Non-invasive ¹³C-glucose breath test using residual gas analyzer-mass spectrometry: a novel tool for screening individuals with pre-diabetes and type 2 diabetes

Type 2 diabetes mellitus is the most common deleterious metabolic disease in the 21st century and has become one of the most pressing human health concerns all over the world, with an estimated high prevalence of 101 million individuals by the year 2030 in India [1–3]. Several pieces of evidence suggest that insulin resistance and pancreatic β-cell dysfunction play an important role in the pathogenesis of type 2 diabetes [4, 5]. Insulin resistance is also associated with a cluster of risk factors (referred to as metabolic syndromes) for increased cardiovascular disease [6]. It is still the subject of debate when or how to recognize individuals at high risk of altered insulin action or during the preclinical phase of type 2 diabetes [7]. Hence an accurate and fast practical diagnosis of pre-diabetes prior to the onset of type 2 diabetes remains a challenge.

Recently, the ¹³C-glucose breath test (¹³C-GBT) [8] has been proposed as a non-invasive method for assessing impaired glucose tolerance in comparison to the gold standard direct invasive method called the hyperinsulinemic-euglycemic clamp [5, 8] and the surrogate methods, such as homeostasis model assessment (HOMA) and quantitative insulin sensitivity check index (QUICKI) [9], estimated from the measurement of blood glucose and plasma insulin. The ¹³C-GBT is based on the principle that when a dose of ¹³C-labelled glucose substrate is orally ingested and metabolized, ¹³CO₂ is exhaled by the respiratory system. In diabetes mellitus, it is expected that less ¹³CO₂ will be exhaled because of impaired glucose uptake by the cells. However, the widespread clinical efficacy of the ¹³C-GBT, exhibiting diagnostic sensitivity, specificity, test accuracy, and positive and negative predictive values along with optimal diagnostic cut-off points for large-scale screening individuals with non-diabetes, pre-diabetes and type 2 diabetes mellitus, has not yet been explored. Furthermore, to the best of our knowledge, no studies to date have elucidated whether endogenous CO₂ production related to the basal metabolic rates (BMR) in individuals would affect the diagnostic accuracy of the ¹³C-GBT for pre-diabetes and type 2 diabetes. Therefore there is a pressing need to accurately evaluate the ¹³C-GBT for routine clinical practice.

The ¹³C-GBT is used to determine the ¹³CO₂/¹²CO₂ isotope ratios in exhaled breath samples, usually expressed as the delta-over-baseline (DOB) values i.e. δ_DOB¹³C‰, and have previously been demonstrated using conventional high-precision gas-chromatography coupled with isotope ratio mass-spectrometry (GC-IRMS [8]). Although traditional GC-IRMS methodologies are highly reliable, there remain some intrinsic drawbacks. In particular, GC-IRMS systems are fairly costly, not portable, require specialized expertise to operate and maintain, and are not suitable for real-time on-line measurements—facts that have hindered significantly the widespread applicability of GC-IRMS systems for breath analysis as a daily decision-making tool for use at point-of-care (POC) facilities. It is therefore of immense interest to develop a simple, robust and cost-effective alternative non-invasive diagnostic tool for the ¹³C-GBT that can reliably and accurately analyze breath samples in real-time at the point-of-care. We have recently developed and validated a simple, low-cost residual gas analyzer-mass spectrometry (RGA-MS) method for non-invasive diagnosis of H. pylori infection [10]. The aim of the present study was therefore to standardize and validate this novel analytical RGA-MS method in the ¹³C-GBT for non-invasive assessment of pre-diabetes and type 2 diabetes mellitus in real time, and eventually to explore its true potential for routine clinical practice.

In the present study, we report for the first time the clinical utility of the ¹³C-GBT using a simple RGA-MS system as an alternative non-invasive method for screening individuals at risk for diabetes. The diagnostic accuracy and the feasibility of ¹³C-GBT by RGA-MS were also validated by comparison with an established laser-based cavity-enhanced absorption technique called integrated cavity output spectroscopy (ICOS). Finally, we determined numerous diagnostic, statistically significant parameters of the ¹³C-GBT including sensitivity, specificity, and the optimal cut-off point for the δ_DOB¹³C‰ values to delineate the applicability of the RGA-MS technique as a POC non-invasive diagnostic tool for large-scale diabetes screening tests.

