Gastric secretion does not affect the reliability of the 13C-acetate breath test: A validation of the 13C-acetate breath test by magnetic resonance imaging


Address for Correspondence
Andreas Steingoetter, Division of Gastroenterology and Hepatology, University Hospital Zurich, Raemistrasse 100, CH-8091 Zurich, Switzerland.
Tel.: +41 44 255 5684; fax: +41 44 255 4503;


Background  13C-Acetate labeled meals are widely used to determine meal emptying by means of analyzing resulting 13CO2 exhalation dynamics. In contrast to the underlying metabolic processes, only few 13C breath test meal emptying studies have focused on intragastric processes that may alter 13CO2 exhalation. This work assessed the effect of enhanced gastric secretion on the reliability of half emptying time (t50) measurements by 13C-acetate breath test.

Methods  13CO2 exhalation data were acquired in a double-blind, randomized, cross-over gastric emptying study in 12 healthy volunteers receiving either pentagastrin or placebo intravenously. The standard method proposed by Ghoos et al. was applied to calculate t50 (t50_Ghoos) from 13CO2 exhalation data, which were compared and tested for agreement to meal half emptying times (t50_MV) from concurrent recorded MRI (magnetic resonance imaging) volume data. In addition, the accumulated gastric secretion volumes during infusion as detected by MRI (AUC_SV60) were correlated with the corresponding cumulative percent 13C doses recovered (cPDR60).

Key Results  t50_Ghoos and t50_MV showed a linear correlation with a slope of 1.1 ± 0.3 (r2 = 0.67), however, a positive offset of 136 min for t50_Ghoos. No correlation was detected between AUC_SV60 and cPDR60 (r2 = 0.11). Both, breath test and MRI, revealed a prolonged t50 under pentagastrin infusion with median differences in t50_Ghoos of 45[28–84] min (P = 0.002) and t50_MV of 39[28–52] min (P = 0.002).

Conclusions & Inferences  This study suggests that 13CO2 exhalation after ingestion of a 13C-labeled liquid test meal is not affected by stimulated gastric secretion, but is rather reflecting the dynamics of meal or caloric emptying from the stomach.


magnetic resonance imaging








half emptying time


breath test-derived gastric content half emptying time (min)


MRI-derived gastric meal half emptying time (min)


percentage dose recovered (% h−1)


cumulative percentage dose recovered (%)


gastric content volume (mL)


meal volume (mL)


secretion volume (mL)


contrast agent


area under the curve of secretion volume during infusion period (mL*min).


Breath tests by means of stable 13C-isotopes have increasingly gained importance in hepatic and gastroenterological functional diagnostics. They are attractive, non-invasive, and multifunctional tools used for different clinical applications such as the quantification of organ functions or the determination of gastrointestinal transport.1

To measure gastric emptying, breath tests with 13C labeled fatty acids, such as 13C-acetate, 13C-octanoic acid, or naturally 13C-enriched Spirulina platensis have been proposed to replace radioscintigraphy, which is still the gold standard.2 Ingested 13C-substrates are emptied from the stomach, absorbed in the small intestine, undergo catabolism in the liver, enter the body’s bicarbonate pool, and then are excreted as 13CO2 in the breath, where 13C can finally be detected.2 Assuming that all these steps are intact, the rate-limiting step in this process is the gastric emptying. Validation studies have shown that the performance of breath tests is not affected by variation in hepatic fatty acid oxidation3,4 or metabolic and absorptive changes in critical disease, diabetes mellitus, liver cirrhosis, and Crohn’s disease.5 Furthermore, it has been demonstrated that intestinal marker delivery is accurately and promptly reflected by 13CO2 exhalation and that 13CO2 formation is not influenced by calorie-dependent digestive processes.6 In contrast, the 13C breath test is not fully validated for intragastric influences on 13C-marker metabolism or distribution that could lead to a misinterpretation of gastric emptying measurements. Gastric secretion is stimulated by nutrient intake and known to influence gastric emptying.7 Large amounts of gastric secretion may conceivably lead to dilution and redistribution of the 13C-acetate within the test meal, presumably resulting in alterations of 13CO2 exhalation and thus systematic errors in computed half emptying time. Currently, magnetic resonance imaging (MRI) is the only non-invasive gastric emptying measurement method to separate meal and secretion from overall gastric content and thus to quantify meal emptying, gastric secretion volume, and related intragastric distribution and mixing.6,8–10

This work aimed to investigate the effect of enhanced gastric secretion on the reliability of meal emptying measurements by 13C-acetate breath test. To this end, concurrent measurements of 13C exhalation data and of MRI volume data of gastric secretion and meal from a recent randomized placebo-controlled cross-over study in subjects with and without pentagastrin-stimulated gastric secretion were analyzed and compared.



