Worldwide, over 5 billion humans are estimated to suffer from lactase (LCT) deficiency, which is the most common cause of lactose intolerance. The geographical distribution of lactose intolerance is variable depending mostly on heritage: in Asian countries, close to 100 % of the general population is considered lactose intolerant, whereas in Northern Europe, prevalence reaches about 10–20 % increasing towards the south(Reference Sahi1). Lactose is a disaccharide physiologically cleaved into glucose and galactose by LCT located on the luminal membrane of the enterocytes located in the proximal small bowel. Reduced expression and/or activity of LCT leads to an accelerated intestinal transit and degradation of lactose by colonic bacteria into SCFA, H2 and other metabolites implicated in the pathophysiology of classical abdominal symptoms such as diarrhoea, bloating, borborygmi and abdominal pain.
Most laboratories employ a lactose H2-breath test (H2-BT) in diagnosing LCT deficiency (LD). This test exploits the fact that H2, a by-product of bacterial lactose metabolism, is absorbed through the intestinal mucosa into the venous blood, and transported past the liver into the alveoli of the lungs, from where it is exhaled(Reference Arola2, Reference Launiala, Kuitunen and Visakorpi3). Several studies correlating the amount of exhaled H2 with LCT deficiency documented a good sensitivity (76–94 %) and specificity (77–96 %) of the BT(Reference Arola2).
In 2002, Enattah et al. (Reference Enattah, Sahi and Savilahti4) described single-nucleotide polymorphism 13 910 bp above the structural gene coding for the enzyme LCT on the short arm of chromosome 2q.21-22. This polymorphism consists of the nucleotide switch of T for C, resulting in variants CC, CT or TT− 13 910. Enattah demonstrated CC− 13 910 to be a good predictor of intestinal LCT activity loss. A recent study(Reference Högenauer, Hammer and Mellitzer5) showed a lower sensitivity of genetic testing of only 75 % when compared with H2-breath testing based on ingestion of lactose.Other genetic polymorphisms, with the most prominent one being G/A− 22 018 identified in European and Asian subjects, have also been shown to be of importance, but they have not been commercially marketed to date(Reference Enattah, Sahi and Savilahti4). Recently, polymorphisms especially in the sub-Saharan African population have demonstrated local gene traits(Reference Enattah, Jensen and Nielsen6). Genetic tests fail to detect secondary causes of relative LCT deficiency such as inflammatory or malabsorptive bowel disease, in which reduced mucosal absorptive area, insufficient contact time between substrate and enzyme, and also reduced mucosal expression of LCT lead to incomplete degradation of lactose(Reference Kirschner, DeFavaro and Jensen7). Besides classical symptoms of lactose intolerance, extra-intestinal symptoms such as headache, fatigue and muscle pain have been suggested to occur in response to ingestion of LCT(Reference Chaudhuri8–Reference Matthews, Waud and Roberts10).
The aims of the present study were (1) to compare the results of H2-BT against those of the genetic test; (2) to evaluate the utility of collecting data on intestinal and extra-intestinal symptoms for up to 48 h after oral ingestion of lactose.
Methods
Patients
The present study was conducted according to the guidelines laid down in the Declaration of Helsinki. All study subjects provided written informed consent before being enrolled in the present study. The study was approved by the Ethics Committee of the University Hospital Zurich and of the Canton of Zurich (EK-1225). Consecutive patients referred to our tertiary referral centre for suspected lactose intolerance were included in the study. Subjects were required to be at least 16 years old. Patients with antibiotic use in the past 2 weeks before the procedure, with active chronic or acute inflammatory bowel disease and with drug or alcohol abuse were excluded from the study.
