Comparative Evaluation of the Limulus Assay and the Direct Gram Stain for Detection of Significant Bacteriuria JAMES
H. JORGENSEN, PH.D.,
AND PAMELA
M. JONES, M.T. (ASCP)
Departments of Pathology and Microbiology, The University of Texas Health Science Center at San Antonio and the Bexar County Hospital, San Antonio, Texas 78284
ABSTRACT
I N GENERAL, there have been three separate approaches to development of new screening tests for the detection of significant bacteriuria. One approach has been the introduction of miniaturized inexpensive culture devices. 2,5 However, these devices still require a traditional culture incubation period of 18-24 hours. A second approach has been the detection of bacterial products or enzymes in urine by chemical tests such as the Griess nitrate Received J u n e 13, 1974, revised July 24, 1974; accepted July 24, 1974. Supported in part by NIH-General Research Support Grant # 5 S01RR05654-05. Address reprint requests to Dr. Jorgensen. 142
reductase test17 or the measurement of tetrazolium reductase. 12 These tests have the advantage of providing more rapid results, but are often much less reliable. 6 ' 18 A third approach has been the visualization of bacteria in urinary sediment by use of either the phase-contrast microscope 1 or a Gram stain of fresh uncentrifuged urine. 13 In particular, the Gram stain has been reported to correlate with significant colony counts in 7 5 - 9 5 % of cases.3'11,12 However, considerable judgment must often be applied to the estimation of the number of organisms visualized in Gram-stained smears.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Jorgensen, James H., and Jones, Pamela M.: Comparative evaluation of the Limulus assay and the direct Gram stain for detection of significant bacteriuria. Am. J. Clin. Pathol. 63: 142-148, 1975. A double-blind study comparing the Limulus in-vitro endotoxin assay with the direct Gram stain of uncentrifuged urine for detection of significant bacteriuria was performed. One-thousand seventy-seven urine specimens were examined by the two methods and the results compared with results of quantitative urine cultures. Two hundred three samples produced growth of >10 5 organisms per ml. urine. The Limulus assay detected 86.2% of these specimens, and 98.8% of urines that contained >10 5 Gram-negative bacilli per ml. The Gram stain procedure detected only 69.5% of urines containing >10 5 organisms per ml. and 74.5% of specimens with >10 5 Gram-negative bacteria per ml. urine. T h e Limulus assay demonstrated both greater sensitivity and greater specificity than the Gram stain procedure. Moreover, the Limulus test is much less susceptible to errors of interpretation than methods involving microscopy. (Key words: Endotoxin in urine; Detection of bacteriuria; Gram stain for bacteriuria; Limulus assay for bacteriuria.)
January 1975
143
DETECTION OF BACTERIURIA
Table 1. Criteria for Evaluation of Gram Stains of Urinary Sediment WBC's/Field
Classification
Classification scheme for leukocytes seen with 43 x objective >20 12-19 8-11 4-7 10 6-9 2-5 1 10 5 Organisms per ml. Organism
TOTAL
Limulus Positive*
96 17 8 6 2 2 1 1 1 1 3 2 1 3 3 21 6 5 •1 12 7
94 17 8 6 2 2 1 1 1 1 0 0 0 0 0 21 6 0 2 8 5
(97.9%) (100.0%) (100,0%) (100.0%) (100.0%) (100,0%) (100.0%) (100.0%) (100.0%) (100.0%)
203
175
Gram Stain Positive*
(100.0%)
(50.0%) (61.5%) (71.4%)
78 14 7 5 0 1 0 0 0 0 3 0 0 1 3 12 3 3 1 6 4
(86.2%)
141
(69.5%)
(100.0%) (100.0%)
(81.3%) (82.4%) (87.5%) (83.3%) (50.0%)
(33.3%) (100.0%) (61.1%) (50.0%) (60.0%) (25.0%) (46.2%) (57.1%)
* Percentages indicate 10 5 organisms per ml. urine, as defined by Kass."
