0003-9%9/92 S5.00 + 0.00 Copyright 0 1992 Pcrgamon Press Ltd
Archs oral Biol. Vol. 31, No. I, pp. 559-W. 1992
Printedin Great Britain. All rights reserved
CONTAMINATION OF HUMAN GINGIVAL CREVICULAR FLUID BY PLAQUE AND SALIVA G. S. GRIFFITHS,* J. M. A. WILTON and M. A. Cuuns MRC Dental Research Unit, London Hospital Medical College, Turner Street, London El 2AD, U.K. (Accepted 22 February 1992)
Summary-Dental plaque and saliva are both possible contaminants of gingival crevicular fluid (GCF). Plaque samples from 12 sites in four subjects who had allowed plaque to accumulate for 2448 h were quantified using the Plaque Index and then transferred to tilter paper strips for fluid volume determination using the Periotron 6000. The mean volume of fluid for plaque scores of 0 was 0.02 (kO.01) ~1, whereas for nlaaue scores of 3 the mean volae was 0.15 (+ 0.07) ~1. In a clinical study, GCF samples, from 19/184 sub&b (10.496) were assessed as contaminate&r suspected to be contaminated wig saliva, but only 28/1740 strips (1.6%) were placed in these categories of contamination. An assay to con&m salivary contamination was established, using an immunochemical, double-antibody method with sheep antihuman salivary a-amylase followed by peroxidase-conjugated rabbit anti-sheep immunoglobulin. The detection threshold of the assay was 7.5 ng of a-amylase, which represents approx. 15-25 nl of saliva. The assay was evaluated on 90 GCF samples categorized as ‘known’ (n = 16), ‘suspected’ (n = 16), or ‘not’ contaminated (n = 58) with saliva; 81.25, 5ti1.5 and 5.228.6%, respectively, were positive for salivary a-amylase. In some GCF samples salivary contamination was in excess of 50%. It was concluded that GCF samples seen to be contaminated with saliva should be. discarded, whereas samples considered free from contamination may be used with confidence. Samples suspected of contamination may require an a-amylase assay before further analysis. Key words: a-amylase, plaque, saliva, gin&al crevicular fluid.
INTRODUCI’ION
Collection of GCF may provide information on fluid volume or flow rate in clinical assessments of gingival inflammation. Some regard the volume of GCF as a more accurate and critical means of assessing the state of the gingiva than traditional clinical measurements (L6e and Holm-Pedersen, 1965; Miihlemann and Son, 1971; Holm-Pedersen, Agerbaek and Theilade, 1975). Samples of GCF may then be used in laboratory investigation to analyse a wide variety of components (for reviews see Cimasoni, 1983; Curtis et al., 1989). The possible contaminating effects of dental plaque or saliva on such estimations has seldom been considered. Contamination with saliva would increase both the estimated volume of GCF and the flow rate. As microbial plaque contains 82% w/w of water, approx. 50% intracellular and 32% in the matrix (Jenkins, 1978), this contaminant could also increase the measured volume and flow of GCF (Stoller, Karras and Johnson, 1990). Precautions taken against these problems include, for example, drying the teeth and surrounding tissues by air (L6e and Holm-Pedersen, 1965) or, alternatively, using guard strips inserted into the gingival crevice (Mann, 1963).
*Present address: Department
of Periodontology, Eastman Dental Hospital and Institute of Dental Surgery, 256 Grav’s Inn Road. London WClX 8LD, U.K. Abbrevikms: GCF,’ gingival crevicular fluid; PBS, phos-
phate-buffered saline.
Despite any such precautions, little attention has been paid to the possible role of plaque fluid and saliva in GCF as normal constituents, or from mixing of fluids in the crevice, or as contaminants interfering with the collection of native GCF. Oliver, HolmPedersen and Liie (1969) attributed the detection of trace volumes of GCF at sites with a Gingival Index of 0 to contamination with plaque and saliva. In a similar study, Rudin, Overdiek and Rateitschak (1970) discussed the possibility of contamination of GCF with saliva or plaque, but only as an explanation for the ‘doubtful origin’ of samples collected from healthy marginal gingiva. Stoller et al. (1990) have now presented data which suggest that plaque may significantly increase measurements of GCF made with the Periotron 6000. Studies with fluorescein have provided some information on the origin of GCF and the role of saliva as a possible contaminant (Weinstein et al., 1967; Lije and Holm-Pedersen, 1965). However, these only demonstrate that fluorescein, taken systemically, passes from the circulation to the GCF, whereas fluorescein taken orally cannot penetrate the gingival crevice but may be incorporated into plaque. Immunochemical methods have been used as a more sensitive way of detecting salivary components in GCF (Weinstein et al., 1967; Douglas, Craig and Neil, 1986). Our aims now were to determine whether plaque contamination of filter paper strips influences the volumetric determination of GCF by the Periotron 6000, and to establish whether salivary components were present in GCF samples collected from dried and isolated gingival crevices, as assessed by the presence of salivary a-amylase.
