This document outlines a project to develop a standardized methodology for conducting systematic reviews of mechanistic cancer studies. The goal is to enable more rigorous synthesis of evidence from animal and cell line studies on how dietary factors may influence cancer risk. An international team of experts will hold workshops to design a comprehensive search strategy, quality assessment criteria, and data extraction methods. These will be tested by reviewing evidence on potential mechanisms linking milk consumption to prostate cancer risk, such as effects on hormone and growth factor levels. The resulting review methodology template will provide guidance for systematically evaluating mechanistic evidence across study types.
2. Aim
• The aim of this project is to draw on the expertise
(in systematic reviews of epidemiological studies
and in experimental studies of cancer) in our group
in order to develop and publish a template for
carrying out rigorous systematic reviews of
mechanistic studies.
3. Why is this important?
Systematic reviews are the most rigorous way to synthesise studies
which address a common question.
Methods for conducting and reporting rigorous systematic reviews
remain lacking for mechanistic studies.
But there is a huge amount of published information on the
mechanisms between dietary factors and different types of cancer
from animal and cell line studies.
This will enable more systematic reviews of mechanistic studies, thus
allowing researchers to determine the strength of evidence for a
particular nutrients and cancer risk and also to identify gaps in the
research.
4. The Team
University of Bristol
• PI- Dr Sarah Lewis –Genetic epidemiology/systematic reviews of genetic studies
• Co-PI- Prof Richard Martin –Epidemiology/systematic reviews
• Dr Mona Jeffreys- Cancer Epidemiology/systematic reviews
• Dr Mike Gardner – Animal biology/systematic reviews
• Prof Jeff Holly- Molecular biology – IGF and cancer
• Dr Tom Gaunt – Genetic epidemiology/bioinformatics
• Prof Jonathan Sterne- Meta-analysis and systematic review methodology
• Professor Julian Higgins – Meta-analysis and systematic review methodology
• Prof George Davey Smith – Epidemiology
• Prof Christos Paraskeva –Molecular biology
• Prof Steve Thomas –Epidemiology of head and neck cancer.
• Dr Pauline Emmett - Nutritional epidemiology
• Dr Kate Northstone – Nutritional Epidemiology
• Cath Borwick – Librarian/ Search strategies
University of Cambridge WCRF
Dr Suzanne Turner- Animal models Prof Martin Wiseman
Dr Pangiota Mitrou
International Agency for Research on Cancer Dr Rachel Thompson
Dr Sabina Rinaldi- Hormones and cancer Faye Butler
5. Study objectives
The key objectives which will be addressed in this project are:
• To design a comprehensive search strategy to identify the diverse study
types of relevance.
• To determine which types of study should be included and determine a
hierarchy of evidence.
• To develop quality control criteria and produce a data extraction form to
capture this information.
• To develop methods to identify and quantify publication bias.
• To test our draft methodology (1 to 4 above) in a feasibility study, in which
we will review the evidence on mechanisms underlying associations
between milk and prostate cancer.
6. Hypothesis to be explored:
High milk consumption is a risk factor for prostate
cancer
• Milk consumption has been implicated as a risk factor for prostate cancer
• But milk consumption is measured semi-quantitatively in some studies, with large
differences between individuals in the same group and therefore subject to
attenuation by errors. In addition, milk intake is susceptible to confounding by
other diet and lifestyle factors.
• Systematic reviews of observational studies remain inconclusive (WCRF Second
Expert Report 2007 showed limited/suggestive evidence for milk as a risk factor for
prostate cancer)
• Experimental studies have been carried-out but not taken into account in most
systematic reviews
7. 3 possible mechanisms
• High calcium intake in milk may suppress the conversion of 25(OH) vitamin
D to 1,25(OH)2 vitamin D, which has antiproliferative and differential
effects on human prostate cancer cells.
• Milk is a rich source of estrogens. Animal models have shown that
estrogen and testosterone act in synergy to initiate prostate cancer.
• Insulin-like growth factor I (IGF-I) level is positively associated with
prostate cancer risk. Milk has been shown to increase IGF levels.
• We plan to investigate the evidence for the 3 mechanisms above (and also
to identify other potential mechanisms) using our devised protocol for
systematic reviews of mechanistic studies of diet and cancer.
8. Analytical approach
• A series of 4 workshops to call on experts within our group –
using a mixture of presentations with discussion, small group
exercises, round table discussions.
• On going searches, and development of methods, feedback to
members of the team
• Regular meetings between PIs and research associate
9. At the end of this workshop we will produce:
• A list of study types to be included in the mechanistic review
• A hierarchy dependent on relevance and strength of evidence which can
be drawn from the studies
• A search strategy.