Table 3. Comparison of Limulus Assay and Gram Stain Results on Urine Specimens Containing 104—105 Organisms per ml. Organism
Escherichia coli Klebsiella pneumoniae Pseudomonas aeruginosa Proteus spp. Misc. GNR'st Group D Streptococcus Staphylococcus epidermidis Group B Streptococcus Streptococcus viridans Candida albicans 2 GNR's 3 or >GNR's 2 GP'st 3 or >GP's 1GNR+1GP Mult. GNR's + GP's TOTAL
]Limulus Positive*
Total No. 16 3 3 3 5 10 7 2 3 2 7 2 18 5 13 12 111
7 2 1 2 0 0 0 0 0 0 2 1 5 1 2 3
Gram Stain Positive*
(43.8%) (66.7%) (33.3%) (66.7%)
2 1 0 0 0 0 0 0 0 0 1 0 4 1 1 2
(28.6%) (50.0%) (27.8%) (20.0%) (15.4%) (25.0%)
26 (23.4%)
(12.5%) (33.3%)
(14.3%) (22.2%) (20.0%) (7.7%) (16.7%)
12 (10.8%)
* Percentages indicate % of total number positive. t GNR = Grain-negative rod: GP = Gram-positive organism.
Table 4. Correlation of Gram Stain and Culture Results (>10 5 Organisms per ml.)* Organisms
Correct
Incorrect
Partly Correct
GNR's GPC's GPR's GN + GP Yeast
110 3 0 0 3
0 6 3 0 0
5 4 1 5 1
* GNR = Gram-negative rod; GPC = Gram-positive coccus; GPR = Gram-positive rod.
Controlling Bias A double-blind protocol was employed to minimize bias in comparing the results of Limulus assay and the Gram stain with results of quantitative cultures. The urine cultures were performed in the usual fashion by the routine clinical laboratory technologists. Results of Limulus tests and Gram stains were read and recorded separately by the two authors prior to the availability of the culture results.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Escherichia coli Klebsiella pneumoniae Pseudomonas aeruginosa Proteus mirabilis Enterobacler aerogenes Enterobacter cloacae Serralia tnarcescens Providencia spp. Herellea vaginocola Citrobacter freundii Staphylococcus aureus Group 1) Streptococcus Group B Streptococcus Streptococcus viridaus Candida albicans 2 GNR'st 3 or >GNR's 2 CP'st 8 or >CP's 1 GNR + 1 GP Mull. GNR's + GP's
Total No.
63
January 1975
DETECTION OF BACTERIUR1A
Calculation of Sensitivity and Specificity The definitions of sensitivity and specificity are those used by the World Health Organization 19 : sensitivity = diseased persons with positive test/all persons in population with disease; specificity = non diseased persons with negative test/all persons in population without disease. Results
Table 5. Correlation of Pyuria* with Bacteriuriaf
WBC's
Culture >10 5 per ml.
Culture 10 4 -10 5 per ml.
Culture IO'perml.
* GNR = Gram-negative rod.
