The brunt of acute toxicity from chemotherapy and radiotherapy used by oncologists falls on the organs with rapidly proliferating cell populations and short tissue turnover times, most notably the bone marrow and the gastrointestinal tract. This toxicity often results in treatment delay and/or dose reduction, strategies that may reduce the probability of a favourable outcome. In the past, much attention and research effort has been devoted to haematological toxicity, as a result of which an increasing number of haemopoietic growth factors are being used in clinical practice. Their use and the present trend towards dose escalation, has resulted in a need for a greater focus on damage to the gastrointestinal tract.

Gastrointestinal damage is most visibly expressed in the mouth. Oral complications have long been recognized as a major problem in the treatment of non-head and neck cancer, occurring overall in 39% of adults and 90% of children. The most frequent complications are ulcerative mucositis, infection, pain and haemorrhage. The incidence of oral mucositis ranges from a low of 12% in patients receiving adjuvant chemotherapy to 100% in patients receiving radiation doses greater than 180 cGy/day, 5 days per week. The prevalence of oral mucositis in patients receiving myeloablative chemotherapy has been reported variously as 28.6 to 100%. In our experience, 92% of patients undergoing myeloablative treatment suffer from mucositis, 64% (174/272) have WHO grade 3 or 4 mucositis for a median of 7 days (range 1–24) and 46% require opiate analgesia because of mucositis for a median of 5 days (range 1–28) (A.M. Wardley and J.H. Scarffe, unpublished data).

Severe oral mucositis prevents oral intake, necessitating parenteral nutrition. Diminished nutritional intake may compound the problem as there is an overall decrease in cell renewal and migration after starvation or protein deprivation. Oral mucositis is maximal when there is profound neutropenia, and the mouth is a major source of life-threatening infection; 25 to 64% of identifiable cases of septicaemia originate from the mouth. Although these effects may compound the severity of oral mucositis, the initial damage is the consequence of the cytotoxic effects on the epithelial stem cells. Hitherto, there has been no effective treatment for oral mucositis and measures have been merely palliative or directed at secondary problems such as infection.

There is a pronounced circadian rhythm in the cell-cycle activity of the oral epithelia of rodents. Certain therapeutic agents are cell-cycle specific in terms of cell killing. If one could selectively manipulate the cell-cycle of the stem cells of normal tissues using, for example, growth factors, in such a way as to render the stem cells more resistant to damage before or during treatment, and subsequently stimulate them to proliferate more rapidly after therapy, there might be benefits; treatment could be maintained or dose could be increased. A suitable experimental system is required for the investigation of agents that may fulfil such functions. We have developed, and present here, an in vivo model of mucositis using BDF₁ mice, measuring changes in epithelial cellularity and proliferation that we believe will be of value for such studies.

Male BDF₁ mice housed under conventional conditions (relative humidity 50–55% and temperature 20–22°C with food and water ad libitum) were used for all experiments. Illumination was in a 12-h light: dark cycle with the switch times at 7 a.m./p.m. At the time of experiment, animals were 10–12 week old, with an average weight of 25 g. For convenience, some animals were placed in a 12-h reversed light-cycle room for 2 weeks before the commencement of the experiment; this results in complete and stable reversal of circadian rhythms. Any procedures involving mice in the dark cycle were done under a photographic red-safe light, so as not to disrupt the prevailing circadian rhythm. All times quoted are corrected to the equivalent normal (as opposed to reversed) light cycle. S-phase labelling was with 10 mg intraperitoneal bromodeoxyuridine in 0.5 ml saline or 25 ?Ci (925 kBq, spec. act. 222 GBq/mmol) of methyl tritiated thymidine (Du Pont (UK) Ltd, Stevenage) in 0.1 ml saline at a designated time, and animals were killed 40 min later. All experiments were performed in accordance with the UK Home Office regulations in the Scientific Procedures (1986) Act. All animals were killed by cervical dislocation.