Lactose hydrogen breath test
To minimise basal H2 values, patients were instructed to consume a carbohydrate-deficient diet (especially avoiding beans, soya products, lentil, peas and whole-grain breads) 1 week before the testing, and to fast at least 12 h before the test. Smoking and physical activity were to be avoided at least 1 h before the test and during the entire test procedure. The oral cavity was decontaminated with 2 × 20 ml of Hextril solution (Pfizer, Zurich, Switzerland). End-expiratory H2 values were measured offline using a stationary H2 detection system (Acutronic Medical Systems, Hirzel, Switzerland). For adequate sample collection, patients were instructed to perform two empty breath manoeuvers before each H2 breath sample collection in order to remove dead air space in the respiratory tree. Patients then exhaled normally, and samples were collected mid-expiration using 20 ml plastic syringes. After determining the baseline H2 concentration, patients received 50 g lactose powder, or an equivalent of 1 mg/kg body weight was given for patients weighing less than 50 kg. The lactose powder (Kantonsapotheke Universitätsspital Zürich, Switzerland) was dissolved in 150–200 ml of water. Mid-expiratory H2 breath samples were collected every 15 min for 4 h. At each time point, two breath samples were drawn from each patient, the higher of which was used for analysis. The collected air samples were analysed within 5 min after collection.
The lactose H2-BT was considered positive (i.e. indicative of LCT deficiency) if there was a rise of at least 20 ppm in H2 concentration above baseline in at least two consecutive measurements, starting minimum 30 min into the test.
Symptom questionnaire
Patients were asked to rate five typical symptoms of lactose intolerance (nausea, bloating, diarrhoea, borborygmi and abdominal pain) and four extra-intestinal symptoms (headache, dizziness, fatigue, and muscle and joint pain) on a 9-point Likert symptom scale ranging from 1 (non-existent) to 9 (worst ever). For the duration of the study symptoms, rating was done every 15 min concomitant with H2 measurements. After this, patients were asked to rate both intestinal and extra-intestinal symptoms every 4 h for 48 h after the intake of lactose. Patients were asked to omit values during their sleeping period to avoid any influence of sleep-disturbance effects on symptom recording. Patients left the hospital after 4 h, and they were instructed to refrain from lactose intake (after being given dietary instruction including hidden lactose) for 48 h. Patients were provided pre-stamped envelopes to mail the completed symptom questionnaire to our institute.
Lactulose hydrogen breath test
In patient with discrepant results between genetic test and H2-BT, a lactulose test was performed either to exclude bacterial overgrowth in those with positive H2-BT and negative genotype (see below), or to exclude H2 non-secretory status in patients with negative H2-BT and positive genotype (see below). Pre-test preparation was identical to that of the lactose BT. Patients were asked to drink 25 g of lactulose powder dissolved in 150–200 ml water, and mid-expiratory samples were taken every 15 min for 4 h. The test was considered positive for bacterial overgrowth when there was a rise in H2 peak concentration of more than 20 ppm above baseline at least 15 min before measurement of a colonic peak of at least 10 ppm above baseline(Reference Kerlin and Wong11). Non-excretion indicating a lack of H2-producing intestinal flora was considered in patients with a rise not exceeding 10 ppm above baseline.
Genetic testing
Venous blood samples were drawn from patients into EDTA tubes (Braun, Melsungen, Germany). Blood samples were stored refrigerated at 4°C for a maximum of 5 d before genetic analysis. Genotyping of the LCT C/T− 13 910 polymorphism was done in the diagnostic laboratory of the Institute of Clinical Chemistry of the University Hospital Zurich using melting curve analysis on a Roche LightCycler (Rotkreuz, Switzerland) according to the previously described method by Stolba et al. (Reference Stolba, Rezanka and Eckhard12). Accuracy and robustness of the assay were evaluated in the laboratory using sequenced control DNA in repeated experiments.
Statistics
Sensitivity, specificity, and positive and negative predictive values were calculated between results from H2-BT and genetic testing. Agreement was calculated using Cohen's κ. Symptom intensity was compared using ANOVA with post hoc Bonferroni's correction for multiple testing. Differences were considered statistically significant if P < 0·05.
