Assessment of Tandem Measurements of pH and Total Gut Transit Time in Healthy Volunteers

In 1972, Brown and co-workers utilized radiotelemetric technology with a glass electrode for in vivo measurement of intraluminal gut pH.1 Since then at least eight additional studies have employed similar technology to characterize the pH profile of the gut in healthy individuals. Each of these studies has shown that luminal pH in the proximal small bowel ranges from 5.9 to 6.8, as the alkaline pancreatic secretions buffer the acidic gastric contents. The pH then gradually rises to 7.3–7.7 in the distal ileum, as the small bowel mucosa secretes bicarbonate. At the ileocecal junction, there is an abrupt decrease in the pH of the cecum (5.8–6.7), thought to be reflective of the fermentation of carbohydrates to short chain fatty acids (SCFAs) by colonic bacteria. Finally, there is a described increase in the pH of the left colon and rectum to 6.1–7.2 as the colon absorbs SCFAs and secretes bicarbonate.2 The SmartPill (Covidien, Minneapolis, MN), was introduced in 2006 as a novel radiotelemetric device to record pH, transit time, and pressure within an individual’s gastrointestinal (GI) tract and was FDA (Food and Drug Administration) approved for assessment of gastroparesis. The SmartPill is a wireless motility capsule that transmits data captured from a subject’s GI tract to a data receiver worn by the subject. This device has been used to determine gut pH profiles in healthy volunteers. In such assessments, the mean pH in different locations of the intestine was as follows: proximal small bowel 6.4 (95% confidence interval 6.1–6.8); ileum 7.6 (7.4–7.8); proximal colon 6.2 (5.8–6.5); and distal colon 7.3 (7.0–7.6).3 All the previously mentioned studies were based upon one-time measurements of gut pH; the generalizability of these data is unknown because there is no current evidence to suggest that it is a static measurement. In fact, no study to date has been performed to evaluate the possibility of intrasubject variations in luminal pH or transit time, whether due to physiologic processes or limitations of the radiotelemetric technology. This study assessed intrasubject variability in the pH and transit time of the small and large bowel of healthy volunteers. The pH sensor of the device measures pH in the range of 0.5–9.0 with an accuracy of ±0.5 pH units, meaning that the recorded pH will never differ more/less than 0.5 pH units from the actual value. In the first 24 h, this is actually no more/less than 0.25 pH units. The original Smartpill Corporation determined this accuracy by assessing (in a 5-day period) the measurements reported by capsules submerged in solutions at body temperature with a known pH value that ranged from 0.5 to 9.0. The sensor is an ISFET (ion-sensitive field-effect transistor-based sensor), which utilizes a single-point calibration at room temperature immediately preceding ingestion with a citrate buffer with pH 6.0. The device also measures pressure in the range of 0–350 mm Hg with an accuracy of ±5 mm Hg below and ±10 mm Hg above 100 mm Hg and temperature in the range of 20–40 °C with an accuracy of ±1 °C. The pH is measured every 5 s for the first 24 h, every 10 s for 24–48 h, and every 5 min after 48 h. Pressure is measured every 0.5 s for the first 24 h and every 1 s after 24 h. Temperature is measured every 20 s for the first 24 h and every 40 s after 24 h. The pH, pressure, and temperature measurements, battery voltage, capsule serial number, and a data packet identification number are transmitted at 434 MHz to the subject-worn data receiver every 20 s for the first 24 h of capsule operation, and subsequently, every 40 s for the duration of the capsule use.5 The receiver device has an “Event” button to indicate the timing of food consumption and bowel movements during capsule transmission. The Smartbar, a 260 kcal FDA approved nutrient bar, was used as a replacement for the standard eggbeater meal. It contains the following ingredients: whey crisp, rolled oats, rice syrup, corn syrup, invert sugar, concentrated whey protein, puffed wheat, apple juice concentrate, glycerine, molasses, and 2% or less of: honey, defatted wheat germ, dried apples, dried cranberries, partially hydrogenated vegetable shortening (soy and/or cottonseed oils), natural and artificial flavors, potassium sorbate (as preservative), salt, and vanilla. Twenty-four hours after ingestion of the initial SmartPill, each subject ingested a second capsule in a manner identical to that described above. This capsule was synced to a second, separate data receiver. Post capsule ingestion instructions were identical to the first capsule procedure (Figure 1). To ensure that the data receivers would capture a temperature decrease as it transitioned from the body to the toilet, subjects were instructed to count to 100 after completion of a bowel movement before