559
G. S. GR~FF~THS et al.
560
Table 1. Number and percentage of R.A.F. samples visually assessed as contaminated or suspected of contamination with saliva Suspected Contaminated contaminated Subjects gt; 184) $?f68) (fl = 1740)
Totals
(3&D)
(7.k)
(10.:9x)
(l&)
(3.k)
(5.k)
(O.k.,
(Lk)
(1.Z)
MATERIALS AND METHODS
Plaque mass and GCF volume determination Four subjects were asked to refrain from their normal oral hygiene practices for between 24 and 48 h in order to allow plaque to accumulate. Supragingival plaque was removed from 12 sites in each subject with a periodontal probe. The amount of plaque removed was quantified according to criteria for scoring the presence of supragingival plaque (Silness and Lee, 1964). The probe was then scraped across the end of a 2 x 8 mm Whatman 3MM Chromatography paper strip (Whatman Lab. Sales Ltd, Maidstone, Kent, U.K.) to transfer the plaque deposits to the paper. The strip was then placed in a Periotron 6000 (Harco Electronics, Winnipeg, Canada) and the fluid volume determined by reference to a standard curve, as described by Griffiths, Curtis and Wilton (1988). Detection of a-amylase in samples of human GCF Clinical method. The samples of GCF to be investigated for a-amylase were from three sources. Samples collected from Royal Air Force (R.A.F.) recruits, who were participants in a longitudinal study, were selected for use in the assay; their GCF was collected at two sites on filter paper. Each site was isolated with cotton-wool rolls and a flanged saliva ejector; the surface of the tooth was dried with a fine-bore aspirator. Samples were taken for 5 s, at 1 min intervals for 3 min (strips Pl-P4), followed by a rest period of 6 min after which a final strip (P5) was collected. We had observed that some strips became visibly contaminated with saliva during collection. Such samples were here designated as ‘known contamination’. Some strips also gave readings of
unusually high volume compared to earlier samples from the same site. We suspected this was due to salivary contamination, but there were no signs of failure in isolating the site (such as wet cotton rolls), and no visual evidence of saliva. These samples were designated ‘suspected contamination’. All other samples were considered to be ‘non-contaminated’; however, some strips were from sites where an earlier strip was known, or suspected, to be contaminated and some carry over of contamination between strips was possible. The number and percentage of subjects, sites and strips that came into each category are shown in Table 1. The n value of 1740, rather than 1840, for strips arose because we collected only strips Pl-P4 from the first 50 subjects, and then added the PS sample to the protocol in an effort to increase the overall volume of GCF collected. Samples visibly contaminated with blood were also excluded, as were all of the strips taken from that site. Samples from 10 R.A.F. subjects provided most of those that were considered either known (n = 1l), or suspected (n = 8) to be contaminated with saliva. These were selected for use in the assay for the detection of salivary a-amylase. Other samples were collected from patients attending the Periodontal Department, London Hospital Dental School, and also from laboratory staff. Some of these samples were deliberately contaminated with saliva by relaxing the normal procedure for drying and isolating the teeth. Otherwise samples from these subjects were categorized as described above. This resulted in 90 samples for assay; the number of strips in each category is shown in Table 3: known contaminated (n = 16), suspected contaminated (n = 16) and noncontaminated (n = 58). Samples were recovered from the strips using the centrifugal elution procedure described by Griffiths et al. (1988), but the elution volume was increased from 20 to 25 ~1 of distilled water. Assay method. The 90 samples of GCF were analysed immunochemically in the Bio-Dot microfiltration apparatus (Biorad Laboratories Ltd, Watford, Hertford, U.K.). A dilution series of saliva from the same subject was also analysed. Each experiment also included a dilution series of normal human serum, so as to confnm the absence of cross-reactivity in the developing antisera. Wells containing only PBS (pH 7.2) were included as a negative control to
Table 2. Clinical and laboratory data for the subject illustrated in Fig. 2
strip number 2; : s5
Site 16mp 16mp 43db 25 mb 25mb
Saliva present
Periotron reading
Volume of GCF sample (nl)
+ + + + -
131 54 03 47 06
540 216 24 188 34
+
Detected volume of saliva* (nl)
Contamination (%)
160 25 32 -
74 29 42.5 -
Percentage contamination of saliva was derived from the volume of GCF obtained from Periotron readings and the detected volume of saliva, which was derived from the a -amylase assay. mp, Mesial palatal; db, distal buccal; mb, me&l buccal. *Volume of saliva detected in 20 ~1 of 50 ~1 eluant. + + , Known contamination; + , suspectedcontamination; -, no contamination.
Contamination
of crevicular fluid
561
Table 3. Numbers and percentages of GCF samples co&med to be contaminated with human saliva categorized by clinical assessment of salivary contamination Reason for non-detection GCF Samnle Status Known contaminated Suspected contaminated Noncontaminated
n
Number with saliva detected
Saliva detected (%)
16
13
81.25
16
8
50
3
5
58
3
5.2
23
32
establish the level of background staining. A dilution series of a-amylase (Sigma, Poole, Dorset, U.K.) was always included, both as an antigen-positive control
and to determine the sensitivity of the assay. All dilutions were made with PBS filtered through a 0.25 pm filter, as earlier work had shown falsepositive results in the assay due to non-specific binding of particulate matter. Twenty ~1 of GCF eluate were diluted 1: 5 in PBS, as the larger volume facilitated application in the dot-blot apparatus. A 12 x 8 cm piece of nitrocellulose membrane (Schleicher & Schull, Dassell, Germany), soaked overnight in filtered PBS, was placed in the Bio-dot apparatus. Samples were applied to the membrane and allowed to drain through for 1 h under gravity, with any residuum drawn through by application of a vacuum. The nitrocellulose membrane was then removed from the apparatus and blocked in 5% bovine serum albumin/PBS for 30 min. The membrane was then washed in PBS/O. 1% Tween, with repeated changes of solution, for 1 h. Salivary a-amylase was detected in a doubleantibody technique using sheep anti-human salivary a-amylase, 1:250, for 2 h (Serotec, Bicester, Oxon, U.K.) followed by peroxidase-conjugated rabbit antisheep immunoglobulins, 1: 100, for 2 h (Dakopatts, Glostrup, Denmark) and colour development with 4-chloronaphthol [4CN (Sigma, Poole, Dorset, U.K.)]. Preliminary experiments had shown a high background reactivity of these antisera with human serum. This was reduced to acceptable levels by batch absorption of the anti-human salivary a-amylase with 50% human serum for 30 min and passage of the rabbit anti-sheep immunoglobulins down a column of human gammaglobulin (Sigma, Poole, Dorset, U.K.) coupled to A&Gel 10 (Biorad Laboratories Ltd, Watford, Hertford, U.K.). Densitometric analysis of dot-blots. Photographic colour prints of the dot-blots were analysed with an Ultroscan Laser Densitometer using the Gelscan programme (LKB Instruments, Milton Keynes, Bucks., U.K.). Standard curves were produced for each subject using the saliva dilution series. The lower limit of detection for each subject’s standard curve was determined by the ability to discriminate between the peak heights of higher dilutions and the background levels obtained for both PBS and serum. The peak heights were calculated for each of the GCF samples and by reference to the standard curve; an evaluation of the presence or absence of salivary a-amylase was then made within the limits of
Mow detection level
Volume too low -
3
detection. When salivary a-amylase was shown to be present, an estimation of the volume of salivary contamination of the 20 ~1 sample of GCF was made by reference to the standard curve. This was described as the ‘detected volume of saliva’ and was compared with the ‘volume of GCF sample’, which was derived from measurement in the Periotron, followed by elution into 50~1 of de-ion&d water. This allowed calculations of the percentage contamination to be made, using the formula: detected volume of saliva volume of GCF
50 x5x
100.