Gram stains is summarized in Table 6. The Limulus assay detected 86.2% of urine specimens containing > 1 0 5 organisms per ml., while the Gram stain detected only 69.5%. In the case of urines containing >10 5 Gram-negative bacilli per ml., the Limulus assay detected 98.8%, while 74.5% were detected by the Gram stain. The overall ability to classify urine specimens correctly as containing either 10 5 per ml. indicates that the Limulus assay correctly classified 94.4% of all urines, whereas the Gram stain method accurately classified 88.5% of the samples. Table 7 indicates the results of calculations of the sensitivity and specificity of each test. The Limulus assay has a sensitivity of 86.2% and a specificity of 96.1%. T h e Gram stain procedure has a sensitivity of 69.5% and a specificity of 92.8%. Discussion The need for periodic screening of certaia high-risk groups such as pregnant women, school girls, and diabetics for the presence of significant bacteriuria has been well established. 7,16 T h e pour-plate quantitative culture has been accepted as
63
the most accurate laboratory procedure available for documenting bacteriuria. A streak plate made with a calibrated platinum loop is a suitable alternate method requiring less time and equipment. 7 However, both methods still require skill, time, and equipment not readily available in all office or clinic settings. Consequently, several miniaturized culture devices and chemical tests have been proposed as practical screening devices for bacteriuria. However, most of these procedures have met with rather limited acceptance. 6 T h e findings of the present investigation confirm our previous preliminary observations 9 that the Limulus in-vitro endotoxin assay can be used successfully for the detection of significant bacteriuria. The results of the Limulus test can be available within l-2'/2 hr. of collection of a urine specimen. Although results of Gram stains can be obtained in a shorter period, more time is required of technical personnel for preparation and interpretation of a Gram-stained smear than is required for performance of a Limulus assay. Furthermore, performance of the Limulus test does not require the degree of subjective judgment for accurate quantitation of bacteria required by methods involving microscopy of urinary sediment. This is emphasized by the considerably greater numbers of both false-positive and false-negative results observed with the Gram stain. The sensitivity and specificity of the Limulus assay for endotoxin results in detection of nearly 100% of urines containing >10 5 Gram-negative bacteria per ml. A major deficiency of the Limulus assay for this purpose is that Grampositive bacteria and yeast are undetectable, due to their lack of endotoxin. Therefore, the predictive value of the Limulus assay for significant bacteriuria lies in the normal preponderance of Gram-negative uropathogens, which possess endotoxin.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Limulus Assay s
A.J.C.P.—Vol.
January 1975
DETECTION OF BACTERIURIA
147
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Table 7. Sensitivity* and Specificity* The direct Gram stain of uncentrifuged urinary sediment appears to have a of the Limulus Assay and the false-negative rate of approximately 30% Direct Gram Stain irrespective of the group of organisms Limulus Assay Gram Stain involved. The data in Table 4 indicate that actually more incorrect responses Sensitivity 86.2% 69.5% 96.1% 92.8% regarding the basic identities of organisms Specificity False-positive rate 3.9% 7.2% seen in urinary sediment occurred with False-negative rate 13.8% 30.5% Gram-positive organisms. Therefore, both * Based on calculations described in text. the Limulus assay and the direct Gram stain procedure seem better able to define bacteriuria accurately when Gram-nega- were further investigated by review of the tive organisms are involved. medical charts of the 27 patients comprisNeither the Limulus assay nor direct ing these two groups. Current antimicromicroscopy could be considered an ac- bial chemotherapy was a probable explaceptable method for detection of bac- nation for negative urine cultures from teriurias involving 104—105 organisms per five of these patients. An additional eight ml. However, the Limulus assay again patients with positive Limulus assays were defined significantly more urines in this diabetic and might be expected to have an group than was possible using the Gram increased incidence of bacteriuria; howstain. ever, urine cultures failed to confirm this The percentage agreement of the Gram supposition. No apparent explanation was stain procedure with quantitative cultures found for the positive Limulus tests of reported here is lower than that of certain urine from the remaining 14 patients. previous studies. 