flushing the toilet. Concurrently, subjects were instructed to monitor the light on the receiver that confirms successful data transmission. If the sensing light remained inactive (red) for a period of 30 min, especially following a bowel movement, then the subject was to assume that the capsule had exited the body. The subjects returned the data receivers by mail in a pre-paid, pre-addressed containers. Intrasubject variability was determined by computing the standard deviation (s.d.) between the two pH values of each segment of the GI tract for an individual subject. Agreement between the pH and transit times recorded at the two time points was assessed with Bland-Altman plots, 95% limits of agreement (LoA), and coefficients of repeatability (CRs=1.96 times the standard deviation of intrasubject differences). The LoA (mean intrasubject difference ±1.96 s.d. of intrasubject differences) is an interval within which most individual intrasubject differences are likely to fall. The interval characterized by the LoA is analogous to a 95% confidence interval for the mean intrasubject difference, computed using the standard deviation instead of the standard error. When the mean difference is equal to zero, the LoA is symmetric about zero, and the interval limits are equal in magnitude but have opposite signs. The upper limit of an interval characterized by LoA centered at zero is defined as the CR. For example, if CR=0.5, then most pairs of measurements from the same subject are expected to be no more than 0.5 units apart (i.e., their differences would mostly fall in the interval (−0.5, 0.5)). When the mean intrasubject difference is significantly different from zero (i.e., when computation of the CR is not possible), the asymmetric 95% LoA is provided. Intrasubject variability in transit times was summarized using coefficients of variation. Bland-Altman plots were also used to investigate whether differences between subsequent measurements varied according to the magnitude of the measurements.8 Paired t-tests were also used to compare the mean differences in pH values and transit times for all 10 subjects’ first and second ingestions. As a sensitivity analysis, non-parametric analogs of these tests (i.e., Friedman test and Wilcoxon signed rank test) were also performed. All statistical analyses were performed using StatPlus (AnalystSoft, Alexandria, VA) and R version 3.02 (R Foundation for Statistical Computing, Vienna, Austria). A total of 11 subjects were recruited, screened, and enrolled in this study: seven were male and four were female. Participants’ ages ranged from 18 to 26 years old (mean±s.d.: 21±2.68 years old). Tandem device ingestion and data transmission occurred without any adverse events experienced by the subjects. However, one subject’s data could not be included as data transmission for the second injection did not occur past the stomach. The subject was able to visibly confirm the presence of the second pill following a bowel movement. A similar, though less substantial event occurred in the second ingestion for subject #9 when data transmission was lost during colonic transit. The mean pH values for 30 min after each landmark for each individual ingestion are shown in Table 1. The average pH values for all 20 ingestions together were 5.71±0.44 (standard deviation) for 30 min AGE, 7.35±0.32 for the 30-min BIC valve, 6.07±0.38 for the 30-min AIC valve, and 6.84±0.51 for the 30-min BBE, with median values of 5.79, 7.37, 6.04, and 6.77, respectively. The medians and ranges of the intrasubject standard deviations between the two ingestions for all 10 patients are 0.02 (0.0002–0.2048) for 30 min AGE, 0.06 (0.0002–0.3445) for 30 min BIC, 0.14 (0.0018–0.3042) for 30 min AIC, 0.08 (0.0098–0.5202) for 30 min BBE, and 0.05 (0.0002–0.5202) for all four locations combined (Figure 3). There is a trend of stable standard deviations in the small bowel with an increase in the proximal colon that decreases in the distal colon, but this trend is not statistically significant. The mean differences in pH values between the two ingestions were −0.14±0.30 for 30 min AGE (P=0.18), −0.22±0.36 for 30 min BIC (P=0.09), −0.45±0.31 for 30 min AIC (P=0.0015), and −0.10±0.58 for 30 min BBE (P=0.62). In regards to AGE and BIC, the CRs are 0.60 and 0.70, respectively, meaning that 95% of the absolute differences in pH in these two locations are expected to be less than the values of these coefficients. It was not appropriate to calculate a repeatability coefficient for AIC because the mean difference is significantly different from zero (as demonstrated above). However, most intrasubject pH differences associated with the AIC landmark are expected to fall within the interval characterized by the 95% LoA (−1.06, 0.16). BBE pH has a slightly higher CR (1.14). There was no evidence of a linear relationship between magnitude of the pH difference and