Results
Patient demographics
In the present study, 201 patients were included. Seven patients who had high basal H2 values or had denied genetic testing after consenting to the study were excluded from the analysis. Remaining patients (total 194: 139 females, mean age 38, range 17–79 years, and 55 males, mean age 38, range 18–68, P = 0·926) referred for testing of suspected lactose intolerance underwent a lactose H2-BT and genetic testing for LCT C/T− 13 910. Patients were largely Northern European (154 Northern Swiss and German, and 5 Eastern Europeans), and furthermore, twenty-six patients were Southern Europeans (including Southern Switzerland), eight were of Asian descent (one Chinese and seven Indian) and seven patients were from Southeast Europe and the Middle East.
Lactose hydrogen breath test
Of all the patients referred for the evaluation of lactose intolerance, sixty-three patients (32·5 %) tested positive for LCT deficiency based on exhaled H2 values after ingestion of lactose (Table 1). There were no age (P = 0·226)- or sex (P = 0·865)-specific differences between the patients with normal LCT activity and LCT deficiency.
Table 1 Genotype* and lactose hydrogen breath test (H2-BT) results
* Sensitivity of genotype C/T13 910 compared with H2-BT was 97 %, specificity was 95 %, positive predictive value was 90 % and negative predictive value was 98 %. Cohen's κ for agreement was 0·9.
Genetic test v. lactose hydrogen breath test
Sixty-eight (35·1 %) patients were CC− 13 910 homozygotes, which is the genotype associated with LCT deficiency. Another sixty-eight (35·1 %) patients were CT− 13 910 heterozygotes, and fifty-eight (29·9 %) patients were TT− 13 910 homozygotes indicating normal LCT activity (Table 1).
There was an excellent agreement between the genetic analysis and the H2-BT with only nine (4·6 %) discrepant results. Using the lactose H2-BT as the gold standard, only two patients were misclassified as LCT sufficient (i.e. false negative) in the genetic test, leading to an excellent sensitivity and negative predictive value in diagnosing LCT deficiency (Table 1). Overall, the genetic analysis identified 35·1 % of the patients as LCT deficient, while the H2-BT identified 32·5 % as LCT deficient.
Discrepant results – lactulose breath test
Of the 194 patients included in the analysis, nine (4·6 %) patients had discordant results between lactose H2-BT and genetic analysis. These patients were clinically reassessed, and eight patients agreed to undergo the lactulose H2-BT. Table 2 and Fig. 1 show individual patient exam stratification and results.
Table 2 Discordant patient details
H2-BT, hydrogen breath test; LI, lactose intolerance, LT, lactose tolerance.
Fig. 1 Flowchart showing patient results and stratification. 
, Diagnosis; 
, discrepant results. LD, lactase deficiency; H2-BT, hydrogen breath test.
Seven of nine (78 %) patients with initially discrepant test results were CC− 13 910 homozygotes indicating LCT deficiency, but they had a negative lactose BT.
Upon focused re-assessment, three patients remembered having used antibiotics shorter than 4 weeks before the test (Helicobacter pylori eradication, chronic urinary tract infection), which could have interfered with the test, leading to a false-negative H2-breath test. However, when tests were performed 2 and 4 months after the initial lactose breath test, two of these three patients who underwent a lactulose breath test exhibited a hydrogen-producing colonic flora, possibly reflecting restoration of formerly suppressed colonic flora. Two patients were identified as non-H2 producers by lactulose BT. One patient had a H2 elevation of more than 20 ppm from baseline at only one time point, very little symptoms and only a slight elevation during the lactulose BT. Finally, one patient had a false-positive genetic test as all other test results spoke in favour of lactose deficiency: non-significant H2 elevation, absence of symptoms and a positive lactulose BT indicating H2-producing flora.
Among the two patients with genotypes indicating normolactasia but abnormal lactose BT, one patient had small intestinal bacterial overgrowth and the other patient had intercurrent bouts of inflammatory bowel disease resulting in secondary lactose intolerance.
Symptom analysis – overall interpretation
There were no differences between sexes in symptom distribution (all P>0·05). Patients (both males and females) with genotype CC− 13 910 were symptomatic in 75 % of the cases. Thirty-eight percentage of women were symptomatic with genotype CT− 13 910 v. 24 % of men, and 38 % of women were symptomatic with genotype TT− 13 910 v. 33 % of men.
