Clinical assessment of gastric emptying and sensory function utilizing gamma scintigraphy: Establishment of reference intervals for the liquid and solid components of the Nottingham test meal in healthy subjects

Current investigations of stomach function are based on small test meals that do not reliably induce symptoms and analysis techniques that rarely detect clinically relevant dysfunction. This study presents the reference intervals of the modular “Nottingham test meal” (NTM) for assessment of gastric function by gamma scintigraphy (GSc) in a representative population of healthy volunteers (HVs) stratified for age and sex.


| INTRODUCTION
Postprandial dyspeptic symptoms including uncomfortable fullness, early satiety, bloating, heartburn, nausea and abdominal pain affect up to one in five of the general community and are responsible for up to 25% of referrals to out-patient gastroenterology. [1][2][3] Assessment of gastric function is indicated in patients with normal endoscopy that fail to respond to empirical therapy with acid suppression and antiemetics. Gamma scintigraphy (GSc) provides direct measurements of gastric emptying (GE) and is the reference standard.
Currently, the low-fat, "eggbeater" meal is the best-established test meal used with GSc. 4,5 Using this method, delayed GE can be documented in approximately 40% of patients with functional dyspepsia and up to 75% of patients with chronic unexplained nausea and vomiting. [6][7][8] Severely delayed GE ("gastroparesis" or "gastric failure") is associated with postprandial vomiting, weight loss, poor health status and poor outcome of therapy; [8][9][10] however, such cases are rare and studies have not demonstrated an association of abnormal GE with dyspeptic symptoms. 8,11 Moreover, delayed GE does not consistently predict clinical response to metoclopramide or other prokinetic and antiemetic medications. 12,13 Thus, there is a clear need for clinical investigations suitable for routine practice that can identify the causes of dyspeptic symptoms and direct effective therapy. 14 We hypothesized that the relatively poor association between dyspeptic symptoms and GE measured by GSc may be due to limitations related to (i) the test meal, (ii) analysis of imaging data and (iii) failure to induce and/or record gastric sensations.
To address the first issue, we developed the modular liquid/solid Nottingham test meal (NTM) which has a higher volume and higher fat content than other, established test meals. 5,[15][16][17] In validation studies the NTM triggered normal postprandial sensations in healthy subjects and provided reproducible measurements of GE with non-invasive imaging. 18 The second issue with current GSc studies is that analysis is generally limited to a single outcome measurement (e.g., GE half-time [T 50 ], residual volume at 4 h). 4,5,19 However, most patients with functional dyspepsia report symptoms during gastric filling or immediately after ingestion of the meal and not hours later. [20][21][22][23] These postprandial symptoms are often related to impaired gastric relaxation ("accommodation") and hypersensitivity; 9 however, such elements of gastric function are not assessed by current GSc protocols. The distribution of a liquid test meal in the stomach provides one non-invasive measure of accommodation. 24,25 The effect of accommodation on early-and latephase GE provides another method to assess this process. 23,[26][27][28][29][30] Both are documented in by the GSc analysis presented here. Additionally, the solid component of the NTM is comprised of agar beads with a known breaking strength. Emptying of this solid component provides an assessment of the mechanical work done by the antral contraction waves (trituration). 19 Finally, gastric sensitivity can be assessed by concurrent reporting of symptoms during gastric filling and emptying using validated scores. 31 This study presents reference intervals for the assessment of gastric motor and sensory function by GSc

| Participants
Adult healthy volunteers (HVs) aged 18-80 were recruited by advertisement. Subjects were stratified by age and sex so that a minimum of 10 men and women in three age groups (18-40, 41-60, 61-80) completed the Liquid-NTM. A subset of participants also completed the Mixed-NTM (Liquid and Solid components). All subjects underwent the same screening procedures.
At the initial screening visit, participants completed validated questionnaires regarding their health and wellbeing. These included patient health questionnaire (PHQ 12), hospital anxiety and depression questionnaire (HADS) and the EuroQol 5D ™ (EQ-5D) quality of life questionnaire. 32

| General procedures
Subjects fasted from midnight and abstained from alcohol and strenuous exercise for 24 hours prior to each study day. Smoking was not permitted during the study. Subjects underwent gastric scintigraphy (GSc) and magnetic resonance imaging (MRI) on separate study days, a minimum of 48 hours apart. Each study including the screening visit was completed within a 4-month period. Magnetic resonance imaging data will be reported in separate publications.

