Parabens in Perspective: Part II
PARABENS ARE PRESENT IN BREAST TUMOURS?
This is the second in a series of articles in each of which I will focus on one specific study that has contributed to the current concerns over the use of parabens in cosmetics. If you have not already done so, it may be useful for you to read the introductory article posted earlier before reading this one!
The paper I will critique in this article is:
Concentrations of Parabens in Human Breast Tumours
Darbre, P.D, Aljarrah, A, Miller, W. R, Coldham, N. G, Sauer, M. J. and Pope, G. S, J. Appl. Toxicol. 24, 5 – 13 (2004)
This study consisted of the analysis of 20 tissue samples removed from human breast tumours, to establish if intact parabens were present, with 6 blank control analyses without any tissues present.
To fully analyse this paper, it is necessary, unusually, to go back a few years before the actual publication. One of the drivers for this work being carried out was the study by Routledge et al, covered in the previous article in this series. Partly as a result of the Routledge work, Dr. Darbre had made several public pronouncements and published several speculative reviews propounding her theory that the use of underarm products (deodorants/antiperspirants) could be, in part, responsible for breast cancer. Routledge’s findings provided ammunition for Darbre to include parabens as specific suspects, alongside aluminium (that’s aluminum to my US readers!). Darbre suggested that parabens in the underarm deodorants and antiperspirants could be absorbed by the skin and migrate to the breast tissue close by. This study, therefore, was designed to test the theory by the expectation of finding parabens in human breast tumour tissue.
Before I get to the meat of the study, there are two problems with the rationale above:
1) Antiperspirants do not require any preservation from parabens, or any other preservative; very few deodorants require preservation, and even fewer use parabens
2) Any substances applied to the underarm do NOT migrate to the breast area. Both the blood and the lymphatic circulatory systems flow AWAY from the breast
So we have a very shaky start, and we haven’t even got to the actual study yet!
When this study was first published, it aroused a storm of criticism, and several letters to the editor of the publishing journal. In the interests of brevity (and because I would prefer you to keep reading to the end) I will simply list the flaws in the methodology, data manipulation and conclusions mostly without comment. I will then comment in some details on the most important aspects of this list:
1) A statistically insignificant number of tissue samples tested
2) No patient history recorded (some treatments use parabens as preservatives)
3) No healthy tissue tested as a control
4) Benzylparaben was used to calculate the analytical recovery factor
5) Parabens were detected in the blank control samples (omitting the tissue)
6) The 4 highest results were used to calculate an average parabens concentration
7) The average was then used in a discussion comparing the results from a study that used a very different parabens composition
8) Wrongly compared the parabens concentrations determined with those of PCBs and OCPs found in humans
9) Bioaccumulation was claimed for parabens
The shortcomings of the first 3 items must be fairly obvious, so I will go straight to number 4. When trying to detect any chemical, it is good practise to add a known quantity of the chemical to the target substance (in this case, the tissues) and check that the analysis detects the same concentration that was deliberately added. If only 50% is detected, this is known as the analytical recovery, and this is taken into consideration when calculating the results. The analysis was designed to detect the following parabens: methyl, ethyl, propyl, butyl, isobutyl and benzyl. No benzylparaben was detected in any of the tissues – not surprising as benzylparaben is not used in cosmetics! Benzylparaben is significantly different to the other parabens mentioned in terms of its chemical structure. Whilst it is good practise to determine the analytical recovery, it is extremely poor practise (and totally illogical) to use a substance that is not even present! This gives a meaningless result. Additionally, as benzylparaben is the least soluble of the parabens, it is likely to give the lowest analytical recovery, and this means that the factor applied to the other parabens will result in a higher concentration than was actually the case. The recovery factor should have been determined for each individual paraben – this was very lazy science!
I will come back to flaw number 5 later, and first deal with the incredible decision to ignore all but the 4 highest concentrations of parabens when using the data to calculate an average figure. This constitutes gross distortion of the data. Statistically, it can be acceptable to discount the highest and the lowest figures in a data series as being the least accurate, but it is NEVER acceptable to only take the figures that best fit the argument! The distorted “corrected” average level of 100ng/g (ng = nanogram – 1 billionth of a gram) was then used as a comparator against studies that found levels of c. 150ng/ml of n-propyl, n-butyl and iso-butylparaben stimulated growth of oestrogen-dependant MCF7 human breast cancer (HBC) cells. This results in yet further distortion as the distorted average is then being compared with results for, essentially, different compounds as 62% of the total paraben level detected in the tissues was methylparaben and a further 10% was ethylparaben, for which there are no data on their effect on MCF7 HBC cells. Therefore, only 28% of the comparator figure is an acceptable comparison as, based on many other studies, methyl and ethylparaben are significantly less likely to exhibit any oestrogenic effect than the higher (longer carbon chain) parabens. A more reasonable comparison would be to take the propyl and butylparaben components from the “uncorrected” average to give a figure of c. 5.6ng/g, which is significantly lower than the 150ng/g level claimed to stimulate growth of MCF7 HBC cells. I repeat – gross distortion of the data.