Workshop 1 – 23rd April
10. Workshop 2
At this workshop we will :
• Present the search strategy and results for our chosen hypothesis on milk and
prostate cancer (see below) for discussion
• Produce a list of quality control criteria, by study type and draft inclusion/exclusion
criteria
• Agree on a list of variables which will be collected to assess between-study
heterogeneity, and which will be used to design the data extraction form
• Determine limits to apply to the search strategy, using the hierarchy of studies and
the quality control criteria, and possibly date of publication to ensure that the
search yields a manageable number of publications
11. Workshop 3
At this workshop we will:
• Present a draft data extraction form, inclusion/exclusion criteria, effect of applying
limits, quality control and summary data, along with the results of statistical tests
for publication bias for the review of milk and prostate cancer
• Consider how to analyse and display results. For the most part we envisage that
studies will be too heterogeneous to meta-analyse but where possible we will do
this in order to summarise the data
• Consider the overall potential for publication bias in mechanistic studies and how
to deal with this
12. Workshop 4
• We present the results of our systematic review of mechanistic studies of
milk and cancer, we will discuss any issues arising during this systematic
review and use these to agree on any amendments to the final template.
13. Search strategy
• Epidemiological review – search strategy developed for
nutrients + cancer (WCRF, 2007).
• We will adopt this strategy, but will have an additional step in
our search, as many mechanistic studies will not have cancer
as an outcome but may still be relevant.
• We will develop a search strategy to find studies on
“Nutrient” + “Mechanistic target”
“Nutrient” + “Cancer”
14. Verifying the search strategy
• We will test our search against any thorough
systematic reviews of animal or other mechanistic
studies relating to diet and cancer, to determine
whether we find the same papers.
• We will check reference lists of retrieved papers
• We will run searches on a selection of authors
working in this field and cross check their papers
against our own search.
15. Mechanistic targets
• 1.Genomic instability
• 2. DNA damage/repair (specify)
• 3. Gene mutation (specify)
• 4. Epigenetics (specify)
• 5. Gene expression/miRNA changes (specify)
• 6. Serum hormones/growth factors (specify)
• 7. Urinary or tissue metabolite
• 8. Inflammation
• a. Cytokines (specify)
• b. Tissue markers (specify)
• c. Histological classification (specify)
• 9. Cell proliferation markers (specify)
• 10. Immunologic effects (specify)
• 11. Invasion/metastasis process (specify)
• 12. Angiogenesis effects (specify)
• 13. Cell signaling effects (specify)
• 14. Cell energetics effects (specify)
• 15. Differentiation (specify)
• 16. Cell death (apoptosis/autophagy/necrosis)
• 17. Replicative potential/ Stem cell enrichment
DNA damage
DNA fragmentation
DNA adducts
DNA breaks – chromosome
breakage
Mechanisms Protocol Development Group
16. Types of study found so far
Milk + exp DNA damage (NOT Lactation, also excluded
Cross-Sectional Studies, Cohort Studies and Case-Control
Studies as MESH terms) = 52 studies
Breakdown of study type
11 human
2 human cross-over
15 animal (rats and mice)
20 human cells
4 animal cells
17. different mechanistic targets will yield different
amounts and types of data- Searches to date on Milk
and Mechanistic target give the following numbers of
studies:
• Genomic Instability 4
• DNA Damage 52
• DNA Repair 7
• Mutation 480
• Somatomedins (Insulin-Like Growth Factor I, Insulin-
Like Growth Factor II) 238
• Growth Hormone 376
• Cell Proliferation 577
18. Timetable
• Recruit research associate to work on the project – Jan 2013 - Dr Mike Gardner
recruited and currently working on the project.
• RA will start work on initial exploratory searches – Mar/April 2013
• Organise and hold first workshop – April 2013
• Second workshop – Sept 2013
• Third workshop –Jan 2014
• Fourth workshop –June 2014
• In between workshops we will refine the protocol, carry-out searches, investigate
QC criteria, determine study inclusion/exclusion etc.
19. Challenges
• Developing a one size fits all template –
although we expect that when the template is developed research groups will choose to search for
mechanistic targets relevant to their own question
• Finding the relevant studies –
this relies on having a good search strategy and validating this
• Determining study quality –
some QC criteria could be adopted from epidemiology
• Determining the strength of evidence for different study types
this will be arrived at by discussions within the team, tapping into to our multidisciplinary team
• Determining the relevance to humans
Again we will arrive at a consensus following examination of the studies and discusses within the team
• PUBLICATION BIAS
We will use recognised methods to quantify the extent to which this is likely to have occurred
• Collating the evidence