2.1. Subjects

2.2. Breath sample collection

2.3. Residual gas analyzer-mass spectrometry (RGA-MS)

We employed the residual gas analyzer-mass spectrometry (RGA-MS) that exploits a conventional quadrupole mass filter technology with a Faraday cup detector to measure the masses of ¹³C¹⁶O¹⁶O (45 amu) and ¹²C¹⁶O¹⁶O (44 amu) isotopes in exhaled breath samples. A schematic diagram of the RGA-MS system for the analysis of exhaled breath samples is depicted in figure 1.

Zoom In Zoom Out Reset image size

The RGA-MS system and its potential for measuring carbon isotopes with a typical precision of ±0.25‰ in breath samples has been described in detail in our previous study [10]. In brief, an RGA was coupled with a high vacuum chamber (~9.0 × 10⁻⁸ Torr) and the baseline vacuum was achieved by two turbo-molecular pumps backed up with a diaphragm pump. The vacuum chamber was equipped with two all-metal leak valves and a manually actuated gate valve to control flow of breath samples into the vacuum chamber as well as to sustain the working pressure (2.1–2.4 × 10⁻⁷ Torr) for the measurements. The ion currents for the masses 44 and 45 amu were measured with a scanning rate of 0.1 s amu⁻¹ by Quadera software (Prisma plus, version 4.50) in the selected multiple ion current detection (MID) mode. Typically a total number of 15–20 data points were recorded for each mass to calculate the δ¹³C‰ values and subsequently the results were reported as the δ_DOB¹³C‰ values, i.e. the difference in the δ¹³C‰ values between the baseline and 120 min breath samples as defined below:

$$\begin{eqnarray}&&{{\delta}_{\text{DOB}}}{}_{{}}^{13}\text{C \unicode{x2030}}=\left[\left(\frac{{{R}_{\text{sample}}}}{{{R}_{\text{standard}}}}-1\right)\right]\times 1000 \end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{{\delta}_{\text{DOB}}}{}_{{}}^{13}{{\text{C \unicode{x2030}}}_{13\text{}}}={{\left({{\delta}^{13}}\text{C \unicode{x2030}}\right)}_{t=t \min}}-{{\left({{\delta}^{13}}\text{C \unicode{x2030}}\right)}_{t=0 \min}}\end{eqnarray} \tag{ 2 }$$

where R_sample is the ¹³C/¹² C isotope ratio of the sample and R_standard is the international standard Pee Dee Belemnite (PDB) value i.e. 0.0112372.

2.4. Integrated cavity output spectroscopy (ICOS)

2.5. Validation of δ_DOB¹³C‰ measurements by RGA-MS and ICOS for¹³C-GBT

2.6. Statistical analysis

In the present study, box and whisker plots were utilized to assess the distribution of ¹³CO₂ enrichments in exhaled breath samples for NDC, PD and T2D individuals. Figure 2 illustrates the box and whisker plots of δ_DOB¹³C‰ values measured by both RGA-MS and ICOS methods. We observed that the mean DOB value for the group with T2D (mean δ_DOB¹³C‰ = 13.46‰) measured by RGA-MS was significantly lower (p < 0.001) compared to the groups with NDC (mean δ_DOB¹³C‰ = 35.94‰) and PD (mean δ_DOB¹³C‰ = 24.80‰). In cases of T2D and pre-diabetes, glucose uptake would be impaired because of diminished pancreatic insulin secretion or impaired insulin action on the target tissue [16], resulting in blunted glucose oxidation and consequently the reduced rate of generation of ¹³CO₂ in exhaled breath samples.

Zoom In Zoom Out Reset image size

Figure 2 also demonstrates that the present ¹³C-GBT using the RGA-MS technique was capable of detecting marked differences in ¹³C-enriched glucose-derived δ_DOB¹³C‰ values at 120 min in exhaled breath samples among the groups with NDC, PD and T2D, suggesting a robust approach for large-scale screening testing without the need for invasive blood sampling. We also observed that δ_DOB¹³C‰ values in breath samples analyzed by both mass spectrometry (RGA-MS) and laser-based spectroscopy (ICOS) techniques were statistically almost similar (35.95‰ versus 36.04‰, 24.80‰ versus 24.94‰ and 13.46‰ versus 13.34‰ for NDC, PD and T2D, respectively) as depicted in figure 2, demonstrating the clinical potential of a simple RGA-MS technique as an alternative non-invasive analytical tool for real-time exhaled breath analysis to precisely diagnose individuals with NDC, PD and T2D. Moreover, to compare the δ_DOB¹³C‰ results of the ¹³C-GBT analyzed by RGA-MS to those determined by the ICOS system, regression statistics were employed.

Figure 3 shows a least-squares regression plot of the correlation for δ_DOB¹³C‰ values from both methods. We found a close correlation as indicated by R² = 0.979, validating the RGA-MS technique to accurately detect ¹³CO₂/¹²CO₂ isotope ratio measurements in exhaled breath samples for non-invasive evaluation of glucose metabolism in the diagnosis of diabetes.

Zoom In Zoom Out Reset image size

To investigate the diagnostic performance of the RGA-MS technique and the accuracy of the ¹³C-GBT to precisely distinguish between individuls with NDC and PD, and individuals with PD and T2D, we performed an ROC curve analysis. An ROC curve was constructed by plotting the true positive rate (sensitivity) against the false postive rate (1-specificity) using the δ_DOB¹³C‰ values, and subsequently the highest values of sensitivity and specificity were used to determine the optimal diagnostic 'cut-off point' for the present ¹³C-GBT using the RGA-MS method. Figures 4(a) and (b) depict ROC plots for the ¹³C-GBT to distinguish NDC and PD cases, as well as PD and T2D cases, respectively. The optimal diagnostic cut-off point was determined to be δ_DOB¹³C‰ (cut-off) = 28.81‰ to correctly diagnose individuals with NDC and PD using the RGA-MS method and this corresponded to a sensitivity of 100% and specificity of 94.4%.

Zoom In Zoom Out Reset image size

We also determined another diagnostic cut-off point of δ_DOB¹³C‰ (cut-off) = 19.88‰ between individuals with PD and T2D, which exhibited a sensitivity and specificity of 100% and 95.5%, respectively, establishing a clinical feasibility of the RGA-MS technique as a sufficiently robust detection method for an accurate diagnosis of PD and T2D. Consequently, in the present study individuals above δ_DOB¹³C‰ (cut-off) = 28.81‰ were considered NDC and individuals with δ_DOB¹³C‰ (cut-off) ⩽ 19.88‰ were considered T2D, whereas individuals with 28.81 ⩾ δ_DOB¹³C‰ > 19.88 were considered to be pre-diabetes (PD). In both cases, the area under the ROC curve, which is a global measure of accuracy in a diagnostic test, was determined to be 0.99 (p < 0.001), showing the validity of the ¹³C-GBT. We subsequently explored the positive and negative predictive values (PPV and NPV) of the present ¹³C-GBT by means of an RGA-MS method for diagnostic assessment. The PPV and NPV are usually defined by the patients' probability of getting diseased states when the actual test results are known [17]. We found a diagnostic PPV and NPV of 95.65% and 100%, respectively, between subjects with NDC and PD, and 96.15% and 100%, respectively, between subjects with PD and T2D, indicating excellent diagnostic predictions. Table 2 shows the important diagnostic parameters of the present ¹³C-GBT that may be practical for large-scale screening purposes. Bargueno et al [18] pointed out that an ideal diagnostic method should have sensitivity and specificity each close to 100%; however, our RGA-MS methodology for the ¹³C-GBT manifested a sufficiently robust method as indicated by the acquired values of sensitivity (i.e. 100%) and specificity (i.e. about 95%). Consequently, to the best of our knowledge, the results of the present ¹³C-GBT using the RGA-MS method demonstrate, for the first time, the clinical applicability of a cost-effective diagnostic method as opposed to a GC-IRMS method as previously used.

