Methods
Validation of Changes in Metabolic Fuel Usage Using Lumen Vs. Gold Standard
Three studies were conducted in two separate and independent institutes (Institution 1 and Institution 2). A total of 54 healthy subjects arrived at the clinic after fasting for 12 hours. Their RER was measured
using metabolic cart gold standard devices (ParvoMedics TrueOne 2400 and Quark RMR, Cosmed) and the Lumen device. Next, they drank a high-carb shake and both measurements were taken again
after 90 min and again after 180 min.
Validation of Lumen Repeatability in Different Metabolic States
In the third study (Institution 2, 2018), Lumen was tested before and after measurements were taken with the metabolic cart, for test-retest reliability.
Validation of Changes in Metabolic Fuel Usage Following High Carb Meal Using Lumen
In addition, on day 4 subjects were measured 90 mins after a high-carb meal.
Validation of Changes in Metabolic Fuel Usage Following Exercise Using Lumen
20 healthy subjects performed a running exercise at a maximum HR of 70 bpm for 60 mins. Measurements were taken using Lumen before and after the activity at a resting HR.
Validation of Changes in Metabolic Fuel Usage Using Lumen Vs. Gold Standard
In all three validation studies, Lumen Index exhibited comparable results to gold standard RER. Bland-Altman plots showed that 95% of data points were within agreement range of 2SD, determining Lumen Index shows similar measurement of gold-standard metabolic cart (Figure 1).
Figure 1.
Bland-Altman plots of the three validation studies: (A) Institution 1, 2017; n=39 (B) Institution 2, 2017; n=54 (C) Institution 2, 2018; n=69. Solid line represents the mean bias between the two measures and dotted lines represent ± 2 standard deviations from the mean (limits of agreement).
In all three validation studies, ANOVA of repetitive measurements showed Lumen Index to effectively differentiate between the different metabolic states of an individual following a high-carb meal (P < 0.0001 in all three studies; Figure 2). Tukey’s post hoc analysis revealed significant increase 90 min (P = 0.0001, P < 0.0001, P < 0.0001), and 180 min (P = 0.0032, P = 0.0101, P = 0.0067) after high-carb meal. Moreover, changes in Lumen Index in response to diet corresponded to changes exhibited in gold-standard RER (RM-ANOVA: P < 0.0001 in all three studies). Tukey’s post hoc analysis revealed significant increase 90 min (P < 0.0001, P < 0.0001, P < 0.0001), and 180 (P = 0.0111, P < 0.0001, P< 0.0001) min following a high-carb meal.
Figure 2.
(A) Lumen Institution 1, 2017; n=13. (B) Lumen Institution 2, 2017; n=18. (C) Lumen Institution 2, 2018; n=23. (D) RER Institution 1, 2017; n=13. (E) RER Institution 2, 2017; n=18. (F) RER Institution 2, 2018; n=23. Dots represent mean ± SEM. *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001.
In addition, these changes from baseline were correlated between Lumen Index and gold-standard RER (Pearson: P = 0.0007, P < 0.0001, P = 0.0041; Figure 3).
Figure 3.
A) Institution 1, 2017: Pearson r = 0.6211, p = 0.0007, Y = 0.3240*X + 4.848 (B) Institution 2, 2017: Pearson r = 0.6112, p < 0.0001, Y = 0.3979*X + 1.673 (C) Institution 2, 2018: Pearson r = 0.4295, p = 0.0041, Y = 0.3516*X - 2.616.
Validation of Lumen Repeatability in Different Metabolic States
Lumen showed high test-retest reliability when measured before and after the metabolic cart. Percentage of CO2 levels measured by Lumen were highly correlated in all three timepoints, indicating the repeatability of Lumen (P < 0.0001 in all three states; Figure 4).
Figure 4.
(A) Lumen %CO2 after fasting: Pearson r = 0.7842, p < 0.0001, Y = 0.8059*X + 0.9804 (B) Lumen %CO2 90 min post high-carb meal: Pearson r = 0.8851, p < 0.0001, Y = 0.7798*X + 1.243 (C) Lumen %CO2 180 min post high-carb meal: Pearson r = 0.8740, p < 0.0001, Y = 0.9723*X + 0.2306
Validation of Changes in Metabolic Fuel Usage Following Low & High Carb Diet Using Lumen
In an ANOVA repetitive measurements Lumen Index was found to effectively differentiate between the different metabolic states of an individual following a low and high carb diet (P = 0.0004; Figure 5A). Tukey’s post hoc analysis revealed significant reduction of 36% following a low-carb diet (P = 0.0013), which later increased by 37% following a high-carb diet (P = 0.0009). Moreover, changes in
Lumen Index in response to diet corresponded to changes in weight (RM-ANOVA: P < 0.0001; Figure 5B), which were also reduced following a low-carb diet (P < 0.0001), and increased following a high-carb diet (P = 0.0041).
Figure 5.
A) Distribution of Lumen index change following low and high carb diets. Box plot whiskers represent minimum and maximum. (B) Changes in the Lumen index and weight following low and high carb diets. Dots represent mean ± SEM. **P<0.01, ***P<0.001, ****P<0.0001. n=8
Validation of Changes in Metabolic Fuel Usage Following High Carb Meal Using Lumen
In a t-test analysis, Lumen index was found to effectively differentiate between the different metabolic states of an individual following a high-carb meal (P < 0.0001; Figure 6).
Figure 6.
Distribution of Lumen index change following high-carb meal. Box plot whiskers represent minimum and maximum. ****P<0.0001. n=8.
Validation of Changes in Metabolic Fuel Usage Following Exercise Using Lumen
Lumen Index decreased in 16 out of 20 subjects following physical activity (80%; Figure 7A). In a t-test analysis, Lumen Index was found to effectively differentiate between the different metabolic states of an individual following exercise (P = 0.001; Figure 7B).
Figure 7.
(A) Lumen index change following exercise (n=20). (B) Distribution of Lumen index change following exercise. Box plot whiskers represent minimum and maximum. **P<0.01.