Symptom analysis – intestinal symptoms
A completed 48 h symptom questionnaire was available from 95 % of the patients (185/194). Patients with CC− 13 910 homozygous genotypes (sixty-four complete symptom sets available) reported more intense intestinal symptoms (borborygmi, bloating, abdominal pain and diarrhoea) than patients with CT− 13 910 heterozygous genotypes (fifty-five complete symptom sets available) and TT− 13 910 homozygous genotypes (fifty-six complete symptom sets available). Symptoms became intense 45 min after lactose intake and lasted up to 12 h. Peak intensity was reached after 4–8 h of lactose ingestion as indicated in Fig. 2.
Fig. 2 Intestinal symptom scores over 48 h according to genotyping for lactase C/T− 13 910. (a) Borborygmi, (b) bloating, (c) abdominal pain, (d) diarrhoea and (e) nausea. 
, Genotype CC; –▲–, genotype CT; –■–, genotype TT. * Mean values were significantly different (P < 0·05).
There was a fair agreement between symptomatic hydrogen breath test and CC− 13 910 genotype (κ 0·365, P < 0·001). Of the patients with a positive genetic test, 75 % (fifty-one) with genotype CC13 910 were considered symptomatic compared with 33·8 % (twenty-three) with genotype CT13 910 and 36·2 % (twenty-one) with genotype TT13 910 (P < 0·01).
Symptom analysis – extra-intestinal symptoms
Patients with genotypes indicating LCT deficiency reported similar intensity of dizziness and muscle and joint pain compared with those with genotypes indicating normal LCT activity. Patients with genotype CC reported more intense fatigue 4–12 h after ingestion of lactose compared with those with genotype CT/TT. The intensity of extra-intestinal symptoms are summarised in Fig. 3.
Fig. 3 Extra-intestinal symptom scores over 48 h according to genotyping for lactase C/T− 13 910. (a) Fatigue, (b) dizziness, (c) headache and (d) muscle/joint pain. , Genotype CC; –▲–, genotype CT; –■–, genotype TT. * Mean values were significantly different (P < 0·05).
Discussion
In the present study, we have reported on the relationship between genetic testing and lactose H2-BT in patients evaluated at a Swiss tertiary referral centre. Additionally, in the present study, we have also reported on the intensity of intestinal and extra-intestinal symptoms characteristic for lactose intolerance in LCT-deficient and non-deficient individuals.We found an excellent agreement, sensitivity and specificity between results from testing for genetic polymorphism LCT C/T− 13 910 and lactose H2-BT. In patients with discrepant results, a careful history and lactulose H2-BT clarified the diagnosis. With regard to intestinal symptoms, patients with LCT deficiency reported higher symptom intensity for 2–10 h after oral ingestion of lactose for all symptoms except for nausea. LCT-deficient patients and those with normal LCT activity reported similar intensities of extra-intestinal symptoms with the exception of fatigue. Increased fatigue 4–16 h after oral ingestion of lactose was most likely the result of the increased intestinal symptoms rather than being directly related to lactose intolerance.
The present results are consistent with a previous report by Krawczyk et al. (Reference Krawczyk, Wolska and Schwartz13) who evaluated fifty-eight patients for lactose intolerance in a German tertiary referral centre. They found a 100 % agreement between lactose H2-BT and genotype CC− 13 910. Considering the genetic test as the gold standard, they reported an excellent sensitivity and negative predictive value of the H2-BT.
Nine patients (4·6 %) in our group had discrepant results in the lactose H2-BT. The most common reason (33 %) for the discrepant results was the prior use of antibiotics leading to suppressed intestinal flora and a ‘non-H2 producer’ status.
Few of our patients had a false-positive result. Colonisation of the proximal intestinum with H2-producing bacteria termed small intestinal bacterial overgrowth can lead to false-positive BT results(Reference Pimentel, Kong and Park14). We diagnosed one patient with small intestinal bacterial overgrowth; the other patient's positive lactose H2-BT despite negative genetic test was most likely attributable to intermittent bouts of inflammatory bowel disease, which was not recognised at study inclusion(Reference Kirschner, DeFavaro and Jensen7).
