly be reviewed by a researcher from the central team of investigators to identify any outlier data points early, in order that they can be reviewed before data collection ends. In order to further confirm the integrity of the data being collected, separate trainees will be recruited by the principal investigator for each site to act as "data validators". These trainees will be asked to input duplicate information on a random selection of the patients recruited at their hospital. This reassessment of the patient records will help to confirm whether the data being inputted is sufficiently reliable to be used for the development of a diagnostic test. For the feasibility phase, 10% of patients will be reassessed, but the percentage of patients undergoing further evaluation could be reduced in the full study if initial data validation is reassuring.

During the feasibility phase, a doctoral student (who is also a surgical trainee at University Hospitals Plymouth NHS Trust) will conduct an interview based study, as mentioned previously. Patients and professional participants will be identified in participating sites or by direct approach and will be invited to contact the student using the study email address. Alternatively, at the potential participant's request, the clinician who identified them may make the initial contact on their behalf. The potential participant will then be provided with information about the study to consider, and will have a (usually telephone) conversation with an investigator about what is required of them and what will happen to the information which the participant provides. The participant will be invited to sign a consent form, which will preferably be e-consent (but will have the same content as the paper consent) before they undertake an interview. The interview will either happen face-to-face or online, and will be recorded. The recording will be transcribed (turned into a written document) and the transcript will be de-identified. These transcripts will be analysed using a "pragmatic methodology" in specialist software. The analysis will break the comments of the participant up into coded phrases which are then organised into "themes", which can then be analysed between participants, and participant groups.

Finally, after the feasibility stage is completed, assuming that it shows that the full study can progress, further funding and site recruitment will be undertaken. The full study will be carried out much as the pilot study will be carried out. It is important that the pilot study replicates the full study as closely as possible, so that the rates of patient recruitment and data completeness are likely to be replicated throughout. After the full study has been completed, the data generated (including from the pilot study, unless major alterations have been required) will be analysed in depth, to look for risk factors for incidental gallbladder cancer. Significant risk factors will be combined to make a model of the predictive (diagnostic) factors for incidental gallbladder cancer, and finally a score which could predict the risk of incidental cancer could be produced. This will be provisionally checked (validated) against the dataset as "proof" of its quality.