3,8 However, in the study Another important factor in the evaluaby Hoeprich, 8 a Gram stain was consid- tion of any potential screening test is the ered predictive of bacteriuria of 10,000 cost of performing the procedure. The organisms per ml. or more (and therefore cost of a Limulus assay as performed in positive) "if bacteria were seen in nearly this study would be approximately 75^! every field." Even with these less stringent per specimen. While this seems reasonacriteria, Hoeprich experienced a false- bly inexpensive, it cannot compare to the positive rate of 11.5%. Brupacher and almost negligible cost of preparing a associates 3 considered a Gram-stained Gram-stained smear. smear positive if "any bacteria were T h e lack of predictive value for bacnoted." The latter authors also considered teriuria obtained by visualization of a colony count of 10,000 bacteria per ml. leukocytes in unsedimented urine is in to be significant. Therefore, the more agreement with findings of the previous stringent criteria for positivity (at least study of Pryles and Eliot.14 T h e data in one organism per oil-immersion field and Table 5 emphasize that bacteriuria may be at least 100,000 organisms per ml.) used present in the absence of pyuria as in the present investigation probably ex- defined, and pyuria may be present withplain the lower predictive value of the out bacteriuria. Therefore, the visualizaGram stain in our hands. tion of significant numbers of leukocytes Specimens in this study which produced in urinary sediment may be suggestive of growth of
H. JORGENSEN, PH.D.,
AND PAMELA
M. JONES, M.T. (ASCP)
Departments of Pathology and Microbiology, The University of Texas Health Science Center at San Antonio and the Bexar County Hospital, San Antonio, Texas 78284
ABSTRACT
I N GENERAL, there have been three separate approaches to development of new screening tests for the detection of significant bacteriuria. One approach has been the introduction of miniaturized inexpensive culture devices. 2,5 However, these devices still require a traditional culture incubation period of 18-24 hours. A second approach has been the detection of bacterial products or enzymes in urine by chemical tests such as the Griess nitrate Received J u n e 13, 1974, revised July 24, 1974; accepted July 24, 1974. Supported in part by NIH-General Research Support Grant # 5 S01RR05654-05. Address reprint requests to Dr. Jorgensen. 142
reductase test17 or the measurement of tetrazolium reductase. 12 These tests have the advantage of providing more rapid results, but are often much less reliable. 6 ' 18 A third approach has been the visualization of bacteria in urinary sediment by use of either the phase-contrast microscope 1 or a Gram stain of fresh uncentrifuged urine. 13 In particular, the Gram stain has been reported to correlate with significant colony counts in 7 5 - 9 5 % of cases.3'11,12 However, considerable judgment must often be applied to the estimation of the number of organisms visualized in Gram-stained smears.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Jorgensen, James H., and Jones, Pamela M.: Comparative evaluation of the Limulus assay and the direct Gram stain for detection of significant bacteriuria. Am. J. Clin. Pathol. 63: 142-148, 1975. A double-blind study comparing the Limulus in-vitro endotoxin assay with the direct Gram stain of uncentrifuged urine for detection of significant bacteriuria was performed. One-thousand seventy-seven urine specimens were examined by the two methods and the results compared with results of quantitative urine cultures. Two hundred three samples produced growth of >10 5 organisms per ml. urine. The Limulus assay detected 86.2% of these specimens, and 98.8% of urines that contained >10 5 Gram-negative bacilli per ml. The Gram stain procedure detected only 69.5% of urines containing >10 5 organisms per ml. and 74.5% of specimens with >10 5 Gram-negative bacteria per ml. urine. T h e Limulus assay demonstrated both greater sensitivity and greater specificity than the Gram stain procedure. Moreover, the Limulus test is much less susceptible to errors of interpretation than methods involving microscopy. (Key words: Endotoxin in urine; Detection of bacteriuria; Gram stain for bacteriuria; Limulus assay for bacteriuria.)
