In this series of studies the purpose was to investigate the possible mechanisms affecting the digestion of lactose, as well as the gastrointestinal symptoms induced by undigested lactose. Side by side with this, the prevalent methods for diagnosing hypolactasia and lactose intolerance came under scrutiny. Lactose maldigestion is usually identified by means of an oral lactose tolerance test, in which, following ingestion of 50 g lactose, maldigestion is detected either by a reduction in the increase of an unabsorbed breakdown monosaccharide component of lactose (glucose or galactose) in either the blood or the urine concentration, or by an increase of hydrogen produced by the colonic fermentation of malabsorbed lactose (see Arola 1994).
Great care was taken to minimise all those factors known to affect test variables for detecting hypolactasia. In this we were enormously supported by the extremely co-operative study subjects, all of whom were healthy volunteers. Pre-test restrictions such as the limited use of medication and a controlled diet before and during the test were imposed in order to improve the comparability of the study periods. It is just possible, however, that the avoidance of coffee among coffee drinkers and smoking among smokers may have influenced the symptoms recorded during the test. Fortunately, rather than affecting gastrointestinal symptoms, coffee withdrawal is probably more likely to result in headache and overall discomfort, and there was only a relatively small number of smokers.
Studies were, whenever possible, randomised (II-IV), double-blinded (II, IV) and cross-overed (II-IV). The evaluation of the usefulness of the Micro H2 breath hydrogen analyser for detecting hypolactasia (Study I) was not blinded. The analyser proved to be reliable for this purpose compared with the stable hydrogen measurement unit Quintron MicroLyzer or with the Golden Standard. To compare its ability to measure breath hydrogen concentration, it might also have been possible to measure exhaled hydrogen after another undigestible carbohydrate such as lactulose, as we did in a previous study (Teuri et al 1999). This combination lactose/lactulose breath hydrogen test enables one to calculate lactose malabsorption by comparing areas under the curves of the lactulose breath test (100% malabsorbed) and the lactose breath test. It also enables the measurement of intestinal transit time (Hammer et al 1996).
However, more experiments are needed to standardise the collection of exhaled hydrogen for analysis with the Micro H2 analyser. In this study the collection of hydrogen was controlled by strictly limiting the collection time to 70 s and inhibiting the passage of hydrogen through the nose by using a nose clip, as advised by the manufacturer of the equipment. However, even though the subjects were allowed to exhale slowly through the mouthpiece of the Micro H2 analyser, methods of blowing varied. This possible bias in the inter-individual comparison of results was minimised by using cross-over study designs when no comparison between the subjects was needed.
The total number of study periods in each study was kept to the minimum. This was because of the perennial difficulty in getting lactose maldigesters to participate in a study in which they will have to ingest several large doses of lactose, after which they know, or assume, that they will suffer severe gastrointestinal symptoms. Based on experience from earlier studies, women are more willing to take part in studies such as these than men. It was thought that it would be better to investigate single-sex groups in order to make the study group more homogeneous. General conclusions based on very homogeneous groups should, however, be drawn with circumspection.
There are relatively few reports describing the possible differences in the tolerance of lactose between men and women. Hypolactasia is equally prevalent in both sexes (Sahi 1974b), though female lactose maldigesters seem to be more sensitive to lactose than men (Vesa et al 1998) despite the fact that the amount of malabsorbed lactose measured by the breath hydrogen test is similar in both sexes (Krause et al 1997).
The influence of gender and the menstrual cycle on gastric emptying and gastrointestinal transit has not been clearly demonstrated. Several studies have found gastric emptying to be faster in men than in women (Datz et al 1987, Hutson et al 1989, Degen and Philips 1996, Hermansson and Sivertsson 1996, Knight et al 1997), whereas others found little or no difference (Horowitz et al 1985, Saltzberg et al 1988, Madsen 1992). In a recent study, evidence of the postprandial reduction of gastric emptying was found in women, but no differences between men and women during fasting were observed (Caballero-Plasencia et al 1999). The phases of the menstrual cycle, however, have not been reported as modifying the gastric emptying of solids or liquids (Caballero-Plasencia et al 1999), but Wald et al (1981) reported that the whole gastrointestinal transit was reduced by 25% during the luteal phase compared with the follicular phase. The interesting questions of gender or phases of the menstrual cycle on gastrointestinal transit cannot be answered by our present studies: we were restricted to one gender and the phases of menstrual cycle were not recorded.
