Why I won’t be eating two more portions a day

The health service, nutritionists, the government and your parents are always telling you to eat more fruit and veg.  Its sound advice, it really is, but how can we actually quantify the amount of fruit and veg a person should eat?  The answer is we can’t.  At least not accurately enough to warrant telling people exactly how many ‘portions’ we should be eating in a day.  The ‘5-a-day’ message is a reasonable one though.  Even if it does seem to suggest there is a set amount of fruit and veg the average person should eat in a day, the real message is plainly eat more fruit and veg because it’s good for you.

So why then do I turn on the TV this morning to find the news barking about how a new study has shown we need to eat 7 portions of fruit and veg a day to cut the risk of dying of common diseases?

The news comes from a study carried out by researchers at University College London who analysed questionnaire data collected by the NHS on people’s diet and lifestyle.  Essentially, what has been reported is that the study indicates that the more fruit and vegetables people ate, the less likely they were to die, at any given age.  In other words – fruit and veg is good for you – not a particularly new message.  So why, oh why, then did they have to go and quantify how much you should be eating to avoid death?  It’s just not possible, and not just because it’s difficult to interpret real world quantities from a self-filled questionnaire, but because the study itself has flaws which make it impossible.  This is not a criticism of the research but more a facet of this type of study which should be respected.


The first thing to note about the study is that it collected data from participants from 2001-2008.  Although this is better than a single measurement it still only represents a relatively short amount of time in a persons life. This isn’t particularly useful when a healthy lifestyle is something you have to maintain rather than just be healthy at the exact moment you were asked.  This isn’t to say that this kind of study isn’t useful but its definitely a limitation worth considering.

Confounding factors

The other big factor which needs to be taken into account is other lifestyle and environmental factors which affect general mortality.  The participants in the study are likely to be exposed to a wide range of different factors such as smoking, drinking, exercise, where they live, what their job is etc.  You can try to take these into account when analysing the data but you’ll never be able to remove them from the equation – particular when you are dealing with a large scale cross-sectional study.  There’s also the issue that healthy eaters tend to live healthier lifestyles.  So how do we know that the benefits of 7-a-day aren’t because those people go to the gym more often or don’t smoke?

Risk reduction paradigm

It’s also worth mentioning that lifestyle interventions that reduce risk against anything need to be assessed alongside the impact of other lifestyle factors.  This is important so we can see the ‘weighting’ of each factor on risk reduction, rather than each in isolation, and as such be able to work out how much impact a certain intervention has to an individual.  For example, if you eat 7-a-day but you are also a heavy smoker; does eating 7-a-day even have an impact on your risk of cancer or heart disease?  The effects of eating 7-a-day, over say 5-a-day, could be so small that the increased risk from smoking makes it irrelevant.  In fact, if you are a heavy smoker it might not be any benefit to eat fruit and veg at all.


The BBC have been cautious over touting a 7-a-day message which is great to see but I haven’t yet seen the full media coverage and I expect there will be some which completely overstate the results in this study.  I don’t think the study authors should have allowed a 7-a-day message to accompany their research because I think they are a long way off showing that eating 7-a-day has any benefit over 5-a-day.  Whichever is the case, the important message is that a healthy diet and a healthy lifestyle are going to be your best bet at having some influence over how and when you die, but I wouldn’t get too hung up on it.

Full study article (Open Access, yay!): http://jech.bmj.com/content/early/2014/03/03/jech-2013-203500.full

Making personalised medicine for cancer a reality

Judith Potts appeared in the Telegraph today discussing the future of cancer treatments.  Her article, on defining cancer by its molecular attributes as opposed to region in the body, highlights where research should be going.

It’s becoming clearer to scientists that lumping cancers as ‘breast’, ‘ovarian’ or ‘lung’ may not be useful when it comes to treating a patient.  Within each type of cancer there are many sub-types, which are all categorised depending on their molecular, genetic and physical characteristics.  But there is overlap between sub-types of one cancer, which blurs the boundaries that define them.  Cancers can also change during progression of the disease, masking their categorised features and developing new ones.

Cancers are also individual.  Tumours derive from the patient’s own cells and so each cancer is individually characterised by the genetic and environmental factors that have influenced that person’s life.   We know that certain genetic mutations are more likely to occur in certain cancers but the individuality of cancer means we can’t expect a blanket treatment for all patients with one type of cancer.

Judith proposes that molecular profiling for individual cancer patients is the way forward, and I am inclined to agree.  This method looks at a wide range of molecular markers that each represents a particular weakness in the cancer.  A clinician could then use this information to match up each weakness with a drug to exploit it.  This would be done on data gathered from a tumour of an individual patient, providing a clear road towards fully personalised medicine.  Clinicians could also get around the evolving cancer problem by taking new molecular profiles from the patient at different points in their disease and adapting their treatment accordingly.

This type of care for cancer patients is already available in some countries, if you can afford it, but there will be several problems to overcome when the technology becomes widespread.  Biotech companies that patent molecular profiling kits could increase costs.  Confidentiality around their product could also hide whether or not the molecular profiles are accurate.  This could lead to patients receiving an ineffective treatment or a treatment that causes harm.

There is also the issue that drugs approved for use for say lung cancer, may not have been tested against breast cancer.  Molecular profiling might tell you that a patient’s cancer has a weakness for that drug, but you would struggle to be able to give it to them.  And what about new experimental drugs?  With no evidence from clinical trials it wouldn’t be possible to use one even if you knew that the patients cancer would be sensitive to the treatment.  One way around this would be to reassess how clinical trials are set up and allow for greater flexibility in trial design (a discussion for another time).

Molecular profiling is on the horizon and offers obvious benefits to the way we treat cancer.   However, for it to work, the policy and healthcare system needs to evolve with the science.

Chemical imaging could be the future of cancer tissue analysis

A new method for analysing tumour samples has been revealed by scientists at Imperial College London.  Current methods are laborious, involving the manual testing and interpretation of tumour characteristics by a histologist.  In a new study, published in the Proceedings of the National Academy of Sciences, researchers describe how they are trying to take cancer diagnosis into the digital age.

The technology is based on a widely used technique called mass spectrometry, which is used to work out what biological molecules are present in a sample, such as blood or urine.  Using a modified version called mass spectrometry imaging; the researchers are able to generate an image of a patient’s tissue sample, showing the location and density of different biological molecules.

The technique works by passing a laser beam over a tissue sample, such as a tumour, which reacts with the biological molecules to produce a signal that can be converted into a pixelated image.  Each pixel of the image reveals how much of a specific molecule is present in that region of the tissue – producing a map that can reveal specific characteristics of the diseased tissue.

A cancer patient’s tumour is currently characterised based on structural features that are detected by staining with expensive reagents that require expertise to interpret.  Mass spectrometry imaging characterises tumours based on their molecular features and could provide a more accurate assessment of an individual’s cancer.  This could help guide clinicians to make the best treatment decisions for each patient and support a move towards more personalised medicine.  The technique can also be automated, allowing a computer to rapidly analyse hundreds of different molecules in one go, reducing testing time from over a week to only a few hours.

Dr Kirill Veselkov, corresponding author of the study from the Department of Surgery and Cancer at Imperial College London, said: “MSI is an extremely promising technology, but the analysis required to provide information that doctors or scientists can interpret easily is very complex. This work overcomes some of the obstacles to translating MSI’s potential into the clinic. It’s the first step towards creating the next generation of fully automated histological analysis.”