January 1975
143
DETECTION OF BACTERIURIA
Table 1. Criteria for Evaluation of Gram Stains of Urinary Sediment WBC's/Field
Classification
Classification scheme for leukocytes seen with 43 x objective >20 12-19 8-11 4-7 10 6-9 2-5 1 10 5 Organisms per ml. Organism
TOTAL
Limulus Positive*
96 17 8 6 2 2 1 1 1 1 3 2 1 3 3 21 6 5 •1 12 7
94 17 8 6 2 2 1 1 1 1 0 0 0 0 0 21 6 0 2 8 5
(97.9%) (100.0%) (100,0%) (100.0%) (100.0%) (100,0%) (100.0%) (100.0%) (100.0%) (100.0%)
203
175
Gram Stain Positive*
(100.0%)
(50.0%) (61.5%) (71.4%)
78 14 7 5 0 1 0 0 0 0 3 0 0 1 3 12 3 3 1 6 4
(86.2%)
141
(69.5%)
(100.0%) (100.0%)
(81.3%) (82.4%) (87.5%) (83.3%) (50.0%)
(33.3%) (100.0%) (61.1%) (50.0%) (60.0%) (25.0%) (46.2%) (57.1%)
* Percentages indicate 10 5 organisms per ml. urine, as defined by Kass."
Table 3. Comparison of Limulus Assay and Gram Stain Results on Urine Specimens Containing 104—105 Organisms per ml. Organism
Escherichia coli Klebsiella pneumoniae Pseudomonas aeruginosa Proteus spp. Misc. GNR'st Group D Streptococcus Staphylococcus epidermidis Group B Streptococcus Streptococcus viridans Candida albicans 2 GNR's 3 or >GNR's 2 GP'st 3 or >GP's 1GNR+1GP Mult. GNR's + GP's TOTAL
]Limulus Positive*
Total No. 16 3 3 3 5 10 7 2 3 2 7 2 18 5 13 12 111
7 2 1 2 0 0 0 0 0 0 2 1 5 1 2 3
Gram Stain Positive*
(43.8%) (66.7%) (33.3%) (66.7%)
2 1 0 0 0 0 0 0 0 0 1 0 4 1 1 2
(28.6%) (50.0%) (27.8%) (20.0%) (15.4%) (25.0%)
26 (23.4%)
(12.5%) (33.3%)
(14.3%) (22.2%) (20.0%) (7.7%) (16.7%)
12 (10.8%)
* Percentages indicate % of total number positive. t GNR = Grain-negative rod: GP = Gram-positive organism.
Table 4. Correlation of Gram Stain and Culture Results (>10 5 Organisms per ml.)* Organisms
Correct
Incorrect
Partly Correct
GNR's GPC's GPR's GN + GP Yeast
110 3 0 0 3
0 6 3 0 0
5 4 1 5 1
* GNR = Gram-negative rod; GPC = Gram-positive coccus; GPR = Gram-positive rod.
Controlling Bias A double-blind protocol was employed to minimize bias in comparing the results of Limulus assay and the Gram stain with results of quantitative cultures. The urine cultures were performed in the usual fashion by the routine clinical laboratory technologists. Results of Limulus tests and Gram stains were read and recorded separately by the two authors prior to the availability of the culture results.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Escherichia coli Klebsiella pneumoniae Pseudomonas aeruginosa Proteus mirabilis Enterobacler aerogenes Enterobacter cloacae Serralia tnarcescens Providencia spp. Herellea vaginocola Citrobacter freundii Staphylococcus aureus Group 1) Streptococcus Group B Streptococcus Streptococcus viridaus Candida albicans 2 GNR'st 3 or >GNR's 2 CP'st 8 or >CP's 1 GNR + 1 GP Mull. GNR's + GP's
Total No.
63
January 1975
DETECTION OF BACTERIUR1A
Calculation of Sensitivity and Specificity The definitions of sensitivity and specificity are those used by the World Health Organization 19 : sensitivity = diseased persons with positive test/all persons in population with disease; specificity = non diseased persons with negative test/all persons in population without disease. Results
Table 5. Correlation of Pyuria* with Bacteriuriaf
WBC's
Culture >10 5 per ml.
Culture 10 4 -10 5 per ml.
Culture IO'perml.
* GNR = Gram-negative rod.