Högenauer et al. reported a lower sensitivity (75 % opposed to 97 % in our subjects) in the detection of LCT deficiency when comparing genetic test and lactose H2-BT as the gold standard(Reference Högenauer, Hammer and Mellitzer5). Differences in test design (in our study patients had to reach a δ H2 peak above 20 ppm in at least two consecutive measurements) and studied population have to be taken into account when interpreting these data. Similar to our findings, Högenauer reported a high specificity in diagnosing LCT deficiency in patients with genotype CC− 13 910. One patient with negative lactose H2-BT and normal lactulose H2-BT had a positive genetic test. This may be explained by LCT persistence into young adulthood as shown in a recent two-decade follow-up study in a Finnish population(Reference Seppo, Tuure and Korpela15). Furthermore, epigenetic modulation of the DNA with consecutive changes in functional properties or gene–gene interactions have to be discussed, and since we did not test for other genetic polymorphisms possibly associated with LCT persistence such as G22 018, C− 14 010, G− 13 907 and G− 13 915, we cannot rule them out(Reference Enattah, Sahi and Savilahti4, Reference Tishkoff, Reed and Ranciaro16). A second patient with negative H2-BT, few symptoms and slightly elevated breath H2 concentrations had a positive genetic test. This patient was considered a low-H2 producer, but transition into LCT deficiency as mentioned earlier cannot be excluded. A technical limitation in the diagnosis of the LCT C/T− 13 910 genotype has been discussed in systems employing LightCycler diagnosis and melting curve analysis(Reference Weiskirchen, Tag and Mengsteab17). However, since our population was largely European and five of seven patients with genotype CC13 910 and negative BT were clearly symptomatic indicating lactose intolerance, we believe that this should not affect our diagnoses and derived statements. Adding to the finding of previous investigators who compared genetic testing and phenotype, our data show a high negative predictive value of 98 % for genetic testing, supporting a low likelihood of LCT deficiency in patients with genotypes CT and TT13 910.
Symptom analysis is an important part of establishing a clinical diagnosis of lactose intolerance. Symptoms typically attributed to lactose intolerance include abdominal pain, nausea, diarrhoea, bloating and borborygmi. In addition to these, it has been suggested that systemic extra-intestinal symptoms such as headache and light headedness (up to 86 %), loss of concentration and short-term memory loss (82 %), fatigue (63 %) and muscle and joint pain (71 %) as well as allergic reactions (40 %, including eczema, pruritus and airway reaction) could be the results of LCT deficiency(Reference Chaudhuri8–Reference Matthews, Waud and Roberts10). In addition to the typical intestinal symptoms, we chose four extra-intestinal symptoms (headache, fatigue, dizziness, and muscle and joint pain) and a time frame of 48 h to explore symptoms which may occur late after oral ingestion of lactose.
For typical intestinal symptoms, the results were consistently different for all the symptoms except for nausea. Patients with genotype indicative of LCT deficiency had more intense symptoms starting from 45 min (bloating) to 90 min (diarrhoea) and lasting 8 h (abdominal pain) to 28 h (borborygmi) after oral ingestion of lactose. Roughly one-third of our patients with lactose persistence had typical symptoms of lactose intolerance although showing LCT persistence. Interestingly, one-quarter of the patients without adequate LCT production remained largely asymptomatic. It has been shown that there is a considerable overlap between irritable bowel syndrome and lactose intolerance plus the placebo effect as patients may have been biased to test positive for different reasons. On the other hand, it has been shown that most patients with LCT deficiency do not develop symptoms after intake of usual servings of lactose (12·5 g as in one cup of milk)(Reference Savaiano, Boushey and McCabe18). Even the considerably larger dose of 50 g, which is considered adequate in a Northern European population, only showed a sensitivity of 75 % to detect hypolactasia by symptoms. This underlines the individual perception of symptoms and may also be caused by a certain tolerance of larger lactose doses, as portion sizes increase around the globe(Reference Nielsen and Popkin19), and evidence exists that patients with lactose intolerance can adapt over time to larger doses of lactose by adaptation of their microflora(Reference Hertzler and Savaiano20).