| Blood glucose measurement
Subjects participating in the Mixed-NTM study had capillary blood glucose measurements taken from the ear lobe at baseline (i.e., before ingestion of MNTM), 30, 60 and 120 minutes postingestion of the Mixed-NTM (Optium Xceed ™ system, Medisense, Abbott Laboratories Ltd, Berkshire, UK). Before each sample was taken a small amount of rubifactant (Deep heat rub, Mentholatum, UK) was applied to the ear lobe approximately 1-3 minutes prior to the test. The rubifactant was used to improve circulation ("arterialization") of the capillary bed to ensure more reliable measurements of blood glucose. The rubifactant was removed from the area prior with moist cotton wool or gauze prior to the test. Further, in a subset of 8 HVs participating in the Liquid-NTM a 5 mL venous blood sample and blood glucose measurements were taken at baseline, 30 minutes, 60 minutes and 120 minutes post Liquid-NTM ingestion. Venous cannulas were placed on study mornings prior to ingestion of the test meal. A small amount of blood from each sample was applied to the Optium Xceed ™ blood glucose test strips. The average of three measurements from each sample was then recorded.

| Study protocol
Radioactive markers were affixed to the subject at the right costal

| Liquid Nottingham test meal (Liquid-NTM)
After baseline imaging, 200 mL of the liquid test meal was ingested from a series of beakers containing 50 mL liquid nutrient over 5 minutes. The subject was then imaged (−5 minutes scan). The remaining 200 mL of the test meal was then given in the same manner so that the entire test meal was consumed over 10 minutes and the subject imaged again (0 minutes scan). Gastric scintigraphy images were then acquired at 5, 10, 15, 30, 45, 60, 75, 90, 120 minutes. At baseline and after each scan the subjects were asked to score satiety, fullness, bloating, heartburn, nausea and epigastric pain using a previously validated visual analogue scale (VAS 0-100 mm). 31

| Mixed Nottingham test meal (Mixed-NTM)
The first 200 mL of the liquid test meal was ingested as described above and the subject imaged (−5 minutes scan). The remaining nutrient drink was then given with 12 agar beads swallowed whole (3 beads with every 50 mL beaker). 18 This two-stage methodology allowed the In-111 Cl 3 gamma photon scatter within the Tc-99m channel on the GSc images to be calculated. Imaging continued for 120 minutes as for the liquid meal but with an additional 75-and 115-minute time point. After the 115-minute scan 200 mL of water was given ahead of the final scan at 120 minutes (procedure required for MRI study).
Additional scans were performed in a subset of patients at 30-minute intervals until a maximum of 4 hour. Gastric sensation was assessed as for the Liquid-NTM. Note: the data from some of the subjects included in this study has been previously published in pilot studies. 18

| Liquid-NTM study
Liquid GE begins almost immediately during ingestion and, in HVs, is almost always seen at completion of the test meal. 30  Similarly, to the whole gastric content the absolute volume (mL) of the test meal in the proximal and distal section was calculated. All counts were corrected for background radiation and radioactive decay. The early phase GE is expressed as the postprandial volume immediately after completion of the meal from the V 0 parameter (GCV0) and late phase GE is expressed as GE rate (mL/min) at T 50 (GErateT 50 ). The early phase GE rate (mL/min) was determined by calculating the remaining volume of the test meal present in the stomach at GCV0 and dividing by the 10 minutes required to dose the meal (Early phase GE rate). The fraction of the NTM in the proximal stomach (proximal/ whole) was used to present meal distribution within the stomach.

| Statistical analysis
Demographic results are reported as median with [interquartile range] and Wilcoxon tests were used for between group comparisons.
Data analysis of blood glucose measurements was performed using GraphPad Prism version 6.0 (GraphPad Software Inc., La Jolla, CA, USA). Blood glucose measurements are reported as mean with 95% upper and lower (confidence intervals).
Gastric volume data obtained over 120 minutes by GSc were fitted to linear exponential curves with a population-based approach using Stan source code (Data S1). 36 Solid (agar bead) emptying data was fitted by a power exponential curve. 37 This approach ensures that all curves can be fitted and give regularized coefficient estimates.
The reference intervals of the GE parameters were determined by the robust method as given in the Clinical and Laboratory Standards Institute's guideline with Box-Cox transformation and reported as the upper and lower 95% reference intervals (RI) of the population. 38 The median was also provided alongside the upper and lower 95% confidence intervals (CI) of the mean. To compute the reference interval for percentage retention of agar bead a logit/antilogit transformation was used to constrain fractions to the range of 0-100%. The initial emptying phase of agar bead emptying was determined by the parameter beta from the power exponential function. 37 The absolute percentage of liquid meal emptying calculated from unfitted data is reported as mean with 95% confidence intervals.

| Participants
In total 91 subjects consented to the studies. Seventeen subjects were excluded during the screening process due to previous history of GI Demographic, anthropometric and health questionnaire data for all subjects stratified by age and sex are provided in Table 1. A small number had evidence of a psychological disorder (n=4, HADS>11). Self-rated health status was very good-excellent (>75 VAS in EQ-5D) in all subjects.
There were no significant differences between the sub groups for either the HADS, PHQ or EQ-5D self-rated questionnaires Table 1. All subjects tolerated the complete 400 mL Liquid-NTM and Mixed-NTM.