All specimens were fixed in Carnoy’s solution (ethanol:chloroform:propionic acid, 6:3:1) for 40 min and transferred to 70% alcohol until required. The whole tongue was excised at its base, fixed in Carnoy’s for 5 min, sectioned in the median saggital plane and returned to Carnoy’s for 40–60 min, to ensure rapid and uniform fixation. The buccal mucosa and lips were carefully dissected out and placed flat on filter paper for fixing. Specimens were routinely embedded in paraffin and 3-?m sections cut and either stained with haematoxylin and eosin, dipped (Ilford K5 emulsion diluted 1:1) for autoradiography (2-week exposure, D19 developer, 8 min) or dried overnight at 37°C and processed for immunostaining (bromodeoxyuridine). [³H]thymide labelling was only used in some initial studies relating to technical validation and for some of the circadian-rhythm studies.

Slides were dewaxed overnight in fresh xylene, then washed in 100% ethanol; endogenous peroxidase activity was blocked with 1% hydrogen peroxide in methanol for 30 min. To expose the DNA, slides were placed in 1 M HCl at 60°C for 8 min, followed by 5 min in boric acid buffer, or digested for 20 min in 0.01% pepsin solution in 0.1 M HCl followed by 20 min in 0.8 M HCl. Initial studies (not presented here) showed no difference in number of labelled cells between thymidine and bromodeoxyuridine studies. Furthermore, no difference was seen between the labelling after either of the two denaturation processes. In most of the bromodeoxyuridine labelling we used the HCl denaturation process. However, the pepsin denaturation was most appropriate for the high-dose samples (30 Gy) and was used for some of the repeat 20-Gy samples. Slides were blocked with 1/20 normal rabbit serum for 30 min and then incubated with primary antibody (rat anti-bromodeoxyuridine 1/5) (Harlan Sera-Lab, Peterborough) added for 1 h. After washing, secondary antibody (rabbit antirat peroxidase 1/100 in 10% normal mouse serum) was added for 1 h. Diaminobenzidine solution (0.5 mg/ml) was added for 6 min and slides counterstained with thionine.

For the circadian-rhythm experiment a Zeiss microscope (×40 Planapo oil-immersion objective) was used. One thousand or more consecutive basal cells were counted, and, according to staining, designated positive or negative for bromodeoxyuridine (as judged by eye) or [³H]thymidine (more than three silver grains) uptake. Previous studies had shown no difference in labelling index when these two approaches were used. For ventral tongue epithelium, counting commenced 2 mm from the distal tip and proceeded proximally so that the area bearing filiform papillae was excluded. The inner labial mucosa was the start-point for counting in the lip, followed by the transition zone and then the epidermal aspect where only the interfollicular epidermis was scored. For the buccal mucosa, only non-folded areas sectioned perpendicular to the long axis of the epithelium were included in the analysis.

Histometric measurements were made with a Zeiss Axiohome^TM computer-assisted microscope. This enables one to trace around an object or area (e.g. nucleated epidermis) that is displayed on a computer screen and simultaneously seen superimposed down the microscope. The area of the defined region could then be calculated. Alternatively, an irregular line (e.g. the basement membrane) could be traced and its length calculated. Cell nuclei were marked with different icons according to their category. Using these facilities, an area or length of epithelium was designated and the nuclei were recorded as basal or suprabasal, bromodeoxyuridine-positive or -negative, and the number of cells/mm² of epithelium or basal cells/mm of basement membrane was calculated. Only the nucleated epithelium from the basement membrane to the stratum granulosum/corneum interface was analysed. For the ventral tongue, counting always began at the same designated point, 2 mm proximal to the tongue tip, and an area of either 0.05 mm² or 0.1 mm², or at least 500 cells, were counted for each of the two halves of the tongue. Buccal mucosa does not maintain its anatomical integrity to the same extent as tongue and tends to become distorted. Consequently, as many structurally preserved areas as possible were included for counting purposes.