This analysis used data acquired during a recently published study by Goetze et al.11 The study was carried out according to Good Clinical Practice and the Declaration of Helsinki. The study protocol was approved by the local Ethics Committee and the Swiss National Agency for Therapeutic Products (registration numbers 1152 and 2005dr2207). All measurements were completed without complications or adverse events.

Test meal

A sterilized glucose solution (500 mL, 200 kcal; Fresenius Kabi AG, Stans, Switzerland) was mixed homogeneously with 5 g locust bean gum powder (Rapunzel Naturkost AG, Legau, Germany). The test meal was labeled with 1200 μmol L−1 of the paramagnetic contrast agent Gadolinium-DOTA (Dotarem®; Laboratoire Guerbet, Roissy CDG, France) and 200 mg L−1 of [1–13C] sodium acetate (Euriso-Top, Saint-Aubin Cedex, France). The mixture was ingested at 37 °C, where it had a viscosity of 202 mPa s−1 and an osmolality of 505 mOsm L−1.

Volunteers and study design

Primary outcome of this work was the difference in meal emptying half time between 13C-acetate breath test and MRI. The sample size of 12 participants was determined based on the study of Treier et al.,12 a two-sided 5% significance level and a power of 90% allowing for the detection of a difference of 15 mL in gastric secretion volume. Eligible participants were healthy adults aged 18–55 years, recruited via public announcement. Exclusion criteria were pregnancy and lactation, previous abdominal surgery (except appendectomy), intake of any medications apart from oral contraceptives, significant abnormality on a 12-lead ECG, claustrophobia and presence of metallic implants, devices or foreign bodies. Twelve healthy volunteers (seven males) with a mean age of 28 years (22–34 years) and a mean body mass index of 23.2 kg m−2 (20.3–28.0 kg m−2) were investigated between February and May 2006 in a single center, double-blind, randomized, placebo-controlled and cross-over study design. Randomization was assured by a computer generated random list and stratified with a 1 : 1 allocation using random block sizes of six. The two study days were separated by 2–14 days and each subject was investigated after an 8-h fasting period. Either pentagastrin (pgs; 0.6 μg kg−1 h−1; Cambridge Laboratories, Wallsend, UK) or placebo (plc; sodium chloride 0.9% Baxter®; Baxter, Volketswil, Switzerland) was infused intravenously (IV) at time point t = 0 min for 60 min. Volunteers ingested the test meal within 3–5 min after IV infusion was started. Breath samples were taken in fasting state and then after meal intake at regular time intervals until 180 min. Concurrent MRI measurements were performed until 90 min after meal intake. Heart rate and blood pressure were monitored over the whole study period.

13C-acetate breath test and MRI data

To focus on the effect of stimulated gastric secretion on 13CO2 excretion, this work analyzed breath test and MRI data during the pentagastrin study arm only for the IV infusion period, i.e., until 60 min. After the end of pentagastrin infusion, MRI volume data showed a change in the dynamics of secretion and also for meal emptying and, therefore, standard analytical methods could no longer be applied for the entire dataset.

13C-acetate breath test data  13CO2/12CO2 ratios were determined by molecular correlation spectroscopy (BreathID, Exalenz, Jerusalem, Israel) approximately once every 3 min with an automatic nasal breath sampling device under continuous capnographic control. The results were expressed as delta (δ) value per mil and delta over baseline (DOBt = δSample0).13 DOBt was used to determine the percentage dose of 13C recovered [PDR (% h−1)] and the cumulative PDR [cPDR (%)] at each measurement time point until the end of analysis period and if indicated as cPDR60 until the end of IV infusion period.13–15 According to the standard breath test analysis methods described by Ghoos et al.,13 meal emptying was quantified from the cPDR data by computing the parameter t50 (i.e., half emptying time) with t50_Ghoos  = (−k−1) × ln(1–2−1/β) using the model cPDR(t) = m × [1–e(−k·t)]β. Using non-linear mixed-effect modeling, k and β were determined per subject and per visit whereas m was estimated for the population.

