Comparison of Volatile Sulfur Compound Concentrations Measured with a Sulfide Detector vs . Gas Chromatography

Abstract

The accuracy of the Halimeter®, an inexpensive, simple instrument that measures total breath volatile sulfur compounds (VSCs), has not been adequately tested. We compared Halimeter® measurements with those obtained with a specific and sensitive gas chromatographic (GC) technique. The Halimeter® gave different, bi-exponential responses to a constant concentration of different VSCs: The relative response rate and sensitivity were hydrogen sulfide > methyl mercaptan > dimethylsulfide. The transient peak VSC concentration of oral samples was reached long before the sulfide detector fully responded. The GC measurement of initial total VSCs in breath samples was 2.7 ± 0.48 times greater than the peak concentration of the Halimeter®. However, the plateau phase measurement of the Halimeter® was 25% greater than that of GC. While GC and Halimeter® measurements positively correlated, appreciable differences were observed. In studies where relatively precise VSC measurements are required, GC is the preferable technique.

INTRODUCTION

The concentration of volatile sulfur-containing compounds (VSCs) in breath-gas correlates with organoleptic assessment of breath malodor (Tonzetich and Ng, 1976; Schmidt et al., 1978; Rosenberg et al., 1991b; Niles and Gaffar, 1993), and VSCs are considered to be a major cause of breath malodor (Schmidt et al., 1978; Kleinberg and Codipilly, 1995; Tonzetich, 1971, 1977; Yaegaki and Sanada, 1992). As a result, VSC concentrations are used as a simple means of objectively documenting the existence of bad breath and assessing the benefit of therapy.

Many investigators, as well as practicing dentists, use an instrument known by its trade name, the Halimeter® (Interscan Corporation, Chatsworth, CA, USA), to measure oral VSCs (Rosenberg and McCulloch, 1992; Richter, 1996; Frascella et al., 2000). This instrument provides a digital readout of the total VSC concentration in gas aspirated from the oral cavity. While relatively inexpensive and easy to use, there is a paucity of data concerning the accuracy of measurements obtained with this instrument. In this study, we compared Halimeter® measurements with those obtained by means of a gas chromatograph (GC) equipped with a sulfur detector.

MATERIALS & METHODS

VSC Concentrations of Standard Gases

The four VSCs observed in human breath were hydrogen sulfide (H₂S), methyl mercaptan (CH₃SH), dimethylsulfide (CH₃SCH₃), and carbon disulfide (CS₂). Authentic H₂S, CH₃SH, CH₃SCH₃ (purchased from BOC Gases, Port Allen, LA, USA), and CS₂ (purchased from Sigma, St. Louis, MO, USA) were added to polypropylene bags containing 5000 mL of air. The final VSC concentrations were approximately 50, 300, and 800 ppb. Each bag was sealed with a three-way stopcock, one arm of which was attached to the intake line of the Halimeter® via a straw (as recommended by the manufacturer to aspirate oral samples). A piece of rubber tubing had previously been fit snugly around the straw to create an airtight seal with needle penetration. The instrument was zeroed as the Halimeter® aspirated air through the open arm of the three-way stopcock. The stopcock was then adjusted to allow the detector to aspirate the standard gas into the Halimeter®. At 1.5 min, a needle attached to a 5-mL polypropylene syringe was inserted through the rubber tubing into the lumen of the straw. A 2.5-mL sample of gas was aspirated and re-injected (to eliminate dead space effect), a 4-mL sample was aspirated, and the syringe was sealed. This gas was analyzed for VSC concentration by GC. Studies were carried out in triplicate.

For a mathematical description of sulfide detector output curves, recordings were scanned into a computer, and coordinates of multiple points on the time vs. concentration curves were determined by means of a computerized cursor. We determined the equation describing the curve by fitting the coordinates with a two-exponential, four-parameter function using the Levenberg-Marquardt method (More, 1977).