REsuLTs
Figure 1 shows the mean and SDS of the volume determinations in the Periotron 6000 of plaque samples applied to 2 x 8 mm chromatography paper strips. As expected, there was a consistent increase in the volume detected with increasing plaque mass applied to the strips. The mean volume when plaque was coded as zero was 0.02 (+ 0.01) p 1,whereas when plaque was coded as 3 the mean volume was 0.15 (kO.07) pl. Table 1 shows data on the visual assessment of GCF contamination of strips from the longitudinal study of R.A.F. recruits. We considered that 3.3% of the recruits had samples contaminated with saliva and that 7.1% were suspect for contamination; this
a25 r MO-
s
0.15 -
E z
0.10 -
a05 -
0
1 Plaque
2
n=21 I 3
Index
Fig. 1. The effect of plaque mass on the volume determination of GCF using the Periotron 6000.
G. S. Gamma
--r--Calibration ~~~.m.+.Thrsrhold
I 20
0
I 40
Volume
I
I
I
I
60
60
100
120
of saliva
(nl)
Fig. 2. Standard curve of salivary volume estimation using an immunochemical assay of cc-amylase. represents 10.4% of the population studied. As two sites per subject were sampled the site rather than the subject data may have produced different figures for contamination. However, as no subject was deemed to have both sites contaminated or suspected, the absolute figures were the same as for the subject data but the percentage values were decreased. When data were further broken down to the level of individual strips, the n value rose to 1740 and the increase in absolute values of contaminated or suspect contaminated strips was due to several strips being contaminated at an individual site. Figure 2 shows the peak heights estimated by densitometry of dots in a dilution series of saliva from a representative subject. These were plotted against the amount of saliva added to each well and a ‘best fit’ equation calculated and plotted. Calculations of the a-amylase concentration of saliva samples were made from comparisons with a similar dilution series of a-amylase (Sigma, Poole, Dorset, U.K.). The sample containing 10nl of saliva, which had a peak height of 0.046, was distinguishable from the background absorbance values obtained with both PBS and blank wells, which represented the absorbance value of the nitrocellulose membrane alone. Extrapolation of this to the line of ‘best fit’ indicates that the threshold for detection in this subject was approx. 15 nl. Figure 3 shows the peak heights for all the GCF samples (Sl-SS) in this subject. Samples S3 and SS gave values less than 0.046 and were therefore below the threshold of detection. Samples Sl, S2 and S4 were all above this point and reference to the equation for
-
z .o
Detection
threshold
0.10
r g
0.05
k! 0
si
s2 GCF
s3 sample
S4
SS
number
Fig. 3. Concentration of a-amylase in GCF samples as assessedby laser densitometry. GCF samples were collected from the subject shown in Fig. 2.
et al.
the line of best fit, derived from Fig. 2, allowed the contribution of saliva in the sample to be calculated. These calculations and the relevant clinical data are presented in Table 2. Both Sl and S2, which were expected to contain saliva showed detectable levels that represented 74 and 29%, respectively, of the total GCF sample volume. The other sample showing a-amylase, S4, was suspected of being contaminated and was found to have 42.5% salivary contamination. Neither S3 nor S5 showed detectable a -amylase as expected. Of the 90 samples tested (Table 3), 81.25% of the sites known to be contaminated with saliva had detectable levels of cr-amylase. The remaining 18.75%, although clinically designated as containing saliva, were below the threshold of detection in this immunoassay. Of the samples suspected of salivary contamination, 50% had detectable levels of a -amylase, although 3/l 6 samples (19%) had too low a volume of sample to achieve detection of a-amylase even if it had consisted only of saliva. Of the 58 samples designated clinically as uncontaminated with saliva only, 3/58 (5.2%) were found to have detectable levels of a-amylase. However, at 23/58 sites (40%) too low a volume of GCF had been collected to enable detection of a-amylase.
DISCUSSION
We show a potential source of inaccuracy from the effect of plaque when using the Periotron 6000 to measure the volume of GCF. A small plaque mass gave an insignificant fluid volume, but a more substantial plaque mass clearly gave larger volumes of fluid that would make an important contribution to the estimated volume of a GCF sample. This is in agreement with observations by Stoller et al. (1990), who found that the presence of supragingival plaque caused a significant increase in measurements of GCF in the Periotron 6000. Some of our R.A.F. subjects had sampling sites that scored 2 or 3 on the plaque index (Silness and Lee, 1964). Whilst a careful technique that removes plaque before sampling will not involve contamination on this scale, it is important that the operator should be aware of the possible contaminating effects of smaller amounts of plaque on the volume determination and measures should be taken to minimize this. Before the work of Stoller et al. (1990) and ourselves, there had been only scant reference made to the possibility of plaque interfering with the collection of GCF. However, Oliver et al. (1969) and Rudin et al. (1970) did attribute their detection of GCF at healthy sites to the contribution of plaque contamination, and Sidi and Ashley (1984) listed saliva, blood and plaque fluid as possible contaminants. Table 1 gives an indication of the likely influence of saliva on estimations of GCF volume in a clinical investigation. Despite careful isolation, which included cotton-wool rolls and a flanged saliva ejector, 10% of our subjects were categorized as having either contamination or suspected contamination with saliva. Per strip the figure is mom acceptable, as only 1.6% of strips were either known or suspected to be contaminated with saliva.
