Article Text

PDF

Defective denominators
  1. R A Goodlad
  1. Correspondence to:
    Dr R A Goodlad
    Cancer Research UK, 44 Lincoln’s Inn Fields, London WC2A 3PX, UK; goodladcancer.org.uk
  1. J Cummings1,
  2. N Coleman2
  1. 1Ninewells Hospital and Medical School, Dundee, UK
  2. 2Hutchison/MRC Centre, Cambridge, UK
  1. R A Goodlad3,
  2. D Alferez3
  1. 3Cancer Research UK, London, UK
  1. J Cummings4,
  2. N Coleman5
  1. 4Ninewells Hospital and Medical School, Dundee, UK
  2. 5Hutchison/MRC Centre, Cambridge, UK

Statistics from Altmetric.com

I was interested in the paper by Langlands et al in which “prebiotic” carbohydrates altered the mucosal flora but apparently had no effect on cell proliferation (Gut 2004;53:1610–16). The matter is of some importance as the products of in vivo fermentation (short chain fatty acids) may increase epithelial cell proliferation, leading to the possibility that such supplements could actually enhance the risk of colorectal cancer.1,2

The authors state that methodology (of gut microflora study) is always an important issue and I argue that this also applies to cell proliferation studies, as the results of the present work may be misleading on two counts. Firstly, I would never recommend the use of proliferating cell nuclear antigen as a marker of cell proliferation as: (1) the method is difficult to standardise; (2) the antigen has a long half life; and (3) anomalous expression has been demonstrated in non-cycling near tumours and after administration of growth factors.3 For sections, Ki67 is far better however even using this antibody the results of the present study are unlikely to be conclusive as only 2–4 crypts could be scored; for most studies I would recommend scoring 30 hemi crypts.

The second point is that reliance on labelling indices can be misleading as lack of difference does not necessarily mean no proliferative change as both sides of the ratio (labelled cells divided by number of cells) could have altered. This was demonstrated in our studies of epidermal growth factor in parenterally fed rats where no differences in labelling index between orally fed and parenterally fed rats could be seen despite halving tissue weight and crypt cell production. When the data were re-expressed as labelling per crypt, the effects of treatment became apparent4; a similar effect was seen in the stomach following misoprostol treatment.5,6

There is however a far easier and well validated method available for the study of human tissue. This is the so-called microdissection technique in which small pieces of stained material are teased apart and mitotic figures scored.7 This literally allows one to score over 100 crypts (if so wished) and as the results are expressed per crypt the effects of changes in denominator are automatically accounted for.

References

Author’s reply

We thank Goodlad for his interest in our article. In our study (Gut 2004;53:1610–16), we assessed expression of the three markers most commonly used to indicate cell cycle entry in tissue sections. Importantly, there was no difference in the data obtained for all three. We agree that proliferating cell nuclear antigen is of limited value for the reasons mentioned by Goodlad and also the fact that the protein has a role in DNA repair, which reduces its specificity as a cell cycle marker. Similarly, Ki67 is not expressed by all cycling cells, may be downregulated by nutritional deprivation, and may also be involved in non-cell cycle related processes, such as ribosomal biosynthesis.1

We consider that the most useful markers of cycling cells are the minichromosome maintenance (MCM) proteins, which are abundant at all phases of the cell cycle and are downregulated following exit into quiescence, differentiation, or senescence.1,2 MCMs therefore provide a sensitive and specific indication of cell cycle entry. In our opinion these markers are preferable to counting mitotic figures, which is a subjective and error prone exercise that by definition provides a limited phase specific indication of cell cycle state in histological sections.

We agree that proliferation indices can be misleading and that when assessing large bowel crypts it is important to determine the number of labelled cells per crypt.3 We confirm that the mucosa in all subjects in our study was microscopically normal, as well as macroscopically normal, as stated. In particular, there was no difference in crypt length and number of cells per crypt between the study groups. The labelling indices determined were therefore valid indicators of cell cycle entry in the samples investigated.

