Objectives Precision medicine requires an in-depth understanding of cancer genomics and a rigorous approach to tumour tissue evaluation. Accounting for tumour heterogeneity, which is a feature of most cancers, is essential when assessing somatic mutations. Genomic instability produces multiple tumour clones within a given tumour tissue. Each clone has different characteristics (e.g. histology, growth, genetic mutations) and ultimately, can have different clinical outcomes. The challenge in oncology is to work out which clone is the greatest threat to life and target treatment appropriately. Current approaches to biomarker testing focus on interpatient heterogeneity but not intra-tumour heterogeneity.

We aim to evaluate how somatic mutation studies report and validate tumour tissue samples and whether they consider tumour heterogeneity. Secondly, we aim to search the literature to identify existing guidance on tumour heterogeneity.

Method We used a set of nine studies (12 datasets) taken from a recent systematic review on the prevalence of somatic gene mutations in cancer. We assessed whether the authors of the studies reported on:

1. Tissue sampling:

   1. the clarity of reporting for sampling and pathology methods

   2. whether the pathology of the sample or the patient is presented.

2. Tumour heterogeneity or clonality:

   1. the use of multiple samples from the same patient;

   2. the investigation of clonality or heterogeneity.

3. Tumour content (purity):

   1. the use of microdissection;

   2. the reporting of the tumour content of the sample.

We searched MEDLINE, Pubmed and Embase and seven guideline- and systematic review-specific databases from inception to January 2019 to identify guidelines reporting on tumour heterogeneity.

Results Seven (58%) datasets did not report sufficient information with regards to tumour sampling and pathology to confirm that the tumour sample represented the patient’s pathology. The use of this information in a systematic review would introduce uncertainty.

None of the datasets reported multiple prevalence results per sample. The use of this finding in a review would illustrate an evidence gap that tumour heterogeneity was not investigated only patient heterogeneity.

Ten datasets (83%) reported measuring the tumour content of the sample; seven (58%) reported that the tumour content could be as low as 10%. The use of this finding in a review would illustrate that the samples are at risk of contamination and the results cannot be considered reliable.

Results for searching of guidelines will be presented at the conference.

Conclusions Systematic reviewers need to be much more cautious about the use of patient derived tissue samples. It cannot be assumed that a tumour sample replicates the pathology of the patient, or that somatic tumour samples are homogeneous for a given biomarker or mutation. Systematic reviewers need to assess whether a) the study reports sufficient information to confirm that the sample reflects the pathology of interest b) the sample is free of normal, benign or other contaminant cells c) tumour heterogeneity has been considered. The assessment of tumour samples and tumour heterogeneity is important for systematic reviews of epidemiology, diagnostic and prognostic studies.