magnitude of the absolute measurements in the Bland-Altman plots (Figure 4). Regression analyses of the relationship between differences and means for the pairs of observed pH values (not shown) did not reveal any evidence of a systematic bias. The mean transit times for each individual ingestion are shown in Table 2. The mean transit time across all 20 ingestions for the small bowel was 288.6±52.7 min (4.81 h), 1,210.2±873.7 min (20.17 h) for the large bowel, and 1667.3±871.46 min (27.8 h) for the entire gut, with median values of 295.5, 1,004.0, and 1,428.0, respectively. The mean coefficients of variation between the two ingestions for all 10 subjects were 12.0±7.4% for the small bowel, 25.8±15.8% for the large bowel (P=0.01), and 19.4±10.2% for the entire gut. Transit time coefficients of reliability in the small bowel, large bowel, and overall (87.57, 936.05, and 1,044.68, respectively) are quite large, suggesting that transit times significantly vary from 1 day to the next. There was no evidence of a linear relationship between transit time difference and magnitude in the Bland-Altman plots (Figure 5). This study uniquely illustrates intrasubject variations in luminal pH and transit time in healthy volunteers. Our data show that the gut pH profile remains stable for an individual, with more variation seen in the colon compared with the small bowel. We also demonstrated that small bowel transit time is less variable than colonic transit time. Finally, we have revealed the safety and feasibility of the SmartPill device for tandem ingestion and investigative purposes in healthy volunteers. Understanding reliable pH and transit times of the bowel is important for many reasons. There is increased appreciation for the role of underlying motility disorders in functional bowel problems. Accurate measurements of these variations are required to better describe the abnormalities and define the parameters of "normalcy." In addition, previous studies have demonstrated alterations in intraluminal gut pH that is associated with disease. These observations contribute to an understanding of pathophysiologic processes. Finally, many medications incorporate an oral delivery system that depends on appropriate gut transit and changes in pH. Further knowledge of variations among healthy and diseased individuals will guide interpretation of therapy effectiveness and design for future delivery systems. Our subjects’ pH profiles are similar to previous studies of the physiology of normal GI tracts.2, 4 We measured an acidic pH in the stomach, which abruptly increased in the proximal small bowel before a gradual increase in pH with distal movement. Likewise we noted a sharp decrease in pH in the cecum, which increased gradually with progression though the colon. We did, however, observe lower pH values in the proximal small bowel and distal colon compared with other publications. We hypothesize that this is due to the standardized meal or differences in techniques using SmartPill as compared with earlier studies.2, 3 The median s.d. of intrapatient differences in pH is 0.05 (0.0002–0.5202). Furthermore, the LoA was small, with absolute differences less than 0.6, 0.7, and 1.14 units in the AGE, BIC, and BBE landmarks, respectively, and pH values in the AIC landmark are generally within 1.06 units of one another. These data demonstrate that the intraluminal pH does not markedly fluctuate in a given region of the gut over the course of 24 h and that repeat measurements of mean pH for a given region in the same subject are highly reproducible. However, although the LoA suggests that the individual intrasubject differences were relatively small, we did find that the mean intrasubject pH difference between the two ingestions was significantly different from zero. The largest standard deviation was observed in the proximal colon. These data in addition to the LoA for AIC (−1.06, 0.16) and BBE (−1.14, 1.14) being greater in absolute value than AGE (−0.60, 0.60) and BIC (−0.70, 0.70) suggest that the largest variations in pH differences between ingestions occur in the proximal and distal colon. We hypothesize that increased or decreased transit time caused transient changes in pH because of either accelerated or delayed clearance of alkaline bicarbonate secretions. An additional possibility is that these differences are secondary to measurement limitations of the device, which has an accuracy of ±0.5 pH units. We also considered a third possibility that variations in the amount of dietary insoluble fiber caused these differences. Brinkworth et al.9 has shown that a very low fiber diet is associated with a significant reduction in the fecal concentration of SCFA and a non-significant increase in stool pH as compared with a very high fiber diet. Furthermore, SCFA production seems to be maximal in the cecum and ascending colon, and then decreases progressively in