| Measurement of gastric sensation
There was no difference in the sensation of fullness or satiety reported between the two NTM meals (P=1.0 and P=.46, respectively), as shown in Figure 1. The mean VAS scores and 95% confidence intervals of fullness and satiety are presented in Table S1. At baseline, most subjects reported less than mild fullness (0-30 mm VAS). After completing the 400 mL test meal most subjects reported more than mild but less than moderate fullness (i.e., between >30 but <60 mm VAS). More than mild bloating (>30 mm VAS) was reported by only two subjects. No other dyspeptic symptoms (i.e., nausea, heartburn, pain) were reported as more than mild throughout the study.

| Blood glucose measurement
Blood glucose measurements were obtained from 32 subjects that

| Liquid gastric emptying
Representative images and data from the Liquid-NTM from a HV participant are shown in Figure 2. One subject from the mixed-NTM T A B L E 1 Demographic, anthropometric and health questionnaire data for healthy volunteers by age and sex reported as the median and [interquartile range]. Wilcoxon tests were used for between group comparisons of sex stratified groups Mixed-NTM study due to poor fit). Gastric emptying measured by GSc followed the same pattern of liquid emptying previously reported. 18 The initial meal volume measured immediately after NTM ingestion was generally less than 400 mL indicating "early phase" GE that occurs during meal ingestion. Subsequently there is a linearexponential decrease in gastric meal volume over time (Figure 3).
Liquid GE reference intervals are presented in Table 2 for both the Liquid-NTM and the Mixed-NTM. The absolute percentage of liquid gastric retention from the unfitted data is presented in Table S2. As previously reported the agar beads had a small but significant effect on the GE parameters. Bayesian Modeling determined that the T 50

| Solid meal gastric emptying
The GE of the solid component of the Mixed-NTM is demonstrated in Figure 4. The reference interval ranges for solid agar bead empting are presented in Table 3. Solid GE was characterized by an initial, slow emptying phase followed by an essentially linear decrease in volume (Table 3). There was a weak correlation between emptying of solid beads and of the liquid emptying T 50 meal (r=.18).

| Interaction of early phase gastric empting on late phase gastric emptying
There was a moderate negative correlation between early phase GE and late phase GE with the Mixed-NTM (−0.43, P=.004). Thus, the Mixed-NTM demonstrated that a faster early phase GE rate was associated with a slower late phase GE rate Figure S1. This interaction was not significant for the Liquid-NTM (P=.47).

| Meal distribution within the stomach
Image data for GCV0, proximal and distal gastric volumes derived from 10 subjects who ingested the Liquid-NTM were analyzed by three independent observers. The inter-observer agreement for the ROIs for the whole stomach, antrum and fundus were calculated (Table S4)

| Relationship of gastric filling and distribution with gastric sensation
The presence of the liquid meal in the proximal stomach of both the Liquid-NTM and Mixed-NTM had no correlation with sensation (fullness, or satiety) at T 0 minute in health Table S5. However, in Figure 5 the VAS scores of fullness and satiety decreases in relation to volume within the proximal stomach.

| The effect of patient factors on gastric emptying
Bayesian model averaging was used to determine the effect of demographic and anthropometric factors on the GE parameters GCV0, T 50 and GErateT 50 . A total of 24 combinations of parameters were tested and the best three alternate models are provided in Table S2. The results indicate that there was no single predictor for both GCV0 and GErateT 50. There was a minor effect of age on T 50 in that for a 10-year increase of age the T 50 increased by 2 minutes. Similarly, an increase of 1-cm waist circumference was associated with T 50 increase by 30s.
In addition, male sex decreased T 50 by 7-minute although this predictor had a low probability in the model.

| DISCUSSION
This study provides reference intervals for the clinical assessment of gastric motor and sensory function by gastric scintigraphy (GSc) using the large, modular "Nottingham test meal" (NTM). Values are provided for liquid (Table 2) and solid (Table 3) components of the NTM from a large cohort of healthy participants.