Magnetic resonance imaging (MRI) data  Magnetic resonance imaging measurements were performed in supine body position approximately once every 10 min on a whole-body MRI system (1.5 T Achieva; Philips Medical Systems, Best, The Netherlands). Imaging techniques were described previously.6 The secretion volume (SV) and the contrast agent (CA) labeled meal volume (MV) were extracted from the previously determined gastric content volume (GCV) according to an established and validated procedure.9,10 In brief, based on MRI T1 mapping data of the GCV and a ‘calibration curve’ that describes the interrelation between the measured intragastric T1 values and the CA concentration in MV, the percentage MV (%meal) was determined for each MR image series recorded. Based on the derived %meal, secretion and MV were then calculated as SV = GCV − MV and MV = GCV × %meal, respectively. As MV did not show a lag effect, MV data were fitted using a power-exponential mixed-effect model, inline image. To allow the comparison with the breath data that reflects the emptying of the 13C labeled meal, the half emptying time of MV, t50_MV = tempt × ln(2) was calculated. In addition, the area under the curve of SV until end of IV infusion period (AUC_SV60) was calculated by the trapezoid method.


Data processing, statistical analyses, and plots were done by R 2.13.1 (R Foundation for Statistical Computing, Vienna, Austria) and GraphPad Prism® 5 (La Jolla, CA, USA). Non-linear mixed-effect modeling was performed using the nlme function of R.16 A linear regression model was used to test for correlation between t50 values derived from breath test and MRI data, and agreement between t50 values was visualized by the Bland-Altman plot. All nlme and linear regression parameter estimates are given as value ± standard error. Effects of pentagastrin infusion on half emptying times were compared using paired Wilcoxon signed-rank test and differences in t50 between the study arms are given as median (95% CI). Correlation between AUC_SV60 and cPDR60 was determined by linear regression analysis. The significance level was set to P = 0.05.


Breath test and MRI data

Visual inspection of the data revealed that 13CO2 exhalation measurements, expressed as cPDR and PDR, were lower during pentagastrin infusion compared to placebo in the majority of the volunteers (Fig. 1A). In addition, concurrent meal volume decrease, as assessed by MRI, appeared to be slower and secretion volume larger during pentagastrin infusion compared to placebo (Fig. 1B). Both, breath test and MRI data, showed interindividual variations during both placebo and pentagastrin infusion.

Figure 1.

 Individual raw data. (A) Individual PDR (% h−1) and cPDR (%) data during infusion with pentagastrin (black) and placebo (gray). PDR was lower after 60 min of pentagastrin infusion in 8 of 11 volunteers; cPDR was lower in 7 of 11 volunteers. (B) Individual MV (mL) and SV (mL) data during infusion with pentagastrin (black) and placebo (gray). Meal volume decrease during pentagastrin infusion was lower in 7 of 12 volunteers. Secretion volume during pentagastrin infusion reached higher levels in 8 of 12 volunteers.

Fits of cPDR and MV overlaid on the original data are presented in Fig. 2A and B, respectively.

Figure 2.

 Individual fits. (A) Individual results for fitted cPDR (%, line) during infusion with pentagastrin (upper row) and placebo (lower row), overlaid on raw data (gray dots). Estimated population parameters with standard errors were as follows: m = 71.34 ± 1.04, log κ = −5.2 ± 0.06, log β = 0.56 ± 0.04. (B) Individual results for fitted MV (mL, line) during infusion with pentagastrin (upper row) and placebo (lower row), overlaid on raw data (gray dots). Estimated population parameters with standard errors were as follows: V0 = 495.9 ± 16.4, tempt = 99.6 ± 8.4.

Correlation and agreement of breath test and MRI data

Due to technical difficulties during pentagastrin treatment, breath test data of one volunteer was missing and excluded from the analyses.

All (i.e. pgs and plc) half emptying times derived from breath test (t50_Ghoos) and all meal half emptying times derived from MRI (t50_MV) showed a linear correlation with a slope of 1.1 ± 0.3 and r2 = 0.67 (Fig. 3A). The Bland-Altman plot revealed a positive offset of 136 min for t50_Ghoos compared with t50_MV (Fig. 3B). No correlation (r2 = 0.11) was detected between all data encompassing AUC_SV60 and cPDR60 (Fig. 3C), suggesting a negligible influence of gastric secretion on 13CO2 exhalation.

Figure 3.

 Comparison of breath test- and MRI-derived data. (A) Correlation of t50_Ghoos (min) and t50_MV (min). The linear relationship was t50_Ghoos = (126 ± 19) min + (1.1 ± 0.3) × t50_MV, with r2 = 0.67. (B) The Bland-Altman plot of t50_Ghoos (min) and t50_MV (min). The plot shows an offset of 136 min for t50_Ghoos with limits of agreement of ±67 min. (C) Correlation of AUC_SV60 (mL*min) and cPDR60 (%). The linear relationship was cPDR60 = (3566 ± 691) − (131 ± 80) × AUC_SV60, with r2 = 0.11.