the distal colon. These observations may also explain our findings of the increase in the coefficient of variability in the proximal colon that then slightly declines in the distal colon. Our mean small bowel and colonic transit time values are similar to those described by Maqbool et al.10 and Sarosiek et al.,6 two studies that also utilized the SmartPill device. Additionally, our intrasubject variability for the small bowel (12.0%) is less than the intrasubject variability for mouth to cecum transit time as determined by the lactulose H2 breath test (mean coefficients of variation of 18.5, 29.7, and 28.3% with 10, 15, and 20 g of lactulose, respectively) and the intrasubject variability for scintigraphically determined small bowel transit time (coefficient of variation of 19%). Our intrasubject variability for colonic transit time was similar to the intrasubject variability of scintigraphically determined large bowel transit time (coefficients of variation of 14–28%).11, 12 Interestingly, there is a significant difference in the intrasubject variability between the small bowel and large bowel transit time (12.0% vs. approximately 25.8%). This difference may be secondary to more tightly regulated physiologic mechanisms in the small bowel vs. the colon, or to the small bowel being less sensitive to the fiber content of a meal than the colon. Two technical malfunctions occurred during this study. The data from one subject could not be included due to an abrupt interruption of data transmission in the stomach during the second ingestion. The same receivers were used successfully by subsequent subjects in this study. This likely represented capsule failure, prolonged separation of the subject from the receiver, or loss of battery strength of the data receiver. During the second device ingestion for subject #9, data transmission also stopped abruptly in the colon. In this instance, given the extended period of time that had transpired (approximately 19 h), the failure in data transmission was thought to be secondary to a loss of battery strength in the data receiver as opposed to a defect in the capsule. Alternatively, it is possible that the subject had passed the pill device with the data receiver beyond the four feet range from toilet and then failed to record an event surrounding the bowel movement. This was a common occurrence as 19/125 subjects of Sarosiek’s study had similar issues.6 Our study is limited by a small sample size of 10 young subjects, ranging from age 18 to 26. Future studies with a larger and more heterogeneous population (including geriatric patients) would further validate the findings of this study and broaden the applicability of the results. Furthermore, despite controlling the meals consumed with each device ingestion, the food consumed in between each ingestion was not standardized. Another variable that could be controlled before and between ingestions is physical activity, which could potentially affect gut transit and pH. Also, although this study characterizes variability with mean pH values, it does not provide insight into the variability of the physiologic mechanisms responsible for the mean pH values calculated at any given location and whether there is intrapatient variation within these mechanisms (e.g., a brisk bicarbonate response with a low baseline pH in AGE and a slow bicarbonate response with a high baseline pH would create similar mean pH values for AGE). Finally, future studies could include two blinded analysts to eliminate any possible bias or errors in the identification of landmarks. In conclusion, this is the first study to assess tandem gut transit and pH assessments in healthy subjects. Our results demonstrate that within 24 h the gut pH profile remains relatively stable for an individual, with more variation seen in the colon as compared with the small bowel, and that small bowel transit time is less variable than colonic transit time. These findings provide important information to define accuracy parameters in physiologic studies and contribute to the information for future research as well. Guarantor of the article: David T. Rubin, MD. Specific author contributions: Designed the study: David T. Rubin; performed the research: Adam E. Mikolajczyk and Bonnie L. Surma; performed the data analysis: Adam E. Mikolajczyk and Sydeaka Watson; wrote the manuscript: David T. Rubin and Adam E. Mikolajczyk. Financial support: This project was funded in part by the Calvin Fentress Fellowship Award through the Pritzker School of Medicine and the Digestive Disease Research Core Center of the University of Chicago (DK42086). Potential competing interest: David T. Rubin previously served as a consultant for SmartPill for an unrelated matter. This current study was entirely investigator initiated. Writing support was provided by Elin Raun-Royer, who was employed by the University of Chicago Medicine.