The symptom questionnaire we used was based on the questionnaires used previously by us in grading gastrointestinal symptoms in lactose intolerance studies (Vesa et al 1996, Teuri et al 1999). This type of questionnaire has not yet been statistically validated in recording lactose-induced gastrointestinal symptoms but has been successfully used by us with almost two hundred lactose intolerant subjects. We also compared the numerical grading of symptoms with a visual analogue scaling (VAS). Since no differences in symptom scores were found between these two grading methods, the numerical grading was chosen, mainly because of the more straightforward analysis of results.
Using experimental animals as a model for investigating hypolactasia provides a valuable tool for detecting mechanisms behind the regulation of the expression and activity of lactase. Even though rodents and humans differ in many ways, the basics of the normal physiological pattern of the lactase enzyme are similar. Thus it makes sense to use the rat as an experimental model for studying hypolactasia. Both human beings and the rat have a high expression of lactase in sucklings, after which the expression and activity both decline to the lower level of adulthood. The time period of this decline differs. In rats the decline begins soon after weaning, as shown by us and others (Büller et al 1989), and in humans, later in childhood (Sahi et al 1972). However, this is a minor difference when adult animals are used, as they were in this study.
Since we wanted to investigate the effect of a lactose supplementation on low adult levels of lactase, the animals we used were 8-9 weeks old. At this age, the level of lactase was already significantly reduced. The lactose content of the milk of nursing rats varies between 1 and 3.5%, depending on the age of the sucklings (Kuhn 1972). The lowest concentration for supplementation, which was 3%, was chosen to imitate this level of lactose in the milk of nursing rats. Two significantly higher concentrations, 10% and 20%, were also used, in order to see clearly the possible dose dependency in the induction of the lactase enzyme.
Water intake and weight gain were reduced by 50-60% in those animals who received large doses of lactose mixed with water. The intake of the lactose-water solution and the overall well-being of the animals were recorded daily. It is possible that the rats disliked the sweet taste of the lactose and thus drank less. Nor can the other possibility of increased gastrointestinal discomfort by maldigested lactose be excluded. No signs of diarrhoea, however, were noticed. It would be interesting, in an experimental hypolactasia model, to have a suitable piece of equipment for recording the gastrointestinal discomfort of the rat, even though experimental models are for investigating the mechanisms, not the symptoms, of lactose maldigestion.
Reduced gastric emptying and total gastrointestinal transit time are thought to increase mucosal contact time and thus to reduce the quantities of unabsorbed lactose in the colon. In addition to this hypothesis, reduced intestinal motility may also affect the colonic flora and thus modify fermentation products, as suggested by El Oufir et al (1996) and Lewis and Heaton (1997). This may also improve tolerance to lactose.
We found that by inhibiting peripheral muscarinic receptors with an oral dose of propantheline, tolerance to lactose, measured as the severity of gastrointestinal symptoms, improved. It is possible that reduced symptoms in our study after propantheline could, at least partly, be due to the non-specific antispasmodic effect of propantheline. However, in addition to reduced gastrointestinal symptoms, the reduced excretion of breath hydrogen and the increased excretion of urine galactose indicate better absorption of lactose. Among other anticholinergic and antispasmodic properties, propantheline has been shown effectively to delay gastric emptying (Hurwitz et al 1977).
Apart from the slight tendency of an earlier appearance of gastrointestinal symptoms, we found no differences in tolerance to lactose after pre-treatment with the prokinetic agent metoclopramide as compared with the placebo. A minor indication of lesser absorption of lactose was seen with an increase in the excretion of breath hydrogen and in the reduced excretion of urine galactose. Gastric emptying of a lactose solution after an overnight fast might be fairly rapid even without a small dose of metoclopramide. Perhaps the dose we used, which was chosen on the basis of previous studies (Massicotte et al 1996) and of the average therapeutic single dose recommendations (Dollery 1991), was not sufficient for all our subjects.
