CUP RFA: Development of systematic review methodology for mechanistic studies

(Co Principal Investigator is Professor Richard Martin)

  • Topic: Cancer-related outcomes
  • Institution: University of Bristol
  • Country: United Kingdom
  • Status: Completed

Scientific abstract

(View plain language abstract)

Background

Many laboratory experiments are performed to identify causal pathways and in doing so inform human health. These mechanistic studies complement epidemiological findings and can offer insights into biological plausibility and pathways between exposure and disease. Systematic reviews are the most robust way to synthesise data which have addressed a common question. Methods for conducting and reporting rigorous systematic reviews of epidemiological studies are well established. However, such methods are lacking for mechanistic studies.

Methods

A multidisciplinary team with expertise in informatics, statistics, epidemiology, systematic reviews, cancer biology and nutrition was assembled and a series of 5 one-day workshops took place involving presentations, group work and discussions, along with smaller meetings and research being carried out in the intervening periods in order to develop a framework for carry-out systematic reviews of studies underlying mechanisms of exposures and cancer. This framework was tested by carrying out a review to investigate the Insulin-like growth factor (IGF) system as a biological mediator of the milk-prostate cancer link.

Results

We have developed a framework for carrying out rigorous systematic reviews of mechanistic studies, which includes guidance on: a two stage search strategy (the first stage of which is an automated mechanisms discovery search (https://www.temmpo.org.uk/), followed by a targeted search for studies on a specific mechanism), formulating a research question, applying inclusion/exclusion criteria, assessing the relevance of retrieved studies to the research question, assessing the quality of individual studies (using appropriate risk of bias tools), synthesizing the data from individual studies, assessing the strength of the overall body of evidence from human and animal studies separately and integrating the human and animal studies to reach a conclusion.

In the process of carrying out this project we have developed a novel tool for simultaneously viewing, visualizing and prioritizing mechanisms based on the number of publications linking and exposure to an intermediate phenotype and the number of papers linking the same intermediate phenotype to the exposure of interest (TeMMPo). We have also developed a new type of graph for data from epidemiological studies, which addresses a similar question, but which is not sufficiently similar to meta-analyse; this has been termed an albatross plot.

The results from our systematic review of milk, insulin like growth factor (IGF) and prostate cancer suggest that there is a positive association between IGF-I and PCa risk and comparatively a negative association between IGFBP-3 and PCa risk (excluding advanced PCa risk). Our results reflect those of the previous meta-analysis (Rowlands 2012); however the strength of the associations seen in our current study appear to be smaller.  There was less evidence that IGF-II, IGFBP-1 and IGFBP-2 were associated with PCa risk; very few studies were available for all three peptides and combined estimates had wide confidence intervals. More rigorous risk of bias protocols in our current study have ensured that the quality of the data presented is high and that confounding factors (with particular focus on age) have been accounted for. This may account for differences between results from this study and the Rowlands 2012 meta-analysis, providing a refined and updated assessment of the IGF-PCa association.

Conclusion

The above framework will be available to researchers in the future who wish to conduct robust systematic reviews of the mechanisms which underpin associations between exposures and cancer. We found moderate evidence that milk increases IGF-I levels and some evidence that IGF-I levels increase prostate cancer risk.


Plain language abstract

Background

Many experiments are performed on human, animals and in cells to identify processes leading to disease and in doing so inform human health. These studies complement findings from observational studies in humans and can offer insights into the biology of a disease and pathways between exposure and disease. A review of the current literature using robust methods to identify all previous studies are the best way to collate data which have addressed a common question. Methods for conducting and reporting rigorous systematic reviews of observational studies in humans are well established. However, such methods are lacking for other studies.

Aims & Objectives

The aim of this study was to develop a set of guidelines for collating data from all studies which may provide evidence on the mechanism by which an exposure causes cancer, and to test this by exploring the hypothesis that milk causes prostate cancer via the Insulin-like Growth Factor (IGF) pathway. 

How the study was carried out

A large team with expertise in a broad range of relevant subjects was assembled and a series of 5 one-day workshops took place involving presentations, group work and discussions, along with smaller meetings and research being carried out in the intervening periods. The guidelines were tested by carrying a review of studies on the association between Milk, IGF (a type of growth hormone) and prostate cancer.

Key findings and conclusions

We have developed a template for carrying out rigorous reviews of studies which inform on the pathways to cancer. We have used our template to show that there is some evidence that milk causes cancer by stimulating a type of growth hormone (IGF). This template will be available for researchers to use for similar reviews in the future.