Page 19 - Pulse@UM Issue 1/2020

but also indirect estimates. In the eventuality However, experts caution against placing Bridging Translational Cancer
of AB trials being found/ conducted, the too much emphasis on the ranking outcome
estimates for AB will then become known for a few reasons; one of which relates to Research - Current Status,
as mixed estimates because, the estimate the lack of clinical consideration (such
of AB is now comprised of direct as well as treatment risk profile) if the treatment Challenges and Future
as indirect estimates. A mixed estimate ranking were to be interpreted at ‘face value’
is generally more precise than direct or (Mbuagbaw, L et al 2017). Strategies
indirect estimates alone (e.g. confidence
interval is narrower). Assumptions in NMA are generally similar LOOI MEE LEE NUR AISHAH MOHD TAIB
to assumptions in pairwise meta-analysis (Department of Biomedical Science) (Department of Surgery)
The key benefit of NMA is to pool relevant (i.e. studies are sufficiently similar and
evidence and perform comparison for three homogeneous with regards to disease
or more treatments simultaneously (even effect modifiers and study characteristics).
in the absence of head-to-head studies). However, in NMA, the validity of the
However, NMA may also be used to provide results (especially the indirect evidence) is
a concise summary of how each treatment is underpinned by additional assumptions.
compared to others, permitting researchers These assumptions are assumption of
to identify strength and gaps in the evidence consistency between direct and indirect
base (Figure 3). Additionally, decision estimates (i.e. both estimates should
makers may use NMA to gain information agree) and assumption of transitivity (i.e.
on treatment ranking. Treatment ranking, distribution of effect modifiers is similar
which is normally performed under the across comparisons). Apart from violation
Bayesian framework, may be expressed of these assumptions, methodological
as the probability of each treatment being flaws and inclusion of poorly conducted/
ranked the best. reported clinical trials are among some
factors that could undermine the quality of
NMA results.

It is therefore important to critically
appraise any meta-analysis work even
though it is the highest level of evidence.
Ironically, meta-analysis is in principle an Panel of speakers and organising committee. Organising committee members: Dr. Looi Mee Lee, Dr. Kamariah Ibra-
observational type of study. It is also easily him, Dr. Mohamad Shafiq Azanan, Dr. Jaime Jacqueline Jayapalan.
affected by the biases of observational study On 3 March 2020, in conjunction with interdisciplinary research grant proposals
design. World Cancer Day, the University of for funding agencies in the future. Prof.
Malaya Cancer Research Institute (UMCRi) Dr Nur Aishah Taib, Director of UMCRi
References: organised the “Bridging Translational briefed the participants on UMCRi and
Bannuru et al. Comparative effectiveness of pharmacologic inter- Cancer Research” forum. The forum saw current restructuring exercise towards UM
ventions for knee osteoarthritis: a systematic review and network
meta-analysis. Annals of internal medicine. 2015 Jan 6;162(1):46- participation from UM clinical and pre- Cancer Institute. She welcome members and
54.
Figure 2: Example of a network plot summarizing Chandler et al. Cochrane handbook for systematic reviews of in- clinical researchers; 47 participants were in associate members to participate once the
connectivity of evidence. Treatment groups are terventions. John Wiley & Sons; 2019 Sep 23.
represented by circles (nodes) while the lines between Mbuagbaw et al. Approaches to interpreting and choosing the attendance. The event aimed to provide a restructuring exercise has been completed.
each nodes represents direct comparison. Size of nodes best treatments in network meta-analyses. Systematic reviews. forum for researchers to share and facilitate
2017 Dec;6(1):79.
and width of lines are proportionate to the number Riley et al. Multivariate and network meta-analysis of multiple collaborative research among the UM The forum invited renowned researchers
of participants and number of studies, respectively. outcomes and multiple treatments: rationale, concepts, and ex-
amples. BMJ. 2017 Sep 13;358: j3932.
(Adapted from Bannuru et al, 2015) members and hence potentially motivate from UM to share their bench-to-bedside
18 Research Spotlight 19 Research Spotlight