VSC Concentrations in Mouth Gas

The Institutional Review Board approved the experiments, and informed consent was obtained from all subjects. Eighteen healthy subjects provided breath samples between 9 and 11 a.m. To maximize oral VSCs, subjects did not brush their teeth or ingest solid food the morning of testing. A previously described (Springfield et al., in press) collecting device (the barrel of a 1-mL syringe fixed between portions of a tongue blade) was placed in the mouth. The lips were sealed around the protruding syringe, and the subject breathed through the nose for 3 min. A 5-mL syringe containing about 8 mg of calcium chloride as a drying agent was then connected to the syringe via a stopcock. (Calcium chloride does not adsorb VSCs.) After several rapid aspirations and re-infusions of oral gas, a 5-mL sample was removed for GC analysis. The mouthpiece was immediately connected to the sulfide detector via a straw, and the stopcock turned to allow aspiration of oral gas. Following the manufacturer' instructions, the lips were separated to allow ingress of air into the oral cavity. The detector output was recorded, and when a plateau was reached, a 4-mL gas sample for GC analysis was rapidly withdrawn from the straw via puncture through the rubber tubing.

Volume of Gas in Oral Cavity

To determine the turnover of oral gas during Halimeter® studies, we measured the gas volume of the oral cavity. Subjects closed their lips around the gas-sampling apparatus, and the cheeks were relaxed such that the oral cavity was allowed to assume its usual configuration. A 3-mL syringe was used to instill 2 mL of air containing approximately 20,000 ppm of methane into the mouth. The subjects breathed through the nose for 30 sec, and then 3 mL of oral gas was aspirated and re-infused 3X, and a 3-mL sample was removed for analysis. Methane concentration was determined by GC with the use of a flame ionization detector. Oral gas volume was determined from the observed dilution of methane.

Gas Chromatographic Technique

VSC concentrations were determined by GC with a 2.4 m x 3.1 mm Teflon column packed with Chromosil 330 maintained at 80°C. The carrier gas was nitrogen (flow, 25 mL/min). A sulfur chemiluminescence detector that specifically responds to sulfur (Sievers, Model 355, Boulder, CO, USA) was used. A 0.3-mL aliquot of gas was injected into the column. VSCs were identified by characteristic retention times and quantitated via comparison of peak area with that of dilutions of standards.

RESULTS

The sulfide detector yielded a bimodal response when exposed to a constant concentration of H₂S, CH₃SH, or CH₃SCH₃ (Fig. 1). (The sensor did not respond to CS₂, and CS₂ concentrations are not depicted.) Each curve, which had a rapid response followed by a slower response, could be almost perfectly fit by a bi-exponential equation with 4 parameters: $\batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \[Detector\ response\ =\ A(1-e^{-t/t1})\ +\ B(1-e^{-t/t2})\] \end{document}$ where A and B are the fractions of the total response represented, respectively, by the fast and slow components of the detector, and t1 and t2 are the respective time constants of the rapid and slow responses. The 1/2 times of the two responses were calculated from the time constant/0.693. The relative rate of response to the VSCs was H₂S > CH₃SH > CH₃SCH₃ (Table ).

The sulfide detector' response to aspiration of a subject' oral gas typically was a sharp peak followed by a plateau phase (see Fig. 1 ). The time elapsing between the initiation of the rise in VSC concentration and the peak averaged 4.1 ± 0.44 sec (n = 14).

Sulfide detector measurements at 6 sec after exposure to a constant H₂S concentration were almost identical to that of the GC, while CH₃SH measurements were about 31% less (see Fig. 2 ). In contrast, H₂S measurements at 1.5 min exceeded GC values by about 49%, while CH₃SH values were virtually identical to those of the GC. The sulfide detector markedly underestimated CH₃SCH₃ concentrations at both 6 sec and 1.5 min (Fig. 2 ). The sulfide detector responded linearly to H₂S and CH₃SH, whereas the detector appeared to disproportionately reflect the highest concentration (800 ppb) of CH₃SCH₃ (Fig. 2 ).

While a positive correlation existed between the sum of VSCs as measured by GC and the peak value of the sulfide detector (r = 0.73; p < 0.01), the GC value averaged 2.7 ± 0.48 times greater than that of the sulfide detector (p < 0.01) (see Fig. 3A ). For the 9 samples with a VSC concentration of less than 100 ppb (via GC), the sulfide detector measurement actually was 67% greater than that of the GC. For the 45 samples with VSC concentrations greater than 100 ppb, the GC values averaged 3.4 times that of the sulfide detector.

The sulfide detector slightly overestimated the concentration of VSCs during the plateau phase, averaging 22 ± 4.0% more than the GC (p < 0.001) (Fig. 3B ). Measurements obtained with the two techniques closely correlated (r = 0.97; p < 0.01).

The volume of gas in the oral cavity averaged 27 ± 3 mL (n = 9).

DISCUSSION

Measurements of breath VSCs obtained with the Halimeter® differ from those achieved by other diagnostic analyses. The recommended procedure for this instrument includes an initial three-minute period during which the subject breathes through the nose with the lips sealed. A straw attached to the detector is then inserted into the slightly open mouth, and a pump withdraws oral gas at a rate of roughly 10 mL per sec. Since the oral cavity contains about 27 mL of gas, oral gas is turned over at a rate of about 33% per sec during these measurements. Thus, while most diagnostic measurements are made in the steady-state situation, Halimeter® measurements are obtained in a markedly perturbed, non-steady state created by the instrument. As a result, the sulfide detector records a sharp peak VSC concentration about 4 sec after the initiation of oral aspiration followed by “plateau” concentration (see Fig. 1 ). With some exceptions (Rosenberg et al., 1991a), most published studies have reported only peak values.

Another complicating feature of the Halimeter® is its relatively slow, bi-exponential response to an input of a constant VSC concentration (Fig. 1). This detector acts as if it has two compartments with different capacities and different saturation rates, depending upon the VSC to which it is exposed. For H₂S, the rapid and slow responding phases of the detector, which accounted for about 70% and 30% of the maximal readout, had 1/2 times of 4.1 sec and 69 sec, respectively (Table ). The instrument responded slightly slower to CH₃SH and still more slowly to CH₃SCH₃. This slow response would not be a problem if the instrument were measuring a single VSC in a steady-state situation, since the detector could be exposed to the test gas for a sufficient period to provide a full response. However, since oral aspiration rapidly reduces the baseline level of oral VSCs, a sharp peak concentration is reached at about 4.1 sec, too short a period to allow for the full detector response. The extent to which the observed peak concentration underestimates the true concentration at the detector depends on the calibration technique. If calibrated such that the correct concentration is recorded after a several-second exposure, peak concentration errors will be minimized, while the steady-state concentration will be overestimated. If calibrated to yield the correct concentration when the detector has achieved its maximal deflection, peak concentrations will be drastically underestimated.

The final unusual aspect of the Halimeter® is that the user ordinarily does not calibrate the machine with standards of known concentrations. Rather, the machine is calibrated at the factory, and the user assumes that the detector' readout is correct. As discussed above, exactly how the instrument should be calibrated is not clear.

The present experiments were designed to determine the accuracy with which the Halimeter® measures VSCs in both the experimental situation and in the clinical setting when gas is aspirated from the oral cavity. To this end, Halimeter® measurements were compared with those obtained with a GC equipped with a sulfur chemiluminescence detector. Since the GC was calibrated with known concentrations of standard VSCs, GC was assumed to provide the correct value.

The accuracy of the sulfide detector when exposed to a constant concentration of a VSC varied with the individual VSC and the time point of the measurement. For example, Halimeter® readings for H₂S made at 6 sec were virtually identical to those obtained with the GC, while measurements at 1.5 min averaged 49% greater than the GC value (Fig. 2 ). The response of the Halimeter® to CH₃SH was slightly lower and slower than for H₂S; the six-second measurement underestimated the true value by about 31%, whereas the 1.5-minute measurement was quite accurate. The Halimeter® markedly underestimated CH₃SCH₃ concentrations at both the six-second and 1.5-minute time points.

Given the above findings with constant inputs of VSCs, it was not surprising to find that Halimeter® measurements of peak and plateau concentrations of oral VSCs did not perfectly mirror the results of GC analysis. Halimeter® measurements of peak VSC concentrations markedly underestimated the VSC concentration of oral gas obtained just prior to the initiation of aspiration into the Halimeter®. At clinically significant VSC levels (>100 ppb by GC), the GC measurement averaged 3.4 times the peak value of the sulfide detector. This result is readily explained in that the Halimeter®: (a) dilutes the initial oral gas concentration via aspiration of ambient air into the mouth; (b) responds too slowly for accurate measurement of the true peak concentration; and (c) has a low response to some VSCs measured by the GC (CH₃SCH₃ and CS₂). The finding that, at very low oral VSC levels, the sulfide detector peak values frequently exceeded those of GC suggests that the Halimeter® responded to oral components other than VSCs.

In contrast to the situation with peak measurements, plateau values measured by the sulfide detector were slightly higher than those of the GC (Fig. 3B ). This presumably reflects the overestimation of H₂S (usually the predominant gas) when the sulfide detector becomes equilibrated with this gas.

The appreciable discrepancy between Halimeter® and GC measurements indicates that studies requiring precise knowledge of VSC concentrations require GC analysis. However, the sulfide detector responds linearly to H₂S and CH₃SH, the primary VSCs in breath gas. In addition, the peak and plateau concentrations as recorded by the Halimeter® significantly correlated with the initial and plateau values obtained with GC. Thus, while lacking perfect accuracy, the sulfide detector provides useful data for the clinical studies of oral malodor.

Table. Responses of the Sulfide Detector to a Constant Input of Various Concentrations^a of VSCs

VSC	Rapid Response^b		Slow Response^b
	% of total response	1/2 time (sec)	% of total response	1/2 time (sec)
^a Three measurements were made at three different VSC concentrations (50, 300, and 800 ppb), and the data presented represent the mean ± SEM of 9 output curves for each VSC.
^b The responses of the detector to exposure to a constant concentration of a VSC were fitted to two-exponential, four-parameter equations. The data provided are the percent of the total response and the 1/2 time of the response for the rapid and slow phases of the detector. In a column, values with different superscript letters differ from each other with a p value of < 0.05.
H₂S	69 ± 2.4^c	4.1 ± 0.42^c	31 ± 2.4^c	69 ± 3.4^c
CH₃SH	59 ± 5.5^d	5.9 ± 0.42^d	41 ± 5.5^d	40 ± 4.8^d
CH₃SCH₃	32 ± 5.2^e	12.1 ± 3.2^d	68 ± 5.3^e	728 ± 372^d

Figure 1.

Typical sulfide detector output curves recorded during the input of constant concentrations (about 300 ppb) of H₂S, CH₃SH, or CH₃SCH₃. Superimposed on these curves is a typical response of the sulfide detector during the measurement of oral VSCs in a volunteer. Note the early peak VSC concentration is followed by a lower, relatively constant (“plateau”) concentration.

Figure 2.

Plots of the concentrations of H₂S, CH₃SH, or CH₃SCH₃ measured with the sulfide detector vs. the concentrations measured with GC. Data are shown for sulfide detector measurements recorded at 6 sec and 1.5 min after initiation of a constant input of the individual VSCs at concentrations of about 50, 300, and 800 ppb (measured by GC). The line of identity for paired measurements is indicated by the dotted line.

Figure 3.

(A) Plot of peak oral VSC concentration measured by the sulfide detector vs. the GC-measured VSC concentration in oral gas obtained immediately before initiation of aspiration into the sulfide detector. Data represent triplicate measurements made in 18 individuals. The three determinations made for each individual are denoted by a common symbol. (B) Plot of plateau oral VSC concentration measured by the sulfide detector vs. the GC measurement of total VSC concentration in gas obtained during the plateau phase. Data represent triplicate measurements made in 18 individuals, and the three determinations made for each individual are denoted by a common symbol.