Contamination of crevicular fluid Immunological methods have been used as a more sensitive way of detecting salivary components in GCF. By immunoelectrophoresis, Weinstein et al. (1967) showed no precipitation with samples of GCF when antisera to parotid or submaxillary saliva were used, indicating the absence of salivary proteins in GCF. This method is not, however, very sensitive and could not exclude minor degrees of contamination of GCF with saliva. Douglas er al. (1986) used Western blots of GCF probed with anti-human saliva antiserum. Unfortunately, they did not present the results of these experiments, but commented only that some constituents of GCF were recognized by the anti-salivary antiserum and that the molecular weights of’ the bands detected were totally different from those of mixed saliva. Our immunochemical method could detect 7.5 ng of a-amylase (data not shown), which, based on a mean value of 0.38 mg/ml of cr-amylase in saliva (Jenkins, 1978) represents approx. 20 nl of saliva. The actual threshold of detection within the assay will vary from individual to individual dependent upon the concentration of a-amylase in the saliva. Thus the inclusion of a dilution series of each subject’s saliva in all the assays allowed for easier assessment of the presence of saliva within the GCF samples. In the example presented (Fig. 2) the limit of detection was calculated as 15 nl of saliva. Table 3 shows th.at the detection of a-amylase confirmed the clinica. impression of the presence or absence of saliva. Where saliva was deemed present visually, a-amylase could be detected in the majority of cases. However, in some instances the volume of GCF collected was small and consequently would require over 50% of the sample to be saliva before it reached the threshold of detection in the a-amylase assay. In samples where there was no reason to suspect salivary contamination, a-amylase was only detected in only 5.2% of strips. This value was artificially low because 23 samples had a volume of GCF below the threshold of detection for a -amylase. If these samples are excluded the level of unsuspected contamination is 3/35 (8.6%). However, this may be an overestimation, as this group of samples was biased by the fact that strips categorized as uncontaminated were often from sites where one of the other strips had been categorized as contaminated or suspect contaminated. Nevertheless these figures were still well below the 50% detection of a-amylase found in sites suspected of being contaminated with saliva and the same arguments applied in this category, with three sites yielding a GCF volume below the threshold of detection of a -amylase. The problem of small volumes of GCF collected and the variation between subjects in the concentrations of a-amylase, and therefore the threshold of detection of the a-amylase assay, illustrate the shortcomings of this assay for routine use. If the assay were to be used on all samples in order either to ensure the absence of salivary contamination, or to be able to state, for example, that all samples contained < 10% saliva, it would be necessary to either use the whole sample volume or be restricted to those samples that initially had yielded a large enough volume. In this study, the use of the whole sample AOB 37/7-D
563
volume was not acceptable, as other investigations of the composition of GCF were being made. Restricting the assay to samples with a large enough volume may allow us to confirm or refute whether the unusually large volume collected at a site was either a reflection of GCF volume, or an indication of salivary contamination. This was one of the indications for samples to be designated as ‘suspected contaminated’. We have shown that if GCF samples are collected using a good technique for isolation and drying, then the likelihood that saliva will be a significant contaminant is small. When sites are contaminated with saliva the detection of a -amylase confirms the clinical impression. Where sites were suspected of contamination the assay was able to confirm or refute a significant contribution from such contamination. Table 3 shows that 5/16 samples (31.25%) suspected of salivary contamination would have been confirmed as free of saliva within the detection limits of our assay. Thus if GCF samples are seen to be contaminated with saliva, they should be discarded. If samples are considered free of contamination they may be used with confidence. However, samples suspected of salivary contamination should be treated with caution and may require an a-amylase assay before further analysis. are grateful to the R.A.F. Dental Branch for their cooperation in this study and to the statI of the Institute of Dental Health and Training, R.A.F. Halton, for their unstinting support. This work was supported by the Medical Research Council.
Acknowledgements-We
REFERENCES
Cimasoni G. (1983) Crevicular fluid undated. Monographs in Oral Science, Vol. 12. Karger, Basel. Curtis M. A., Gillett I. R., Grilhths G. S., Maiden M. F. J., Steme J. A. C.. Wilson D. T.. Wilton J. M. A. and Johnson N. W. (1989) Detection of high-risk groups and individuals for periodontal diseases: laboratory markers from analysis of gingival crevicular fluid. J. clin. Periodont. 16, 1-l 1. Douglas C. W. I., Craig G. T. 0. and Neil T. C. A. (1986) A novel antigen in fluid collected from established periodontal pockets. The Borderland between Caries and Periodontal Disease III (Eds Lehner T. and Cimasoni G.) Editions Medecine et Hygiene, Geneve. Griffiths G. S., Curtis M. A. and Wilton J. M. A. (1988) Selection of a hlter paper with optimum properties for the collection of gingival crevicular fluid. J. periodont. Res. 23, 33-38. Holm-Pedersen P., Agerbaek N. and Theilade E. (1975) Experimental gingivitis in young and elderly individuals. J. clin. Periodont. 2, 14-24. Jenkins G. N. (1978) The Physiology and Biochemistry of the Mouth, 4th edn, Chap. IX, p. 286 and Chap. X, p. 371. Blackwell Scientific Publications, Oxford. Liie H. and Holm-Pedersen P. (1965) Absence and presence of fluid from normal and inflamed gingivae. Periodontics 3, 171-177. Mann W. V. (1963) The correlation of gingivitis pocket depth and exudate from the gingival crevice. J. Periodont. 34, 379-387. Miihlemann H. R. and Son S. (1971) Gingival sulcus bleeding-a leading symptom in initial gingivitis. Jfefu. odont. Acta. 15, 107-113.
564
G. S. GRIFFITHS er al.
Oliver R. C., Hahn-Pedersen P. and L6e H. (1969) The correlation between clinical scoring, exudate measurements and microscopic evaluation of inflammation in the gingiva. J. Periodont. 40, 201-209. Rudin H. J., Overdiek H. F. and Rateitschak K. H. (1970) Correlation between sulcus fluid rate and clinical and histological inflammation of the marginal gingiva. Helu. odont. Acta. 14, 21-26. Sidi A. D. and Ashley F. P. (1984) Influence of frequent sugar intakes on experimental gingivitis. J. Periodont. 55, 419423.
Silness J. and Tie H. (1964) Periodontal disease in pregnancy. II. Correlation between oral hygiene and periodontal condition. Acta odont. stand. U, 121-135. Stoller N. H., Karras D. C. and Johnson L. R. (1990) Reliability of crevicular fluid measurements taken in the presence of supragingival plaque. J. Periodont. 61, 67&673. Weinstein E., Mandel I., Salkind A., Oshrain H. I. and Pappas G. D. (1967) Studies of gingival fluid. Periodontics 5. 161-166.