| Gastric emptying
Typical patterns of liquid and solid GE in health were observed. 42 For the liquid NTM "early phase" emptying commenced during ingestion of the meal and was followed by a "late-phase" linear-exponential reduction in meal volume (Figure 3). 18 Detailed measurements by Magnetic Resonance Imaging (MRI) following ingestion of a liquid test meal have shown that "early phase" GE is driven by volume load alone, whereas "late phase" GE is modulated by volume and calorie load (i.e., neurohormonal feedback). 23,[26][27][28][29][30]43 Further studies have shown increased early phase GE in functional dyspepsia patients with impaired accommodation detected by gastric barostat. 26 The same effect was observed also with GSc after ingestion of the 400 mL NTM. 23 In pilot clinical studies rapid early GE after the meal was observed in a proportion of patients with functional dyspepsia and this may indicate the presence of impaired gastric accommodation in this group. 23 Conversely, if symptoms occur in the absence of motor disorders (i.e., abnormal GE) it may be inferred that gastric hypersensitivity is the likely cause of dyspeptic symptoms.
For the solid NTM, GE showed a characteristic initial, slow emptying phase relating to the trituration (breakdown) of the agar beads into smaller particles, 19 followed by an essentially linear pattern of emptying ( Figure 4). This provides an objective assessment of the "mechanical work done" by the antral contraction waves in the postprandial period. 19 The retention of solids in the stomach may be more sensitive than abnormal liquid GE for detection of abnormal gastric function in patients with certain conditions (e.g., diabetic gastroparesis). 44 The NTM is designed to be modular in the sense that the liquid component can be used with or without the non-nutrient solid component. This would be practical because dual-radionuclide studies are relatively expensive and the analysis is complicated due to scatter F I G U R E 4 Spaghetti plot of percentage agar bead retention per subject over time. The outer ribbon is the approximate 95% reference interval of the population. The inner ribbon demonstrates the bootstrapped 95% confidence intervals of the mean.

| Gastric accommodation
The assessment of gastric accommodation is challenging. Barostat measurements provide a direct measurement of gastric volume change at a given pressure; however, this technique is highly invasive and the presence of a large bag in the stomach has effects on GE. 26 Less invasive measurements of intra-gastric pressure using highresolution manometry have been proposed as a surrogate of accommodation; however, pressure change after a 400 mL meal is small (typically <4 mm Hg, at the limit of resolution by current equipment), varies with position relative to the meal and is difficult to interpret without some assessment of volume change. 30  Mixed-NTM such that fast early phase GE tended to be followed by slow late phase GE. This is likely due to relatively rapid initial delivery of nutrients to the small bowel leading to the rapid release of neurohormonal factors that slow GE to ensure efficient absorption and assimilation of the meal (the so-called small bowel or "ileal brake." 46 )

| Gastric sensation
Many studies have shown a linear relation between meal volume filling and gastric filling sensation. 47,48 The sense of fullness is modulated by various factors including the composition of the meal, with dietary fat known to increase visceral sensitivity more than other macronutrients. [47][48][49] The liquid NTM has a larger volume and higher fat content than most other test meals. 5,[15][16][17] Consistent with published pilot data, 18   There was also a minor interaction between T 50 and waist circumference within the healthy range tested. Recently our group reported that GE was slower in patients with clinically stable anorexia nervosa than in healthy subjects or obese patients. 58 Other studies showed no such findings; however, in all cases, the effects of body weight on GE were small. In summary, we consider that the reference values for the NTM can be applied without adaption for demographic or anthropometric factors.

| Limitations
This study provides reference intervals for gastric function from a large, representative population of healthy individuals (n=74); however, the Clinical Laboratory Standard Institute has recommended a minimum of 120 patients to establish normal values in a system with large inter-individual variation and a large degree of physiological redundancy. 38 An alternative approach applied by other, well-established measurements of gastrointestinal physiology such as the Chicago Classification of esophageal motility disorders, is to apply this data alongside patient data to determine thresholds that define not the "normal range," but definitively pathological function. 59 These studies are in progress and will also establish the

| Potential application in clinical practice
Gastric function was measured using standard GSc technology without the need for specialized equipment. The time required to complete GSc with the liquid NTM is 2 hours which is less than that employing other radiolabeled test meals. 4,16,53 Gastric volume data was fitted and analyzed using open-access software (Menne Biomed, Tübingen, Germany). This is optimal with regards to data fitting; however, standard imaging analysis can be applied. 66 Finally, the NTM is inexpensive, simple to prepare and easy to administer. It is also suitable for use with those special dietary requirements (e.g., vegetarians) and does not contain lactose, gluten, eggs or other food substances linked with food intolerance or allergy. These attributes of the NTM ensure that there should be few barriers to implementation.
In upcoming publications based on data obtained by the NTM in routine clinical practice, we will assess whether non-invasive imaging of gastric function can identify the causes of symptoms in patients with functional dyspepsia, gastroparesis and related conditions. As in other areas of medicine, it is likely that definition of clinical phenotypes based on objective measurement is a key step to effective and specific treatment of these challenging conditions.

ACKNOWLEDGMENTS
We thank David Kirby for assisting with scintigraphic imaging.

AUTHOR CONTRIBUTION
The authors' responsibilities were as follows: HLP, ET, and EB were involved in planning and performance of study, data collection, interpretation and analysis. CLH and DM data interpretation and analysis; LM and AP were involved in planning of study and data interpretation;