Even though we found no correlation between the total gastrointestinal transit time and the development of gastrointestinal symptoms, the possibility cannot be excluded that the reduction of gastric emptying also modified intestinal motility. In another study, the anti-diarrhoeal drug loperamide reduced intestinal transit in lactose maldigesters by about 30 min (Szilagyi et al 1996). This prolongation has been found not only to diminish but also to some extent to delay the development of gastrointestinal symptoms (Szilagyi et al 1996), but we were not able to confirm this.
Carmine dye and the subjective estimation of its appearance in the faeces may not have been a sufficiently sensitive way to detect gastric emptying reduction. This method for measuring gastrointestinal transit time was chosen because it is considered a fairly simple, cheap and reliable method (Read et al 1980, Marlett et al 1981) and was readily accepted by the study subjects. Even though there may have been little value in this method of measuring transit time, it may have increased the compliance of the study subjects since it is a good, and at the same time objective, means which can be observed by them personally. There are more exact methods for measuring gastric emptying, such as the radioscintigraphic technique, the 14C octanoic acid breath test, and the use of ultrasonography or radiopaque capsules, but since these methods are unsuitable for out-patients or need special equipment, a greater number of study periods or the collection of faeces over a long period, the estimation of transit time with dye was chosen.
We thus proved with a simple design, without the affecting factors of any other diet or milk components, that by retarding gastric emptying, tolerance to pure lactose was improved. This confirms previous observations of diminished symptoms after retarding gastric emptying by dietary modifications (Welsh and Hall 1977, Nguyen et al 1982, Solomons et al 1985, Martini and Savaiano 1988, Marteau et al 1990, Mahe et al 1994, Dehkordi et al 1995, Vesa et al 1997a, Vesa et al 1997b).
In order to investigate the possible role of inflammation in the symptoms of lactose intolerance, we compared the effects of lactose with a digestible disaccharide, sucrose. The synthesis of prostaglandins, as a marker of inflammation, was reduced by administering ibuprofen at the same time as lactose. Prostaglandins are synthesised via the cyclo-oxygenase pathway (COX) in response to stimuli which, in the gut, include such things as mechanical stimulation, cell trauma and inflammation (see Hawkey and Rampton 1985). Ibuprofen is a nonselective COX inhibitor, affecting both constitutive COX-1 and inducible COX-2.
In the urinary excretion of the prostaglandin metabolites PGE2-M and 6-keto prostaglandin F1α, an increase of about one third was noticed between the lactose and the sucrose intakes. By reducing the synthesis of prostaglanins by inhibiting the COX pathway with ibuprofen, this difference between a purely lactose intake and lactose ingested with ibuprofen was not eliminated. Nor were any differences seen in the plasma PGE2-M excretion. On the other hand, gastrointestinal symptoms and their severity increased after ibuprofen, both with lactose and with sucrose. Thus it seems probable that gastrointestinal symptoms after ibuprofen are not related to lactose ingestion but rather to COX1 inhibition, and thus they cannot be abolished by reducing the synthesis of prostaglandins, even though slight changes in the excreted concentrations of prostaglandin were noticed.
Hence our findings do not support the hypothesis that lactose-induced symptoms are related to an increased production of prostaglandins, as suggested previously by Buissert et al (1978), Lieb (1978) and Lieb (1980). In agreement with Flatz and Lie (1982) we found no relief in gastrointestinal symptoms after the inhibition of prostaglandin synthesis. Thus the role of prostaglandins in mediating the symptoms of lactose intolerance is minor, if it exists at all.
To test the role of another possible marker of intestinal inflammation, the enhanced production of nitric oxide, we measured the concentrations of its excreted metabolites after a lactose challenge. To our knowledge, no previous studies of the role of NO in lactose intolerance exist. In this study the effect of exogenous nitrate was reduced by restricting the dietary intake of nitrate for 48 h preceding the test, as suggested by Wennmalm (1995). In the excretions of the urinary concentrations of nitrate + nitrite (NOx), which is a stable metabolite of NO, and cyclic GMP, which is a second messenger of NO, no differences related to lactose ingestion were found.
A new method has been recently introduced to measure by chemiluminescence the luminal concentrations of gaseous NO from rectal gas samples, with no need for colonoscopy (Herulf et al 1999). In gastro-enteritis patients, rectal NO measurements taken this way were a hundred times higher than in the healthy controls (Herulf et al 1999). This method could also perhaps be a new means of investigating the possible role of NO in lactose intolerance. However, when measuring excreted metabolites, there is always the problem of whether the results refer to the production or to the metabolism of the substance being investigated.
Taken together, even though inflammation-related changes such as hyperemia and edema have been detected in jejunoscopies after a lactose challenge (Banai et al 1984), and even though the symptoms of lactose intolerance resemble those seen in inflammatory gastrointestinal diseases, our findings do not support the theory that inflammation is related to lactose-induced symptoms. The method we used to investigate this was not very refined, but if there had been enhanced production of prostaglandins or NO, which are well documented as being associated with chronic inflammation of the intestine (see Stark and Szurszewski 1992, Lefebvre 1995), this would almost certainly have been apparent with the study design and the methods used.
The effect of dietary modification on the expression or activity of the lactase enzyme was tested by supplementing the diet of 8-week-old rats with lactose for seven days. Both the expression and the activity increased compared to the control animals, who received no lactose. The induction of lactase was most obvious in the proximal and the middle parts of the jejunum. This accords with previous experimental studies in which the reduced activity of lactase found in adult animals was at most doubled by dietary modifications (Bolin et al 1969, Wen et al 1973, Goldstein et al 1974, Goda et al 1984, Thoreux et al 1998).
In human studies, on the other hand, the continuous intake of lactose did not modify the activity found in post-test biopsies (Cuatrecasas et al 1965, Newcomer and McGill 1967, Kreusch et al 1969, Gilat et al 1972). But these biopsies were only taken from the proximal small intestine before and after the study. As shown recently, the activity of lactase varies greatly on the villus enterocytes depending on the biopsy site (Maiuri et al 1992, Rossi et al 1997). Thus, one biopsy sample is not sufficient to describe the overall ability of the subject, improved or otherwise, to digest lactose.
To reduce the variability of the activity due to the sampling site, and to observe the variability between the different parts of the intestine, several samples were taken longitudinally the length of the small intestine. To describe better the total capacity to digest lactose we produced an indicator simply by adding the activities of single biopsies. When this indicator was compared proportionately to the control group we proved that lactase activity increased overall by about 35% in those animals who received the largest dose of lactose. This method could be further improved if the effects of age, the total length of the small intestine and the sampling site were noted in this indicator, for example by using suitable coefficients. In this study the effect of these factors was minimised by using age-, weight- and sex-matched animals.
The dose of lactose with which we showed the induction of the enzyme varied between 10-12 g/kg body weight/day. This accords with previous studies, as far as can be concluded from the original articles (Bolin et al 1969, Wen et al 1973, Goldstein et al 1974, Goda et al 1984, Thoreux et al 1998). In human studies with lactase activity measured from a biopsy, the daily doses have been about one tenth of this quantity (Cuatrecasas et al 1965, Newcomer and McGill 1967, Kreusch et al 1969, Gilat et al 1972). But the smaller dose of lactose, about 0.5-1 g/kg body weight/day, reduced the excretion of breath hydrogen and diminished gastrointestinal symptoms in lactose maldigesters after a few weeks of a lactose-containing diet (Johnson et al 1993b, Hertzler and Savaiano 1996, Briet et al 1997). At the same time faecal β-galactosidase activity doubled (Briet et al 1997) or even tripled (Hertzler and Savaiano 1996) compared with the control periods. These authors (Johnson et al 1993b, Hertzler and Savaiano 1996, Briet et al 1997) suggest that, rather than the induced activity of lactase, colonic bacterial adaptation was responsible. A similar inference was made after a continuous intake of indigestible lactulose (Florent et al 1985, Flourie et al 1993).
The results of our study thus confirm the hypothesis that, in an experimental model, lactase is induced by the continuous intake of large dose of lactose. This can also be seen in the Western Blot analysis of intestinal homogenates from rats which received milk for several weeks (unpublished data). Probably because of the far smaller doses used, increased activity has not been demonstrated with humans. Continuous, more normal doses of lactose, less than 1 g/kg body weight, have, however, reduced gastrointestinal symptoms and the excretion of breath hydrogen and increased faecal β-galactosidase activity (Johnson et al 1993b, Hertzler and Savaiano 1996, Briet et al 1997), all of which are indicators of a better tolerance to lactose.
Taking conjointly our three hypotheses of the possible tools for reducing lactose-induced symptoms, it seems that, rather than being of inflammational origin, lactose intolerance is, at least partly, the result of motility disorders. This has also been suggested by Hammer et al (1996), who emphasised the importance of transit time in the occurrence of the symptoms of lactose intolerance, rather than the importance of the amount of malabsorbed lactose. If either gastric emptying or gastrointestinal transit time or both can be retarded, the reduced level of the lactase enzyme in the intestinal epithelium of a lactose maldigester may be in better state of balance with the normal dietary intake of lactose. This reduction of motility could be achieved by any factors, dietary or otherwise, which may retard gastric emptying and/or reduce intestinal motility. Continuous daily intakes of lactose may further improve tolerance to lactose by inducing the lactase enzyme, or at least by adapting the colonic bacteria so that they ferment undigested lactose in the colon more effectively and thus reduce the development of gastrointestinal symptoms.
Interestingly, the self-diagnosis of lactose intolerance gave poor results. Only one third of the self-diagnosed subjects proved to be lactose maldigesters, and some of them, indeed, proved to be asymptomatic. Our findings accord closely with previous studies in which lower doses of lactose (15-25 g) were used (Rosado et al 1987, Johnson et al 1993a, Suarez et al 1995, Carraccio et al 1998, Saltzman et al 1999). In these earlier studies the actual number of symptomatic lactose maldigesters among self-reported maldigesters varied between 10% and 70%. Since the incidence of true maldigesters in our studies and studies by other authors varies greatly, probably depending mainly on the inclusion and exclusion criteria of the study subjects, this phenomen indicates the very real need for carefully controlled clinical lactose tolerance tests.
We also found that about one third of previously tested lactose intolerant subjects were actually lactose digesters, according to our Golden Standard. This large ratio of changed status in tolerance to lactose could be explained either by a truly changed status due to the high incidence of secondary hypolactasia, or by an incorrectly made diagnosis, either an earlier one or the new one.
The incidence of secondary hypolactasia, as well as many probable factors resulting in it by damage to the epithelial mucosa or by the modification of gastrointestinal motility, is not known. As the severity of inherited hypolactasia may vary, so the severity of secondary hypolactasia may also vary according to the power and the length of the challenge that caused it. Lactase is sensitive to factors harmful to the mucosa, because it is located on the brush border of the mature enterocytes on the top of the villi, and is therefore more exposed than other brush-border enzymes to various factors affecting the jejunal mucosa (see Villako and Maaroos 1994).
There are several factors that can lead to an incorrect diagnosis. We tried to minimise these by using the combination of three independent test variables, and by controlling those factors which may affect gastrointestinal motility, or may affect the symptoms, such as dietary intake and the prior use of medication. We also used the standard dose of lactose (50.0 g) dissolved in the standard volume of boiling water.
In Finland the lactose tolerance test is usually performed with 50 g lactose dissolved in 200-400 ml water after an overnight fast (Peuhkuri et al 2000a). This dose equals the lactose of a whole litre of milk and cannot be assumed to imitate the normal symptoms produced by dietary lactose. The dose of 50 g is suitable for detecting lactose maldigestion, but too abnormal for detecting lactose intolerance. There are studies that describe the use a dose of 25 g with breath hydrogen analysers for detecting lactose maldigesters (see Hamilton 1998), but more detailed studies of an appropriate dose are needed in order to find more effectively those subjects who really benefit from the restriction of lactose in their diets.