Archs oral Biol. Vol. 31, No. I, pp. 559-W. 1992
Printedin Great Britain. All rights reserved
CONTAMINATION OF HUMAN GINGIVAL CREVICULAR FLUID BY PLAQUE AND SALIVA G. S. GRIFFITHS,* J. M. A. WILTON and M. A. Cuuns MRC Dental Research Unit, London Hospital Medical College, Turner Street, London El 2AD, U.K. (Accepted 22 February 1992)
Summary-Dental plaque and saliva are both possible contaminants of gingival crevicular fluid (GCF). Plaque samples from 12 sites in four subjects who had allowed plaque to accumulate for 2448 h were quantified using the Plaque Index and then transferred to tilter paper strips for fluid volume determination using the Periotron 6000. The mean volume of fluid for plaque scores of 0 was 0.02 (kO.01) ~1, whereas for nlaaue scores of 3 the mean volae was 0.15 (+ 0.07) ~1. In a clinical study, GCF samples, from 19/184 sub&b (10.496) were assessed as contaminate&r suspected to be contaminated wig saliva, but only 28/1740 strips (1.6%) were placed in these categories of contamination. An assay to con&m salivary contamination was established, using an immunochemical, double-antibody method with sheep antihuman salivary a-amylase followed by peroxidase-conjugated rabbit anti-sheep immunoglobulin. The detection threshold of the assay was 7.5 ng of a-amylase, which represents approx. 15-25 nl of saliva. The assay was evaluated on 90 GCF samples categorized as ‘known’ (n = 16), ‘suspected’ (n = 16), or ‘not’ contaminated (n = 58) with saliva; 81.25, 5ti1.5 and 5.228.6%, respectively, were positive for salivary a-amylase. In some GCF samples salivary contamination was in excess of 50%. It was concluded that GCF samples seen to be contaminated with saliva should be. discarded, whereas samples considered free from contamination may be used with confidence. Samples suspected of contamination may require an a-amylase assay before further analysis. Key words: a-amylase, plaque, saliva, gin&al crevicular fluid.
INTRODUCI’ION
Collection of GCF may provide information on fluid volume or flow rate in clinical assessments of gingival inflammation. Some regard the volume of GCF as a more accurate and critical means of assessing the state of the gingiva than traditional clinical measurements (L6e and Holm-Pedersen, 1965; Miihlemann and Son, 1971; Holm-Pedersen, Agerbaek and Theilade, 1975). Samples of GCF may then be used in laboratory investigation to analyse a wide variety of components (for reviews see Cimasoni, 1983; Curtis et al., 1989). The possible contaminating effects of dental plaque or saliva on such estimations has seldom been considered. Contamination with saliva would increase both the estimated volume of GCF and the flow rate. As microbial plaque contains 82% w/w of water, approx. 50% intracellular and 32% in the matrix (Jenkins, 1978), this contaminant could also increase the measured volume and flow of GCF (Stoller, Karras and Johnson, 1990). Precautions taken against these problems include, for example, drying the teeth and surrounding tissues by air (L6e and Holm-Pedersen, 1965) or, alternatively, using guard strips inserted into the gingival crevice (Mann, 1963).
*Present address: Department
of Periodontology, Eastman Dental Hospital and Institute of Dental Surgery, 256 Grav’s Inn Road. London WClX 8LD, U.K. Abbrevikms: GCF,’ gingival crevicular fluid; PBS, phos-
phate-buffered saline.
Despite any such precautions, little attention has been paid to the possible role of plaque fluid and saliva in GCF as normal constituents, or from mixing of fluids in the crevice, or as contaminants interfering with the collection of native GCF. Oliver, HolmPedersen and Liie (1969) attributed the detection of trace volumes of GCF at sites with a Gingival Index of 0 to contamination with plaque and saliva. In a similar study, Rudin, Overdiek and Rateitschak (1970) discussed the possibility of contamination of GCF with saliva or plaque, but only as an explanation for the ‘doubtful origin’ of samples collected from healthy marginal gingiva. Stoller et al. (1990) have now presented data which suggest that plaque may significantly increase measurements of GCF made with the Periotron 6000. Studies with fluorescein have provided some information on the origin of GCF and the role of saliva as a possible contaminant (Weinstein et al., 1967; Lije and Holm-Pedersen, 1965). However, these only demonstrate that fluorescein, taken systemically, passes from the circulation to the GCF, whereas fluorescein taken orally cannot penetrate the gingival crevice but may be incorporated into plaque. Immunochemical methods have been used as a more sensitive way of detecting salivary components in GCF (Weinstein et al., 1967; Douglas, Craig and Neil, 1986). Our aims now were to determine whether plaque contamination of filter paper strips influences the volumetric determination of GCF by the Periotron 6000, and to establish whether salivary components were present in GCF samples collected from dried and isolated gingival crevices, as assessed by the presence of salivary a-amylase.