There was no difference in symptom expression between genotypes CT13 910 and TT13 910 at any time point. Conversely to this, Matthews et al. (Reference Matthews, Waud and Roberts10) reported differences in sensitivity of H2 lactose testing between genotypes CT and TT13 910 in a large patient population. The reason for this discrepancy is unclear, and it may be due to patient population as they examined patients from an irritable bowel syndrome cohort possibly suffering in part from underlying motility disorder or small intestinal bacterial overgrowth leading to secondary lactose intolerance of false-positive results. Furthermore, the authors performed 6 h H2 recordings, whereas we tested patients for a maximum of 4 h. Though our time window of symptom assessment of 48 h was clinically important, in our patient population clinically suspected to have lactose intolerance in the first 6 h or later up to 48 h, differences in neither intestinal symptoms nor extra-intestinal symptoms were noted based on difference in genotypes CT and TT13 910. Kuokkanen et al. (Reference Kuokkanen, Enattah and Oksanen21) noted different levels of intestinal brush border LCT with 4–9 U/g in CC− 13 910 carriers and with 13–49 and 18–87 U/g in CT and TT carriers, respectively. We believe that the observed large overlap in LCT expression between CT and TT− 13 910 carriers explains the similar symptom pattern in patients, and that it does not seem to translate into clinically significant differences in our patient population. This is in accordance to the data reported by Högenauer et al. (Reference Högenauer, Hammer and Mellitzer5), although they showed a higher percentage of CT-carrying patients than of TT− 13 910-carrying patients with lactose H2-BT indicating LCT deficiency.
Extra-intestinal symptoms did not yield different results based on genetic testing for LCT deficiency either during the initial 4 h window, where symptoms were quantified every 15 min, or in the 44 h thereafter. Only fatigue was noted more often in patients with LCT deficiency 4–16 h after ingestion. This may be due to a vagal nerve system response due to increased intestinal activity with respect to the osmotic challenge by undigested lactose or a reaction to typical intestinal symptoms during the first hours of the test that may be painful and stressful in patients with lactose intolerance.
The long duration of classical symptoms was somewhat surprising, as from a mechanistic standpoint, undigested lactose passes quicker through the intestine due to its osmotic properties. This may be explained by two mechanisms. First, apart from lactose residues still causing mechanical symptoms, the concept of a local reaction via impairment of the endoplasmic reticulum in the intestinal wall after provocation with large amounts of lactose emerged, which may be comparable to changes induced by viral infections(Reference Naim and Naim22). Secondly, direct effects on the mucosa comparable to those caused by bacterial enterotoxins have been discussed, but have not been validated yet(Reference Matthews, Waud and Roberts10, Reference Campbell, Wann and Matthews23).
In conclusion, our data show that the testing of LCT C/T− 13 910 genotype could become the first-line test for investigating patients with symptoms suggestive of lactose intolerance. The length and self-limitation of symptoms provoked by a lactose H2-BT are clinically important information to be shared with the patients, and if needed, pain and/or antispasmodic medication should be provided. Extended symptom questionnaires do not yield meaningful information on patients with LCT deficiency.
Acknowledgements
The authors would like to thank Brigitte Gabathuler, Esther Karlen and Diana Jovanovic for their careful execution of H2-BT and blood sampling. The contributions of the authors are as follows: D. P. was involved in the study design, grant application, execution, data analysis and writing of the manuscript. E. S. was involved in the data analysis and writing of the manuscript. M. H. was involved in the study design, execution and finalisation of the manuscript. Z. B. was involved in the data analysis and finalisation of the manuscript. B. S. was involved in the study design and data management. O. G., A. v. E. and M. F. were involved in the study design and finalisation of the manuscript. R. T. was involved in the data analysis and writing of the manuscript. The authors have no conflicts of interest. The present study was supported by the Swiss Foundation for Nutritional Research (Schweizerische Gesellschaft für Ernährung; SGE), grant number 358.
