Is there really anything we can do to prevent cancer; are we better packing it up and conceding that at some point we will all succumb to the emperor or all maladies? A well reportedpaper by the media last week in the journal, Science, does appear to have suggested this, and is surely a rude surprise to cancer researchers globally (including myself)!The paper in question found that 2/3 of the lifetime risk of developing cancer could be explained by only the random errors that accumulate overtime when cells replicate. This is indeed a shocking finding, but lets unpack it a little and see whether it is really saying all it has been reported as in the media - hint, it is not!Recent article reports cancer is due to mostly stochastic errors in cell divisions.Lung cancer, however, falls out of this category, hence it cannot be explained solely by the luck factor.source :http://tinyurl.com/okdtt3wFirstly, the study accounts for 31 cell-types in its analysis and measured the association with the corresponding physiological location, a far cry from having considered cancer as a whole. In fact, common cancers such as breast or prostate were omitted. Explicitly, the study took stem cells (the basic building block for cells of all types) from these organs and estimated the number of cell replications that a particular organ would likely undergo in a single lifetime. The premise is that different organs will require different rates of cell replication just for the mainantance of the baseline function. Each time a cell replicates there is a certain probability that an error will occur that may or may not lead to cancer, and so organs that duplicate more should be more likely to observe cancer risk - and this is what they observed by correlating the estimate of cell replication with lifetime risk.However, lets quickly remember an old adage "correlation does not equal causation". It is difficult to suggest that this correlation is anything more than a very plausible association, and it is certainly not appropriate to suggest that this correlation rules out factors that we can affect like diet and lifestyle. More skepticism should also be given to the result as it didn't include the most common cancers that would be household concerns like prostate and breast, and so suggesting as the media did that cancer as a whole is stochastic is wholly misleading - such descriptions of cancer as one homogenous entity is just flat out incorrect, and will I'm sure be the topic of a future post! We should note also that the study inferred mutation rates that likely lead to cancerous cells from the estimated number of cell divisions, and did not explicitly measure said mutations. While this is an acceptable method for a paper, it is so only in the knowledge that average mutation rates will not be useful for predicting individual cell DNA mutations, and further, not all mutations lead to cancer - so we cannot say that these cell replications will predict with a large degree of accuracy which cell will have which mutation, and whether it will be cancerous.The final passage of the paper discusses how this correlation can be useful when deciding how to direct research funding and public policy. They use a clustering algorithm to group the cell types they used into two categories: largely driven by random mutations, and largely not driven by random mutations. The suggestion is that research should be directed to discovering modifiable risk factors (diet and lifestyle, etc) for only those cancers where they will be likely to be found. In the light of the caveats I raise with interpreting their findings, I hope it will not be a surprise that I am worried greatly by their claim that the single correlation should drive policy! Cancer is an umbrella term for a set of very complex and harmful diseases, and research should be driven by a more robust set of results with careful analysis than these.For our podcast discussing this article look here: https://ia902600.us.archive.org/…/cancer…/cancer_podcast.mp3 !Itunes channel:http://feeds.feedburner.com/Popscienceplease
Is there really anything we can do to prevent cancer; are we better packing it up and conceding that at some point we will all succumb to the emperor or all maladies? A well reported paper by the media last week in the journal, Science, does appear to have suggested this, and is surely a rude surprise to cancer researchers globally (including myself)! The paper in question found that 2/3 of the lifetime risk of developing cancer could be explained by only the random errors that accumulate overtime when cells replicate. This is indeed a shocking finding, but lets unpack it a little and see whether it is really saying all it has been reported as in the media - hint, it is not! Recent article reports cancer is due to mostly stochastic errors in cell divisions.Lung cancer, however, falls out of this category, hence it cannot be explained solely by the luck factor.source :http://tinyurl.com/okdtt3w Firstly, the study accounts for 31 cell-types in its analysis and measured the association with the corresponding physiological location, a far cry from having considered cancer as a whole. In fact, common cancers such as breast or prostate were omitted. Explicitly, the study took stem cells (the basic building block for cells of all types) from these organs and estimated the number of cell replications that a particular organ would likely undergo in a single lifetime. The premise is that different organs will require different rates of cell replication just for the mainantance of the baseline function. Each time a cell replicates there is a certain probability that an error will occur that may or may not lead to cancer, and so organs that duplicate more should be more likely to observe cancer risk - and this is what they observed by correlating the estimate of cell replication with lifetime risk. However, lets quickly remember an old adage "correlation does not equal causation". It is difficult to suggest that this correlation is anything more than a very plausible association, and it is certainly not appropriate to suggest that this correlation rules out factors that we can affect like diet and lifestyle. More skepticism should also be given to the result as it didn't include the most common cancers that would be household concerns like prostate and breast, and so suggesting as the media did that cancer as a whole is stochastic is wholly misleading - such descriptions of cancer as one homogenous entity is just flat out incorrect, and will I'm sure be the topic of a future post! We should note also that the study inferred mutation rates that likely lead to cancerous cells from the estimated number of cell divisions, and did not explicitly measure said mutations. While this is an acceptable method for a paper, it is so only in the knowledge that average mutation rates will not be useful for predicting individual cell DNA mutations, and further, not all mutations lead to cancer - so we cannot say that these cell replications will predict with a large degree of accuracy which cell will have which mutation, and whether it will be cancerous. The final passage of the paper discusses how this correlation can be useful when deciding how to direct research funding and public policy. They use a clustering algorithm to group the cell types they used into two categories: largely driven by random mutations, and largely not driven by random mutations. The suggestion is that research should be directed to discovering modifiable risk factors (diet and lifestyle, etc) for only those cancers where they will be likely to be found. In the light of the caveats I raise with interpreting their findings, I hope it will not be a surprise that I am worried greatly by their claim that the single correlation should drive policy! Cancer is an umbrella term for a set of very complex and harmful diseases, and research should be driven by a more robust set of results with careful analysis than these. For our podcast discussing this article look here: