Abstract
The aim of this work was to develop a spectrophotometric method for the measurement of salivary β-galactosidase for the evaluation of oral malodor. A comparison between different methods for estimating oral malodor has been conducted on 94 healthy adult volunteers. For organoleptic measurements, the subjects were instructed to exhale briefly through the mouth at a distance of approximately 10 cm from the noses of two trained judges. The evaluation of β-galactosidase activity was accomplished in unstimulated whole saliva of all participants through the colorimetric method and spectrophotometry. A significant association among spectrophotometric evaluation of β-galactosidase and organoleptic measurements was assessed by Spearman correlations. Although colorimetric and spectrophotometric assessments of β-galactosidases were estimated to have the same sensitivity, the last method is characterized by a higher specificity. The results suggest that the use of the UV-vis spectrophotometer increases the specificity of the evaluation of salivary β-galactosidases.
Export citation and abstract BibTeX RIS
Introduction
Halitosis is a common problem affecting approximately 30% of the whole population. 85% of bad breath causes originate in the oral cavity (Outhouse et al 2006). The odor intensity of breath has been correlated with the concentration of volatile sulfur compounds (VSCs) consisting of hydrogen sulfide, methyl mercaptan, dimethyl sulfide and dimethyl disulfide (Tonzetich 1977).
VSCs are the result of a complex interaction between gram-positive and gram-negative bacteria which degrade sulfur-containing amino acids of salivary glycoproteins (Delanghe et al 1997, Rosenberg 1996). Gram-positive bacteria produce enzymes responsible for the initial degradation of salivary glycoproteins by removing the carbohydrate side chains with O- and N-bonds. Thus, split-proteins become much more easily digestible by proteases produced by gram-negative bacteria (Haraldsson and Holbrook 1999). One of these enzymes is the β-galactosidase (De Jong and Van Der Hoeven 1987).
The main producers of these enzymes are oral streptococci, especially Streptococcus salivarius but other oral pathogens such as Porphyromonas gingivalis, Prevotella intermedia and Prevotella nigrescens also have the ability to produce β-galactosidases (Haraszthy et al 2008, Homer et al 1994, Xu et al 2010). Sterer assessed that whereas P. gingivalis alone produced substantial levels of odor and volatile sulfides, pre-incubation in the presence of β-galactosidases resulted in a significant increase in volatile sulfides, as compared to mucin incubation in the presence of P. gingivalis alone (Sterer and Rosenberg 2006). This process is inhibited by the addition of d-glucose, a typical non-competitive inhibitor, as well as the substrate analog inhibitor, p-aminophenyl-β-d-thiogalactopyranoside (p-APTG) (Wilson et al 2008, Won et al 2001).
Many studies indicate that the β-galactosidase activity plays an important role in malodor production, but its detection is based on a color intensity scale and results are recorded by the judges (Xu et al 2010, Wilson et al 2008). This activity assay is based on the intensity of the blue product, 5,5'-dibromo-4,4'-dichloro-indigo, produced by separation of the substrate bromo-chloro-indolyl-galactopyranoside (x-gal) by salivary β-galactosidases (Rosenberg et al 1991).
This procedure has poor reproducibility as it relies on judges' subjective evaluation and the precision of the results are limited because the scale contains only three values from 0 to 2 (Sterer et al 2002).
The development of a simple and objective technique for the β-galactosidase activity assay is necessary, both for research and clinical assessment. This communication proposes a spectrophotometric assay to evaluate salivary β-galactosidase activity for the assessment of oral malodor.
Materials and methods
The study population consisted of 94 healthy adult volunteers, 68 males and 26 females (average age ± standard deviation: 29.94 ± 13.83) who were recruited from the Department of Oral Sciences, Nano and Biotechnology of G. d'Annunzio University between 2009 and 2011. The study was conducted in accordance with the principles embodied in the Declaration of Helsinki and the protocol was reviewed and approved by the ethics committee of University G. D'Annunzio of Chieti. The inclusion criteria were: adult males or females from 20 to 65 years of age.
Each participant completed a medical and dental history questionnaire; those who received antibiotic treatments within 1 month prior to the study or showing the evidence of diseases concerning respiratory or gastrointestinal tract, diabetes, liver or kidney problems that may influence breath odor were excluded from this study. Each participant followed a protocol which included abstaining from certain food and drugs at least 72 h before testing and procedures of oral hygiene during the 3 h earlier (table 1).
Table 1. The protocol followed by each participant of the study.
| Always | Abstain from drugs: |
| Antihypertensive | |
| Antidepressants | |
| Antimuscarinic | |
| Antidiarrheals | |
| Antiparkinsonian | |
| 1 month before | Abstain from: |
| Antibiotics | |
| 72 h before testing | Abstain from food or drugs: |
| Garlic | |
| Onion | |
| Spices | |
| Fried | |
| Alcohol | |
| 12 h before testing | No smoking (for smokers) |
| Do not use perfumes and alcoholic | |
| mouthwashes | |
| 4 h before testing | Do not drink beverages other than water |
| 2 h before testing | Do not use toothbrush or floss |
Collection of unstimulated whole saliva
Unstimulated whole saliva was collected using the spitting method. Samples were collected from the participants between 10:00 a.m. and 11:00 a.m. to minimize the effects of diurnal variability in salivary composition.
Organoleptic measurements (odor judge scores)
After recording the subject's medical history, each subject received an oral health check-up and the organoleptic measurement. The purpose of the oral check-up was to exclude oral factors which could improve oral malodor, such as incongruous prosthesis or dental appliance. Before taking organoleptic measurements, the two blinded judges completed the Nachnani's training protocol: (i) introduction to sensory scales, n-butanol reference, sniffing techniques; (ii) pretraining measurements; 20 samples of varying intensities of four unpleasant and three pleasant odorants; (iii) exercises assessing quality, intensity, ranking and matching; and (iv) post–training measurements (Nachnani et al 2005). For judge's scoring of whole-mouth malodor, subjects were instructed to exhale for 5 s through the mouth, at a distance of approximately 10 cm from the nose of the judges.
Oral malodor was recorded as follows: 0 = absence of odor; 1 = questionable malodor; 2 = slight; 3 = moderate; 4 = strong; and 5 = severe (Rosenberg et al 1991).
For Spearman's correlation and receiver operating characteristic (ROC) analyses, mean judges' scores were dichotomized into two groups: 'malodor −' (scores < 2) and 'malodor +' (scores ≥ 2).
Colorimetric evaluation of β-galactosidases (Cβ-g)
Each saliva sample (20 µl) was immediately applied to a paper disk, prepared according to the procedure described by Sterer et al (2002). After 10 min of incubation at room temperature, β-galactosidases activity was estimated by evaluating the variation of the color that appeared on the diskettes. Results were recorded by a judge as follows: 0—no color, 1—light blue color, 2—dark blue color. For the later Spearman correlation and ROC analyses, β-galactosidases scores were dichotomized into two groups: 'Cβ-g −' no detectable colorimetric reaction (scores 0) and 'Cβ-g + detectable colorimetric reaction (cut-off ≥ 1).
Spectrophotometric assessment of β-galactosidase activity assay (Sβ-g)
After colorimetric analysis, paper disks were incubated (15 min) in 400 µl of N,N-dimethylformamide (DMF, Sigma-Aldrich) (Woo 2000). After the extraction of 5,5'-dibromo-4,4'-dichloro-indigo, the disks were removed and the obtained mixture was centrifuged for 10 min at 12 000 rpm at 37 °C (MIKRO 22R, Hettich Zentrifugen, Germany). The obtained supernatant was separated from the pellet and then analyzed using a diode array spectrophotometer (HP/Agilent 8453 UV-vis diode spectrophotometer G1103A), connected to a computer with UV-vis license, Chemstation 8. Wavelengths of maximum absorbance (λmax) between 560 nm and 627 nm have been considered because spectral data of indigo derivatives have a broad absorption λmax comprehended between 600 and 650 nm (Min et al 2009, Travasso et al 2003).
Each sample was analyzed three times and the average value was considered.
The calculated delta, i.e. the difference between wavelengths of maximum absorbance (λmax) at 627 nm and 560 nm, was recorded. For Spearman's correlation and ROC analyses, spectrophotometric scores were dichotomized into two groups: 'Sβ-g −' (delta < 0.05) and 'Sβ-g +' (delta ≥ 0.05).
Statistical analysis
MedCalc Software, version 11.6.0 (Broekstraat 52, 9030 Mariakerke, Belgium), was used for diagnostic test evaluation. ROC curves and complete sensitivity/specificity reports were calculated. The DeLong method for the calculation of the standard error of the area under the curve was considered (DeLong et al 1988).
Data were also analyzed with ANOVA and post hoc Fisher's PLSD, Scheffe and Bonferroni/Dunn methods. The two-tailed Spearman rank order correlation coefficient was used to determine the correlations between variables. Cohen's kappa statistical approach was used to determine the level of agreement between the judges' scores (Randolph 2008).
All tests were conducted at the 5% significance level. The Stat View 4.0 statistical software package was used for Spearman's correlation, Cohen's kappa statistics, Anova and post hoc tests (Abacus Concepts, Berkeley, USA).
Results
To verify the use of spectrophotometric assessment of β-galactosidase activity, a comparison between different methods for estimating oral malodor has been conducted. Based on mean judge scores, 35 subjects (37.2%) suffered from oral malodor (at a cut-off ≥ 2).
The level of agreement between the examiners, Cohen's kappa, was 0.596 and the Spearman correlation was 0.968. The performance of the spectrophotometric assay has been assessed by calculating the sensitivity, specificity and ROC curves (figure 1). The area under the ROC curve (AUC) measures discrimination, that is, the ability of the test to correctly classify those with and without the disease. The area relating to Sβ-g (0.807) is higher than Cβ-g (0.790).
Figure 1. ROC curves: (A) Sβ s (at a cutoff delta ≥ 0.05) and (B) Cβ s (at a cutoff ≥ 1). The ROC curve is a plot of the true positive rate against the false positive rate based on judge scores (at a cut-off ≥ 2). Both Cβ-g and Sβ-g were estimated to have a sensibility of the 80% (with 95% CI); specificity tests were of 77.97% (with 95% CI) and 81.36% (with 95% CI) respectively. Accuracy, which is measured by the area under the ROC curve (AUC), is higher in Sβ-g than in Cβ-g.
Download figure:
Standard imageBoth Cβ-g and Sβ-g were estimated to have a sensitivity of 80% (with 95% confidence interval); specificity tests were 77.97% (with 95% CI) and 81.36% (with 95% CI), respectively.
The mean odor judge scores (± standard deviation) were 1.223 ± 1.109. The mean Cβ-g and Sβ-g scores were 0.585 ± 0.739 and 0.012 ± 0.025, respectively.
The two-way repeated-measure ANOVA showed that the results between spectrophotometric and colorimetric analyses are statistically significant (figures 2 and 3). Figures 2 and 3 show the mean scores for colorimetric and spectrophotometric analyses based on odor judge scores. Results obtained with Sβ-g are arranged in order of increasing value; in contrast, this does not occur in figure 2. The mean value of Cβ-g in patients who received 3 at odor judges' scores was higher (1.40) than subjects who were scored with 4 (1.33).
Figure 2. Mean values of colorimetric measurements organized based on odor judge scores (significance level < 0.05).
Download figure:
Standard image
Figure 3. Mean values of spectrophotometric measurements organized based on odor judge scores (significance level < 0.05).
Download figure:
Standard imageAll analyses were compared with each other. The strength of association among the various parameters was assessed by Spearman's correlations (table 2).
Table 2. The two-tailed Spearman rank order correlation between judge (Judge s), colorimetry (Cβ s) and spectrophotometry (Sβ s) scores. All parameters are highly associated indeed Sβ s are more correlated to Judge scores with respect to Cβ s.
| Judge s | Cβ s | Sβ s | |
|---|---|---|---|
| Judge s | – | r = 0.73* | r = 0.78* |
| p < 0.0001 | p < 0.0001 | ||
| Cβ s | – | – | r = 0.90* |
| p < 000.1 |
As expected, Cβ-g and Sβ-g were closely correlated (Spearman, r = 0.90, p < 0.0001). Sβ-g and judge scores were also compared (Spearman, r = 0.78, p < 0.0001). A significant association was also found between Cβ-g and judge scores (Spearman, r = 0.73, p < 0.0001).
Discussion
Organoleptic scores and Cβ-g were significantly associated with one another, although the correlations were slightly higher than in previous reports. Rosenberg found a significant association between odor judge scores and Cβ-g (Spearman, r = 0.59; p < 0.01). 89% of the people who was positive to organoleptics were also positive for Cβ-g (sensitivity). Specificity and accuracy were 75% and 79%, respectively (Rosenberg et al 2007).
Yoneda has also found a positive correlation between Cβ-g and malodor strength (organoleptic score, portable sulfide monitor score, VSC concentrations and tongue coating) (Yoneda et al 2010).
These results indicate that β-galactosidase activity plays an important role in malodor production, but all these evaluations were performed using properly trained and calibrated odor judges.
Indeed, different operators could evaluate results obtained by the colorimetric test in a different way, so colorimetry is a poor method of analysis for clinical use.
Figure 2 shows a diagnostic error which could occur with the colorimetry; in fact, subjects with an odor judge score of 4 are characterized by a lower colorimetric score than those who received 3. This phenomenon did not occur with the spectrophotometric analysis (figure 3). Unlike human evaluation, the spectrophotometer does not rely on judgment or environmental conditions to evaluate color variations. The spectrophotometric analysis is not influenced by human variables such as eye fatigue, age, experience and other physiological factors such as color blindness and the number of rods and cones contained in the eye. In addition, the colorimetric evaluation is highly subjective and time-consuming (Horn et al 1998).
Although Cβ-g and Sβ-g have been estimated to have the same sensitivity, the last test is characterized by a higher specificity (figure 1). If the AUC concerning Sβ-g is higher than Cβ-g, then spectrophotometric evaluation presents a higher percentage of correctly classifying two patients in a random pair.
Spectrophotometry scores present a higher correlation to organoleptic evaluations than colorimetric methods (table 1).
This means that dentists who are not an expert in colorimetry can request spectrophotometric analysis of β-galactosidases by sending a salivary sample to a scientific laboratory. This could be a better aid for halitosis diagnosis after the organoleptic evaluation.
Conclusions
Results suggest that the spectrophotometry increases performance of the β-galactosidase activity assay. In particular, the advantages of this method with respect to colorimetry are
- higher correlation with organoleptic scores,
- higher specificity,
- results being in a numerical form and objective, and
- no need of a trained judge for the evaluation of the colorimetric reaction.
Acknowledgments
The authors would like to thank Dr Claudia Di Giandomenico, Hun Ting Diu and Aurora Berardinelli.