An exact recommendation for the volume of water in which lactose should be dissolved is also needed. As the volume ingested affects gastric emptying (see Cooke 1975, Malagelada 1990), it can be assumed that tolerance to lactose tested with a volume of 400 ml differs from a test performed with a smaller volume. The optimum volume depends on the dose of lactose used. In principle, 1 g lactose can be dissolved in 5 ml room-temperature water or in 2.6 ml boiling water. If the lactose is dissolved in boiling water, a proper cooling to room temperature may improve the reliability of the test, since gastrointestinal symptoms produced by the room-temperature solutions are minor compared with those produced by cold or hot solutions, as we have shown. The volume used may be taken from everyday drinking patterns. 'A glass' (200 ml) seems to be the natural volume that can be swallowed without distaste and without the symptoms induced by the larger volume of liquid after an overnight fast.
We found that, compared to the Golden Standard, on its own none of the test variables used in this study was particularly good for recognising both the lactose maldigesters and the digesters, though on average, the sensitivity and the specificity of the variables used were good. The urine galactose test showed the best sensitivity and the breath hydrogen test the best specificity. The total number of misdiagnosed subjects was at most 9% if the blood glucose test was used on its own. However, it should be borne in mind that the predictive value of each test depends on the prevalence of lactose maldigestion in the study group. In our study the incidence of lactose intolerance was much higher than in a normal cross-section of the Finnish population.
If these two variables are taken in conjunction, the reliability of the lactose tolerance test may improve. The best combination may be to consider the measurements both of excreted breath hydrogen and of urinary galactose. The urinary galactose test showed the highest sensitivity, recognising 100% of the lactose maldigesters; the breath hydrogen test was best for specificity, detecting 94% of the lactose digesters. It would seem possible to look at the results of these two tests together, since a three-hour urine collection is recommended for excreted galactose measurement (Arola 1988b; Grant et al 1989, Buttery and Ratnaike 1995, Alvarez-Coca et al 1996) and it is recommended that breath hydrogen measurement should be continued until the baseline value is exceeded by at least 20 ppm or for up to three hours (see Hamilton 1998). The combined 13CO2/H2 breath test is in its present form far too cumbersome. If, however, this could developed further, improving both sensitivity and specificity, simplifying the equipment needed, and reducing costs, this could offer a suitable method in the future.
When diagnosing lactose intolerance (symptomatic lactose maldigestion), the recording of gastrointestinal symptoms should not be disparaged. Both the severity and the intensity of symptoms increase for several hours from the ingestion of the lactose solution, the maximum symptom scores predicted being achieved some hours after the beginning of the test. The most common symptoms, both those that were expected and those actually experienced, seem to be flatulence, borborygmi and abdominal bloating. The recording of symptoms during a lactose tolerance test should be based on the true incidence of single symptoms and should last several hours, preferably of minimum of three.
As was seen in our study, both those who are and those who suppose themselves to be lactose maldigesters expect to experience severe symptoms if they drink a glass of milk. A lactose tolerance test as such may thus, in some subjects, increase the severity and the number of gastrointestinal symptoms. One possible mechanism for this may be via increased intestinal motor activity induced by the stress factor, which in this case is the test situation itself. The modifying effect of emotional factors on intestinal motility may also explain the variability of the symptoms of lactose intolerance, as has also been shown with gastrointestinal symptoms in IBS patients (see Stam et al 1997). This may be avoided or reduced by using a blinded test, as we did whenever possible. This blinding could be achieved by using lactose-free milk or another indigestible disaccharide such as lactulose.
To conclude, in the diagnosing of lactose intolerance, based on our results and those of other investigators, any of the laboratory test variables used for detecting lactose intolerance is superior to self-diagnosing. A breath hydrogen measurement together with a urine galactose test after ingestion of a room-temperature test solution in a natural volume of water is to be recommended. The possibility of reducing the dose of lactose from the prevalent 50 g to a more normal dietary level should be studied further. The recording of gastrointestinal symptoms is an essential part of the tolerance test, even if on its own it has little value in diagnosing lactose intolerance.