Effect of pentagastrin stimulation on gastric meal emptying half time

Breath test and MRI data both revealed a longer t50 for the pentagastrin study arm (Fig. 4A). Table 1 shows the individual differences in t50 between pentagastrin and placebo. Median (95% CI) differences in t50_Ghoos and t50_MV were 45 min (28–84), = 0.002, and 39 min (28–52), P = 0.002, respectively (Fig. 4B).

Figure 4.

 Differences in half emptying times (t50) between pentagastrin and placebo. (A) Individual differences in t50 and (B) median and 95% CI of the differences in t50 between pentagastrin (pgs) and placebo (plc), as assessed by breath test (t50_Ghoos) and MRI (t50_MV). Both figures highlight the wider range found for t50_Ghoos.

Table 1.   Differences between pentagastrin- and placebo-derived half emptying times in minutes for breath test and MRI data
InitialsΔt50_Ghoos (min)Δt50_MV (min)


To investigate the effect of enhanced gastric secretion on the reliability of meal emptying measurements by 13C-acetate breath test, this work analyzed simultaneously acquired 13C-acetate breath test data and meal emptying data recorded by MRI. Independent of the amount of gastric secretion, a linear correlation for the derived half emptying times (t50) was found between both methods and, compared with MRI, a large positive offset in t50 was determined for the 13C breath test data analyzed by the method proposed by Ghoos et al.13 As expected, pentagastrin caused a prolongation in meal half emptying times11,17 that was detected with both methods. The prolongation in 13C-acetate breath test half emptying time observed during pentagastrin infusion was comparable with the prolongation detected by MRI, further suggesting no confounding effect of gastric secretion on 13CO2 exhalation data.

Currently, MRI is the only method to non-invasively assess gastric SV and related distribution and mixing of gastric secretion.6,8–10 The applied quantitative MRI technique6,12 enables the extraction of meal and secretion volume from imaged GCV, thus allowing a separate quantification of meal emptying from and secretion production into the stomach. Magnetic resonance imaging detected an onset of different meal emptying and gastric secretion dynamics after the end of the pentagastrin infusion period (data not presented here). This characteristic of the meal emptying prevented incorporation of the entire dataset of the pentagastrin study arm into a standard non-linear mixed-effect model analysis. Therefore, as already pointed out, analysis of data from the pentagastrin study arm was limited to the pentagastrin infusion period. Data from MRI revealed that gastric SV was increased by pentagastrin infusion with interindividual differences. From visual inspection of MRI data, the intended secretion stimulating effect was detectable within the first 20–30 min after start of infusion (Fig. 1B), which is in accordance with past findings, where an increase in gastric acid output after stimulation with gastrin derivates was seen after a comparable period.18,19

In all subjects, the detected cumulative 13CO2 exhalation dynamic was more related to the meal emptying dynamics than to the integrated volume of gastric secretion (Fig. 3). Breath test-derived half emptying times (t50_Ghoos) were well correlated with the MRI-derived meal half emptying times (t50_MV) (Fig. 3A); however, exhibited a considerable positive offset of 136 min in the Bland–Altman plot (Fig. 3B).