559
G. S. GR~FF~THS et al.
560
Table 1. Number and percentage of R.A.F. samples visually assessed as contaminated or suspected of contamination with saliva Suspected Contaminated contaminated Subjects gt; 184) $?f68) (fl = 1740)
Totals
(3&D)
(7.k)
(10.:9x)
(l&)
(3.k)
(5.k)
(O.k.,
(Lk)
(1.Z)
MATERIALS AND METHODS
Plaque mass and GCF volume determination Four subjects were asked to refrain from their normal oral hygiene practices for between 24 and 48 h in order to allow plaque to accumulate. Supragingival plaque was removed from 12 sites in each subject with a periodontal probe. The amount of plaque removed was quantified according to criteria for scoring the presence of supragingival plaque (Silness and Lee, 1964). The probe was then scraped across the end of a 2 x 8 mm Whatman 3MM Chromatography paper strip (Whatman Lab. Sales Ltd, Maidstone, Kent, U.K.) to transfer the plaque deposits to the paper. The strip was then placed in a Periotron 6000 (Harco Electronics, Winnipeg, Canada) and the fluid volume determined by reference to a standard curve, as described by Griffiths, Curtis and Wilton (1988). Detection of a-amylase in samples of human GCF Clinical method. The samples of GCF to be investigated for a-amylase were from three sources. Samples collected from Royal Air Force (R.A.F.) recruits, who were participants in a longitudinal study, were selected for use in the assay; their GCF was collected at two sites on filter paper. Each site was isolated with cotton-wool rolls and a flanged saliva ejector; the surface of the tooth was dried with a fine-bore aspirator. Samples were taken for 5 s, at 1 min intervals for 3 min (strips Pl-P4), followed by a rest period of 6 min after which a final strip (P5) was collected. We had observed that some strips became visibly contaminated with saliva during collection. Such samples were here designated as ‘known contamination’. Some strips also gave readings of
unusually high volume compared to earlier samples from the same site. We suspected this was due to salivary contamination, but there were no signs of failure in isolating the site (such as wet cotton rolls), and no visual evidence of saliva. These samples were designated ‘suspected contamination’. All other samples were considered to be ‘non-contaminated’; however, some strips were from sites where an earlier strip was known, or suspected, to be contaminated and some carry over of contamination between strips was possible. The number and percentage of subjects, sites and strips that came into each category are shown in Table 1. The n value of 1740, rather than 1840, for strips arose because we collected only strips Pl-P4 from the first 50 subjects, and then added the PS sample to the protocol in an effort to increase the overall volume of GCF collected. Samples visibly contaminated with blood were also excluded, as were all of the strips taken from that site. Samples from 10 R.A.F. subjects provided most of those that were considered either known (n = 1l), or suspected (n = 8) to be contaminated with saliva. These were selected for use in the assay for the detection of salivary a-amylase. Other samples were collected from patients attending the Periodontal Department, London Hospital Dental School, and also from laboratory staff. Some of these samples were deliberately contaminated with saliva by relaxing the normal procedure for drying and isolating the teeth. Otherwise samples from these subjects were categorized as described above. This resulted in 90 samples for assay; the number of strips in each category is shown in Table 3: known contaminated (n = 16), suspected contaminated (n = 16) and noncontaminated (n = 58). Samples were recovered from the strips using the centrifugal elution procedure described by Griffiths et al. (1988), but the elution volume was increased from 20 to 25 ~1 of distilled water. Assay method. The 90 samples of GCF were analysed immunochemically in the Bio-Dot microfiltration apparatus (Biorad Laboratories Ltd, Watford, Hertford, U.K.). A dilution series of saliva from the same subject was also analysed. Each experiment also included a dilution series of normal human serum, so as to confnm the absence of cross-reactivity in the developing antisera. Wells containing only PBS (pH 7.2) were included as a negative control to
Table 2. Clinical and laboratory data for the subject illustrated in Fig. 2
strip number 2; : s5
Site 16mp 16mp 43db 25 mb 25mb
Saliva present
Periotron reading
Volume of GCF sample (nl)
+ + + + -
131 54 03 47 06
540 216 24 188 34
+
Detected volume of saliva* (nl)
Contamination (%)
160 25 32 -
74 29 42.5 -
Percentage contamination of saliva was derived from the volume of GCF obtained from Periotron readings and the detected volume of saliva, which was derived from the a -amylase assay. mp, Mesial palatal; db, distal buccal; mb, me&l buccal. *Volume of saliva detected in 20 ~1 of 50 ~1 eluant. + + , Known contamination; + , suspectedcontamination; -, no contamination.
Contamination
of crevicular fluid
561
Table 3. Numbers and percentages of GCF samples co&med to be contaminated with human saliva categorized by clinical assessment of salivary contamination Reason for non-detection GCF Samnle Status Known contaminated Suspected contaminated Noncontaminated
n
Number with saliva detected
Saliva detected (%)
16
13
81.25
16
8
50
3
5
58
3
5.2
23
32
establish the level of background staining. A dilution series of a-amylase (Sigma, Poole, Dorset, U.K.) was always included, both as an antigen-positive control
and to determine the sensitivity of the assay. All dilutions were made with PBS filtered through a 0.25 pm filter, as earlier work had shown falsepositive results in the assay due to non-specific binding of particulate matter. Twenty ~1 of GCF eluate were diluted 1: 5 in PBS, as the larger volume facilitated application in the dot-blot apparatus. A 12 x 8 cm piece of nitrocellulose membrane (Schleicher & Schull, Dassell, Germany), soaked overnight in filtered PBS, was placed in the Bio-dot apparatus. Samples were applied to the membrane and allowed to drain through for 1 h under gravity, with any residuum drawn through by application of a vacuum. The nitrocellulose membrane was then removed from the apparatus and blocked in 5% bovine serum albumin/PBS for 30 min. The membrane was then washed in PBS/O. 1% Tween, with repeated changes of solution, for 1 h. Salivary a-amylase was detected in a doubleantibody technique using sheep anti-human salivary a-amylase, 1:250, for 2 h (Serotec, Bicester, Oxon, U.K.) followed by peroxidase-conjugated rabbit antisheep immunoglobulins, 1: 100, for 2 h (Dakopatts, Glostrup, Denmark) and colour development with 4-chloronaphthol [4CN (Sigma, Poole, Dorset, U.K.)]. Preliminary experiments had shown a high background reactivity of these antisera with human serum. This was reduced to acceptable levels by batch absorption of the anti-human salivary a-amylase with 50% human serum for 30 min and passage of the rabbit anti-sheep immunoglobulins down a column of human gammaglobulin (Sigma, Poole, Dorset, U.K.) coupled to A&Gel 10 (Biorad Laboratories Ltd, Watford, Hertford, U.K.). Densitometric analysis of dot-blots. Photographic colour prints of the dot-blots were analysed with an Ultroscan Laser Densitometer using the Gelscan programme (LKB Instruments, Milton Keynes, Bucks., U.K.). Standard curves were produced for each subject using the saliva dilution series. The lower limit of detection for each subject’s standard curve was determined by the ability to discriminate between the peak heights of higher dilutions and the background levels obtained for both PBS and serum. The peak heights were calculated for each of the GCF samples and by reference to the standard curve; an evaluation of the presence or absence of salivary a-amylase was then made within the limits of
Mow detection level
Volume too low -
3
detection. When salivary a-amylase was shown to be present, an estimation of the volume of salivary contamination of the 20 ~1 sample of GCF was made by reference to the standard curve. This was described as the ‘detected volume of saliva’ and was compared with the ‘volume of GCF sample’, which was derived from measurement in the Periotron, followed by elution into 50~1 of de-ion&d water. This allowed calculations of the percentage contamination to be made, using the formula: detected volume of saliva volume of GCF
50 x5x
100.