We finally investigated whether there was any effect of endogenous CO₂ production related to the basal metabolic rates (BMR) in individuals on the diagnostic accuracy of the present ¹³C-GBT using the RGA-MS method. The accurate determination of the endogenous CO₂ production in the ¹³C-breath test is still a controversial issue [19]. Endogenous CO₂ production is usually influenced by age (adults > children), weight, height and sex (male > female) [20] and consequently the δ_DOB¹³C‰ values are also expected to vary in accordance with these factors. Therefore to assess the effect of endogenous CO₂ production in the present ¹³C-GBT, we applied the Mifflin–St Joer equations [21] to calculate the BMR based on age, weight and height of either sex and the following equations [20] were used:

$$\begin{eqnarray}&&\text{C}{{\text{O}}_{2}}\, \text{production}\, \text{rate} \left(\text{mmol} / \min \right)=\frac{{{V}_{\text{C}{{\text{O}}_{2}}}}\times 1000}{24\times 60} \end{eqnarray} \tag{ 3 }$$

$$\begin{eqnarray}&&{{V}_{\text{C}{{\text{O}}_{2}}}}\left(\text{mol}/\text{day}\right)=\frac{\text{Energy}\, \text{expenditure}\, \left(\text{EE}\right)}{134.25} \end{eqnarray} \tag{ 4 }$$

$$\begin{eqnarray}&&\text{Energy}\,\, \text{expenditure}\,\,\left(\text{EE}\right)\left(\text{kcal}/\text{day}\right)=\text{Physical}\,\,\text{activity}\,\,\text{level}\left(\text{PAL}\right)\times \text{BMR}\end{eqnarray} \tag{ 5 }$$

Figure 5 depicts box and whisker plots of the endogenous CO₂ production rate for three different groups of subjects. We observed that there were no statistically significant differences (p = 0.37 and 0.73) among the endogenous CO₂ production rates estimated for normal, pre-diabetes and type 2 diabetes, indicating the negligible effect of BMR-based endogenous CO₂ production on diagnostic accuracy.

Zoom In Zoom Out Reset image size

This study confirms the clinical feasibility of a novel RGA-MS method for accurate evaluation of the ¹³C-GBT in the diagnosis of non-diabetes control, pre-diabetes and type 2 diabetes in real-time, thus making it a valid and potentially robust non-invasive diagnostic tool for routine clinical practices at the point-of-care. Our results also suggest that the ¹³C-GBT using a simple RGA-MS method can reliably assesses the changes in glucose metabolism in real time. The present ¹³C-breath analysis instrument is compact, easy-to-run, more portable and inexpensive compared to currently available optical spectroscopy and MS-based detection methods, suggesting a simple alternative non-invasive screening tool for the measurement of high-precision δ_DOB¹³C‰ values in exhaled breath samples, not only from diabetes or non-diabetes, but also from any other disease or metabolic disorder. However, the cost of the ¹³C-glucose breath test may be considered to be of concern from the economic point of view, since ¹³C-glucose is much more expensive than blood sugar analysis. This is the main limitation of the ¹³C-glucose breath test (¹³C-GBT) for screening diabetes. Nevertheless, utilizing the RGA-MS methodology rather than GC-IRMS, the overall cost of the ¹³C-GBT will likely not be prohibitive in the near future in order to screen diabetes even in economically developing countries like India.

This work was supported by the S.N. Bose National Centre for Basic Sciences (Grant No: SNB/MP/11–12/69). The authors further acknowledge the Department of Science & Technology (DST, India) for PhD Inspire Fellowships (A Maity and G D Banik), and the S.N. Bose Centre for Junior Research Fellowships (S Som and C Ghosh). We are also grateful to all volunteers who eagerly participated in the present study.