The conclusions of the study claim proof of bioaccumulation of parabens in human breast tumour tissue. It is simply not possible to establish bioaccumulation using only one data point. This can only be achieved by monitoring over a period of time and observing an increase in the concentrations present and this study uses only a single data point. If the parabens truly are present in the tumour tissue, they may be simply background levels present during the metabolic process. The report then draws a comparison between parabens and PCB’s and OCP’s, quoting mean levels of 20, 267 and 707ng/g tissue respectively as further evidence of bioaccumulation of parabens (and wrongly claiming that these levels are similar – they differ by more than an order of magnitude). I believe that this proves precisely the opposite. Human exposure to parabens must be several orders of magnitude greater than exposure to PCB’s or OCP’s, yet the latter have significantly higher residual levels than parabens. If parabens were bioaccumulative, this would be reflected in mean levels significantly higher than PCB’s and OCP’s.
All the information so far points to a very shoddy study, with an implausible rationale, poor methodology and indefensible distortion of data. However, I have left the best until last – Flaw Number 5! It is good scientific practise to carry out control experiments. These can be negative or positive controls. In this instance, it was a negative control designed to check that parabens would not be detected in the absence of the actual tissue samples. But they were! This should have raised alarm bells – parabens were being detected where they shouldn’t be! The alarm bells were silenced by making the assumption that the parabens detected in the blank controls were due to contamination from the glassware (although it is much more likely that the lab workers had cosmetics containing parabens that found their way into the controls). This is a fair assumption, but why was it assumed only that the blanks were contaminated? Why not consider that the tissues were similarly contaminated? Surely it is illogical to think otherwise – especially when you consider the following points:
1) There was no statistical difference between the parabens concentrations found in the tissues and the blanks – the highest “blank” was higher than 12 of the 20 tissue samples; the 2nd highest “blank” higher than 9 of the tissue samples!
2) The parabens ratios detected were what would be expected, given the their relative frequency of use in cosmetics
3) The ratios of the 5 parabens detected were statistically the same in both the tissues and the blanks
Expanding on points 2/3, to detect the parabens in the same ratios is highly suspicious. What this implies is that parabens can be absorbed through the skin, migrate around the body and lodge in breast tumours without any difference in the rate of breakdown. There is no dispute that parabens may be at least partially absorbed through skin, but it is known that the skin and the blood contain enzymes called esterases. These enzymes break down a group of chemicals known as esters – including parabens esters! The rate of breakdown is different for each of the different parabens so, logically, it may be expected that passage through the body (to the breast?) would result in a change of the relative ratios.
The combination of the statistical similarity of both the actual concentrations of parabens AND their ratios to each other in both the tissue and the blank controls is an astonishing coincidence. Could it possibly be the case that the parabens “detected” in the tissue samples were also present due to the same source of contamination as in the blank controls?
Given that this study almost single handedly caused all the hysteria over the use of parabens that has continued to this day, I find it quite worrying that such a deeply flawed paper got into print. Over the past 6 years the results of this study have been distorted and manipulated even further – the Wikipedia entry actually stated that “high concentrations” of parabens had been found in human breast tumours – until recently, that is. It now states the actual concentration in terms of nanograms per gram, with an explanation of just how large a nanogram is. I wonder who made that change. . . . .
In summary – congratulations if you made it this far – that probably makes the two of us!
Whilst I am unable to claim emphatically that parabens were not present in the breast tumour samples, I hope that I can claim that this study is, at best, highly ambiguous, because it does not provide irrefutable proof of the presence of parabens in human breast tumours. I do believe, however, that the evidence is not in favour of parabens having been present in the tissue tested.
Given that the claim of this single study to have detected intact parabens in breast tumours is the basis for much of the subsequent work, and it provides the fundamental cause for concern over the use of parabens, shouldn’t we all be concerned that this is all based on such inconclusive and deeply flawed evidence?
As a final point, even the authors, in their summary, stated that the presence (?) of parabens did not prove a causal link to breast cancer. So why do so many others make this claim?
More about the author: Dene Godfrey has been involved with preservatives for cosmetics since 1981, from both technical and commercial angles and has a degree in chemistry. Read more from this author