REsuLTs
Figure 1 shows the mean and SDS of the volume determinations in the Periotron 6000 of plaque samples applied to 2 x 8 mm chromatography paper strips. As expected, there was a consistent increase in the volume detected with increasing plaque mass applied to the strips. The mean volume when plaque was coded as zero was 0.02 (+ 0.01) p 1,whereas when plaque was coded as 3 the mean volume was 0.15 (kO.07) pl. Table 1 shows data on the visual assessment of GCF contamination of strips from the longitudinal study of R.A.F. recruits. We considered that 3.3% of the recruits had samples contaminated with saliva and that 7.1% were suspect for contamination; this
a25 r MO-
s
0.15 -
E z
0.10 -
a05 -
0
1 Plaque
2
n=21 I 3
Index
Fig. 1. The effect of plaque mass on the volume determination of GCF using the Periotron 6000.
G. S. Gamma
--r--Calibration ~~~.m.+.Thrsrhold
I 20
0
I 40
Volume
I
I
I
I
60
60
100
120
of saliva
(nl)
Fig. 2. Standard curve of salivary volume estimation using an immunochemical assay of cc-amylase. represents 10.4% of the population studied. As two sites per subject were sampled the site rather than the subject data may have produced different figures for contamination. However, as no subject was deemed to have both sites contaminated or suspected, the absolute figures were the same as for the subject data but the percentage values were decreased. When data were further broken down to the level of individual strips, the n value rose to 1740 and the increase in absolute values of contaminated or suspect contaminated strips was due to several strips being contaminated at an individual site. Figure 2 shows the peak heights estimated by densitometry of dots in a dilution series of saliva from a representative subject. These were plotted against the amount of saliva added to each well and a ‘best fit’ equation calculated and plotted. Calculations of the a-amylase concentration of saliva samples were made from comparisons with a similar dilution series of a-amylase (Sigma, Poole, Dorset, U.K.). The sample containing 10nl of saliva, which had a peak height of 0.046, was distinguishable from the background absorbance values obtained with both PBS and blank wells, which represented the absorbance value of the nitrocellulose membrane alone. Extrapolation of this to the line of ‘best fit’ indicates that the threshold for detection in this subject was approx. 15 nl. Figure 3 shows the peak heights for all the GCF samples (Sl-SS) in this subject. Samples S3 and SS gave values less than 0.046 and were therefore below the threshold of detection. Samples Sl, S2 and S4 were all above this point and reference to the equation for
-
z .o
Detection
threshold
0.10
r g
0.05
k! 0
si
s2 GCF
s3 sample
S4
SS
number
Fig. 3. Concentration of a-amylase in GCF samples as assessedby laser densitometry. GCF samples were collected from the subject shown in Fig. 2.
et al.
the line of best fit, derived from Fig. 2, allowed the contribution of saliva in the sample to be calculated. These calculations and the relevant clinical data are presented in Table 2. Both Sl and S2, which were expected to contain saliva showed detectable levels that represented 74 and 29%, respectively, of the total GCF sample volume. The other sample showing a-amylase, S4, was suspected of being contaminated and was found to have 42.5% salivary contamination. Neither S3 nor S5 showed detectable a -amylase as expected. Of the 90 samples tested (Table 3), 81.25% of the sites known to be contaminated with saliva had detectable levels of cr-amylase. The remaining 18.75%, although clinically designated as containing saliva, were below the threshold of detection in this immunoassay. Of the samples suspected of salivary contamination, 50% had detectable levels of a -amylase, although 3/l 6 samples (19%) had too low a volume of sample to achieve detection of a-amylase even if it had consisted only of saliva. Of the 58 samples designated clinically as uncontaminated with saliva only, 3/58 (5.2%) were found to have detectable levels of a-amylase. However, at 23/58 sites (40%) too low a volume of GCF had been collected to enable detection of a-amylase.
DISCUSSION
We show a potential source of inaccuracy from the effect of plaque when using the Periotron 6000 to measure the volume of GCF. A small plaque mass gave an insignificant fluid volume, but a more substantial plaque mass clearly gave larger volumes of fluid that would make an important contribution to the estimated volume of a GCF sample. This is in agreement with observations by Stoller et al. (1990), who found that the presence of supragingival plaque caused a significant increase in measurements of GCF in the Periotron 6000. Some of our R.A.F. subjects had sampling sites that scored 2 or 3 on the plaque index (Silness and Lee, 1964). Whilst a careful technique that removes plaque before sampling will not involve contamination on this scale, it is important that the operator should be aware of the possible contaminating effects of smaller amounts of plaque on the volume determination and measures should be taken to minimize this. Before the work of Stoller et al. (1990) and ourselves, there had been only scant reference made to the possibility of plaque interfering with the collection of GCF. However, Oliver et al. (1969) and Rudin et al. (1970) did attribute their detection of GCF at healthy sites to the contribution of plaque contamination, and Sidi and Ashley (1984) listed saliva, blood and plaque fluid as possible contaminants. Table 1 gives an indication of the likely influence of saliva on estimations of GCF volume in a clinical investigation. Despite careful isolation, which included cotton-wool rolls and a flanged saliva ejector, 10% of our subjects were categorized as having either contamination or suspected contamination with saliva. Per strip the figure is mom acceptable, as only 1.6% of strips were either known or suspected to be contaminated with saliva.