Prebiotic carbohydrates, such as those used in our study, are completely fermented in the large bowel and none is excreted in faeces. The principal products of this fermentation are short chain fatty acids (SCFA). While SCFA have been associated with increased cell proliferation in some animal models, it is hard to believe that what are the major anions in the colon of all mammalian species should enhance the risk of cancer, particularly since one of these fatty acids, butyrate, is thought to be a differentiating agent. Fermented carbohydrates, such as dietary fibre, when measured properly in the diet, appear to protect against colorectal cancer in observational studies.4 The observed lack of effect of prebiotic carbohydrates on colonocyte proliferation in our study suggests that a substantial increase in fermentable carbohydrate intake, as provided by these prebiotics, does not enhance proliferation, as shown in some animal models, and thus might be regarded as adding to the protective role of the fermentable non-starch polysaccharides (fibre).

REFERENCES

Author’s reply

Further to Cummings and Coleman’s reply to my letter above, I would like to question the advocacy of minichromosome maintenance (MCM) proteins as proliferative markers, as the number of MCM positive cells can greatly exceed other labels and they are widely distributed on unreplicated chromatin.1 They would appear to be more of an indicator of replication potential and, as such, are likely to be useful markers of dysplasia.2 In addition, scoring immunohistochemical labelled cells is just as, if not more, “subjective and error prone” than scoring mitotic figures (which are far easier to score in “squash” preparations than in sections). My main concern still stands, as scoring histological sections of human biopsies, unlike squash preparations, leads to the sampling of a very limited number of crypts (2–4 in the present study) which prevents credence of the “observed lack of effect” of prebiotic carbohydrates.

Finally, I think that the jury is still out on the “protective role” of fermentable non-starch polysaccharides (fibre) as while the EPIC study showed a dramatic effect of intrinsically high fibre diets,3 many others have shown null effects and some of these, especially the intervention ones, demonstrated adverse effects. For example, wheat bran supplementation increased polyp recurrence in women4 and ispaghula had a more general adverse effect on polyps.5

REFERENCES

Author’s reply

We sought to identify cells at any point of the cell cycle, regardless of the rate of cycling or the duration of particular cell cycle phases. We therefore elected not to assess individual cell cycle phases in our samples, either by immunostaining or by counting mitotic figures.

While additional roles for minichromosome maintenance (MCM) proteins have been suggested, there is strong evidence that they function as essential replication factors.1 MCMs are displaced from chromatin following DNA replication, yet remain abundant in the nucleus throughout the cell cycle.2 Importantly, several groups have shown that MCMs are lost following cell cycle exit (into differentiation, quiescence, or senescence).3–5 MCMs are therefore useful immunohistochemical markers of cell cycle state. It is not surprising that MCMs are more abundant than Ki67 and proliferating cell nuclear antigen (PCNA), as the latter markers are not detectable in all cycling cells.

Objectivity and reproducibility in the interpretation of immunohistochemical staining are functions of the marker used. Some markers, such as PCNA, produce substantial variation in staining intensity and cause difficulty in slide interpretation. However, our MCM antibodies have not provided us with such difficulties, resulting in low interobserver variation in numerous studies to date.2

Interpreting observational and intervention studies of fibre has filled many journal pages in recent years. There are numerous problems which, in the context of the present discussion, relate primarily to people treating all sources of fibre as being equal, thinking that fibre supplements will have the same effect as fibre present in whole foods in the diet and the amounts of fibre considered to be protective. With regard to the study by Alberts and colleagues,6 the fibre was provided as a supplement and was only of wheat bran. As Goodlad and Alferez correctly note, the EPIC study showed a protective effect for fibre when intrinsically part of the diet, and from mixed sources. In other words, it is a high fibre diet that protects. The Bonithon-Kopp study7 used a fibre supplement, ispaghula, not found in most diets of the world, and at a very small dose of only about 3 g/day.

REFERENCES

View Abstract

Footnotes

  • Conflict of interest: None declared.

Footnotes

  • Conflict of interest: None declared.

Footnotes

  • Conflict of interest: None declared.

Footnotes

  • Conflict of interest: None declared.

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.