Positive offsets in t50 derived by the 13C-acetate breath test have previously been described in other comparative imaging studies using ultrasonography and gamma-scintigraphy.20,21 Using the method proposed by Ghoos et al.,13 these studies found mean offsets of 41 min and 49 min, respectively, in comparable test meals. The authors related their observed offset to all required postgastric processing steps of the 13C-acetate marker preceding 13CO2 exhalation. In studies comparing 13C-octanoate breath test with gamma-scintigraphy, Lee et al.22 and Choi et al.23 reported mean positive offsets in emptying half times derived from 4-h 13C breath test data of 144 and 69 min, respectively. These studies concluded that the offset can partly be assigned to an overestimation of the parameter m which is used for calculation of the t50_Ghoos. This parameter m is equivalent to the maximal value of cPDR, i.e., cPDR(∞), and is dependent on the actual shape of the cPDR curve, particularly if a plateau in 13CO2 exhalation data has not been reached during the acquisition period. The parameter m usually approximates 40–80% of the ingested 13C amount, depending on the irreversible non-pulmonary loss of the 13C-atom.22–24 The limited time window available for the analysis during the pentagastrin study arm did not allow for individual estimation of the parameter m in the majority of subjects. Therefore, m was determined for the population, whereas the parameters k and β were simultaneously fitted for each individual. This represents a valid and common approach in pharmacometrics25,26 and was here performed using the nlme function of the program R. k (the elimination constant of the 13C atom in breath) and β (the steepness of the rise of 13CO2 exhalation) depend on the type of gastric emptying (e.g., if it is fast, slow). The parameter m depends directly on fixation and non-pulmonary loss of the 13C-atom. By determining m only for the population, this parameter was considered a common feature of the study cohort. This simplification was required, and may be considered a conservative approach, to allow for robust individual detection of changes in β and k and thus reliable statistical comparison of t50_Ghoos between the two study arms. This approach was indirectly validated by the fact that it allowed the MRI confirmed detection of the pentagastrin-induced delay in gastric emptying. The estimation of m to 71% lies within the upper range of previously reported values. The shorter cPDR curves of the pentagastrin arm caused some overestimation in the population value m, resulting in a larger offset for the calculated t50_Ghoos. However, the extent by which the offset in t50_Ghoos compared with t50_MV can be explained by the postgastric metabolic processes or may be due to the applied standard analytical ‘Ghoos’ method cannot be clearly deduced from this work. It may be suspected that it is caused to a major part by the ‘Ghoos’ method itself. Derivation of t50 by analyzing breath test data with other published methods, such as the method published by Bluck,27 did also lead to t50 values that correlated with t50_MV, but showed different systematic offsets in half emptying times. Due to these inconsistencies between analysis techniques and the lack of a gold standard, the collected 13C-acetate breath test data were analyzed using the widely accepted ‘Ghoos’ method that has been previously validated for liquid emptying.20

Pentagastrin infusion caused an increase in gastric SV and also in meal emptying times (Fig. 4A). Various mechanisms that explain the observed decrease in meal emptying rate can be pursued: (i) Acidification of the duodenum with pH 1.0–2.7 induced a rapid increase in CCK plasma concentration which in turn is known to inhibit meal emptying.28–33 (ii) Neural reflexes involving acid-sensitive neurons that adjust delivery of acidic gastric content into the duodenum to balance the level of acid and hence to avoid mucosal lesions were triggered.34 (iii) Pentagastrin, irrespective of its secretion stimulating properties, may have had an effect on meal emptying by altering proximal gastric function, i.e., inducing relaxation of the fundus and increasing gastric wall compliance.17 However, the given data did not allow for robust separation of gastric secretion and/or pentagastrin-induced effects on meal emptying. To exclude the potentially overwhelming effect of pentagastrin itself on motor inhibition, future studies on this topic should be designed using physiologic, non-pharmacologic secretion stimulation.

Considering that the range of changes in t50_Ghoos is comparable with the range of changes in t50_MV between pentagastrin and placebo infusion, it can be assumed that the endocrine pentagastrin effect is dominating over the intragastric secretion effect. Both methods detected a comparable prolonging effect of pentagastrin infusion on t50 with median (95% CI) differences in t50_Ghoos of 45 (28–84) min and in t50_MV of 37 (28–52) min. Therefore, it may be speculated that pentagastrin affects the emptying process as a whole (and thereby in a more constant manner), whereas secretion might possibly have led to a partial or local effect in the stomach (and subsequent on the 13C-marker). This would have resulted in a wider range of changes in t50_Ghoos between plc and pentagastrin infusion and could also be another explanation for the missing correlation of AUC_SV60 and cPDR60 (Fig. 3C). In addition, the comparable prolonging effect of pentagastrin on t50_Ghoos clearly demonstrated that a pentagastrin infusion-specific effect on postgastric 13C metabolism can be excluded.

The above findings suggest that 13CO2 exhalation after ingestion of a 13C-acetate labeled liquid test meal is not affected by the stimulated gastric secretion volume, but is rather reflecting the dynamics of meal or caloric emptying from the stomach.

In conclusion, this study represents another step toward the validation of the 13C-acetate breath test for gastric emptying measurements. Even under potent pharmacological modulation of gastric secretion, the 13C breath test can reliably detect intraindividual differences in meal half emptying times.


The authors thank Philips Medical Systems, Best, The Netherlands, for technical and financial support.


This study was supported by Swiss National Science Foundation (SNF) grant number 320000–112006–1, and the Zurich Center for Integrative Human Physiology (ZIHP).


The authors have no competing interests.

Author contributions

SK and AS drafted the manuscript and were involved in data analysis/interpretation; SK, AS, DM, WS, and MFx were involved in critical revision of the manuscript; SK, AS, and DM analyzed the statistics; OG, MF, WS, and MFx contributed to study concept/design; RT, OG, and MFx did the data acquisition; PB, MF, and WS provided technical support and supervised the study.