Contamination of crevicular fluid Immunological methods have been used as a more sensitive way of detecting salivary components in GCF. By immunoelectrophoresis, Weinstein et al. (1967) showed no precipitation with samples of GCF when antisera to parotid or submaxillary saliva were used, indicating the absence of salivary proteins in GCF. This method is not, however, very sensitive and could not exclude minor degrees of contamination of GCF with saliva. Douglas er al. (1986) used Western blots of GCF probed with anti-human saliva antiserum. Unfortunately, they did not present the results of these experiments, but commented only that some constituents of GCF were recognized by the anti-salivary antiserum and that the molecular weights of’ the bands detected were totally different from those of mixed saliva. Our immunochemical method could detect 7.5 ng of a-amylase (data not shown), which, based on a mean value of 0.38 mg/ml of cr-amylase in saliva (Jenkins, 1978) represents approx. 20 nl of saliva. The actual threshold of detection within the assay will vary from individual to individual dependent upon the concentration of a-amylase in the saliva. Thus the inclusion of a dilution series of each subject’s saliva in all the assays allowed for easier assessment of the presence of saliva within the GCF samples. In the example presented (Fig. 2) the limit of detection was calculated as 15 nl of saliva. Table 3 shows th.at the detection of a-amylase confirmed the clinica. impression of the presence or absence of saliva. Where saliva was deemed present visually, a-amylase could be detected in the majority of cases. However, in some instances the volume of GCF collected was small and consequently would require over 50% of the sample to be saliva before it reached the threshold of detection in the a-amylase assay. In samples where there was no reason to suspect salivary contamination, a-amylase was only detected in only 5.2% of strips. This value was artificially low because 23 samples had a volume of GCF below the threshold of detection for a -amylase. If these samples are excluded the level of unsuspected contamination is 3/35 (8.6%). However, this may be an overestimation, as this group of samples was biased by the fact that strips categorized as uncontaminated were often from sites where one of the other strips had been categorized as contaminated or suspect contaminated. Nevertheless these figures were still well below the 50% detection of a-amylase found in sites suspected of being contaminated with saliva and the same arguments applied in this category, with three sites yielding a GCF volume below the threshold of detection of a -amylase. The problem of small volumes of GCF collected and the variation between subjects in the concentrations of a-amylase, and therefore the threshold of detection of the a-amylase assay, illustrate the shortcomings of this assay for routine use. If the assay were to be used on all samples in order either to ensure the absence of salivary contamination, or to be able to state, for example, that all samples contained < 10% saliva, it would be necessary to either use the whole sample volume or be restricted to those samples that initially had yielded a large enough volume. In this study, the use of the whole sample AOB 37/7-D
563
volume was not acceptable, as other investigations of the composition of GCF were being made. Restricting the assay to samples with a large enough volume may allow us to confirm or refute whether the unusually large volume collected at a site was either a reflection of GCF volume, or an indication of salivary contamination. This was one of the indications for samples to be designated as ‘suspected contaminated’. We have shown that if GCF samples are collected using a good technique for isolation and drying, then the likelihood that saliva will be a significant contaminant is small. When sites are contaminated with saliva the detection of a -amylase confirms the clinical impression. Where sites were suspected of contamination the assay was able to confirm or refute a significant contribution from such contamination. Table 3 shows that 5/16 samples (31.25%) suspected of salivary contamination would have been confirmed as free of saliva within the detection limits of our assay. Thus if GCF samples are seen to be contaminated with saliva, they should be discarded. If samples are considered free of contamination they may be used with confidence. However, samples suspected of salivary contamination should be treated with caution and may require an a-amylase assay before further analysis. are grateful to the R.A.F. Dental Branch for their cooperation in this study and to the statI of the Institute of Dental Health and Training, R.A.F. Halton, for their unstinting support. This work was supported by the Medical Research Council.
Acknowledgements-We
REFERENCES
Cimasoni G. (1983) Crevicular fluid undated. Monographs in Oral Science, Vol. 12. Karger, Basel. Curtis M. A., Gillett I. R., Grilhths G. S., Maiden M. F. J., Steme J. A. C.. Wilson D. T.. Wilton J. M. A. and Johnson N. W. (1989) Detection of high-risk groups and individuals for periodontal diseases: laboratory markers from analysis of gingival crevicular fluid. J. clin. Periodont. 16, 1-l 1. Douglas C. W. I., Craig G. T. 0. and Neil T. C. A. (1986) A novel antigen in fluid collected from established periodontal pockets. The Borderland between Caries and Periodontal Disease III (Eds Lehner T. and Cimasoni G.) Editions Medecine et Hygiene, Geneve. Griffiths G. S., Curtis M. A. and Wilton J. M. A. (1988) Selection of a hlter paper with optimum properties for the collection of gingival crevicular fluid. J. periodont. Res. 23, 33-38. Holm-Pedersen P., Agerbaek N. and Theilade E. (1975) Experimental gingivitis in young and elderly individuals. J. clin. Periodont. 2, 14-24. Jenkins G. N. (1978) The Physiology and Biochemistry of the Mouth, 4th edn, Chap. IX, p. 286 and Chap. X, p. 371. Blackwell Scientific Publications, Oxford. Liie H. and Holm-Pedersen P. (1965) Absence and presence of fluid from normal and inflamed gingivae. Periodontics 3, 171-177. Mann W. V. (1963) The correlation of gingivitis pocket depth and exudate from the gingival crevice. J. Periodont. 34, 379-387. Miihlemann H. R. and Son S. (1971) Gingival sulcus bleeding-a leading symptom in initial gingivitis. Jfefu. odont. Acta. 15, 107-113.
564
G. S. GRIFFITHS er al.
Oliver R. C., Hahn-Pedersen P. and L6e H. (1969) The correlation between clinical scoring, exudate measurements and microscopic evaluation of inflammation in the gingiva. J. Periodont. 40, 201-209. Rudin H. J., Overdiek H. F. and Rateitschak K. H. (1970) Correlation between sulcus fluid rate and clinical and histological inflammation of the marginal gingiva. Helu. odont. Acta. 14, 21-26. Sidi A. D. and Ashley F. P. (1984) Influence of frequent sugar intakes on experimental gingivitis. J. Periodont. 55, 419423.
Silness J. and Tie H. (1964) Periodontal disease in pregnancy. II. Correlation between oral hygiene and periodontal condition. Acta odont. stand. U, 121-135. Stoller N. H., Karras D. C. and Johnson L. R. (1990) Reliability of crevicular fluid measurements taken in the presence of supragingival plaque. J. Periodont. 61, 67&673. Weinstein E., Mandel I., Salkind A., Oshrain H. I. and Pappas G. D. (1967) Studies of gingival fluid. Periodontics 5. 161-166.