Gram stains is summarized in Table 6. The Limulus assay detected 86.2% of urine specimens containing > 1 0 5 organisms per ml., while the Gram stain detected only 69.5%. In the case of urines containing >10 5 Gram-negative bacilli per ml., the Limulus assay detected 98.8%, while 74.5% were detected by the Gram stain. The overall ability to classify urine specimens correctly as containing either 10 5 per ml. indicates that the Limulus assay correctly classified 94.4% of all urines, whereas the Gram stain method accurately classified 88.5% of the samples. Table 7 indicates the results of calculations of the sensitivity and specificity of each test. The Limulus assay has a sensitivity of 86.2% and a specificity of 96.1%. T h e Gram stain procedure has a sensitivity of 69.5% and a specificity of 92.8%. Discussion The need for periodic screening of certaia high-risk groups such as pregnant women, school girls, and diabetics for the presence of significant bacteriuria has been well established. 7,16 T h e pour-plate quantitative culture has been accepted as
63
the most accurate laboratory procedure available for documenting bacteriuria. A streak plate made with a calibrated platinum loop is a suitable alternate method requiring less time and equipment. 7 However, both methods still require skill, time, and equipment not readily available in all office or clinic settings. Consequently, several miniaturized culture devices and chemical tests have been proposed as practical screening devices for bacteriuria. However, most of these procedures have met with rather limited acceptance. 6 T h e findings of the present investigation confirm our previous preliminary observations 9 that the Limulus in-vitro endotoxin assay can be used successfully for the detection of significant bacteriuria. The results of the Limulus test can be available within l-2'/2 hr. of collection of a urine specimen. Although results of Gram stains can be obtained in a shorter period, more time is required of technical personnel for preparation and interpretation of a Gram-stained smear than is required for performance of a Limulus assay. Furthermore, performance of the Limulus test does not require the degree of subjective judgment for accurate quantitation of bacteria required by methods involving microscopy of urinary sediment. This is emphasized by the considerably greater numbers of both false-positive and false-negative results observed with the Gram stain. The sensitivity and specificity of the Limulus assay for endotoxin results in detection of nearly 100% of urines containing >10 5 Gram-negative bacteria per ml. A major deficiency of the Limulus assay for this purpose is that Grampositive bacteria and yeast are undetectable, due to their lack of endotoxin. Therefore, the predictive value of the Limulus assay for significant bacteriuria lies in the normal preponderance of Gram-negative uropathogens, which possess endotoxin.
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Limulus Assay s
A.J.C.P.—Vol.
January 1975
DETECTION OF BACTERIURIA
147
Downloaded from http://ajcp.oxfordjournals.org/ by guest on June 5, 2016
Table 7. Sensitivity* and Specificity* The direct Gram stain of uncentrifuged urinary sediment appears to have a of the Limulus Assay and the false-negative rate of approximately 30% Direct Gram Stain irrespective of the group of organisms Limulus Assay Gram Stain involved. The data in Table 4 indicate that actually more incorrect responses Sensitivity 86.2% 69.5% 96.1% 92.8% regarding the basic identities of organisms Specificity False-positive rate 3.9% 7.2% seen in urinary sediment occurred with False-negative rate 13.8% 30.5% Gram-positive organisms. Therefore, both * Based on calculations described in text. the Limulus assay and the direct Gram stain procedure seem better able to define bacteriuria accurately when Gram-nega- were further investigated by review of the tive organisms are involved. medical charts of the 27 patients comprisNeither the Limulus assay nor direct ing these two groups. Current antimicromicroscopy could be considered an ac- bial chemotherapy was a probable explaceptable method for detection of bac- nation for negative urine cultures from teriurias involving 104—105 organisms per five of these patients. An additional eight ml. However, the Limulus assay again patients with positive Limulus assays were defined significantly more urines in this diabetic and might be expected to have an group than was possible using the Gram increased incidence of bacteriuria; howstain. ever, urine cultures failed to confirm this The percentage agreement of the Gram supposition. No apparent explanation was stain procedure with quantitative cultures found for the positive Limulus tests of reported here is lower than that of certain urine from the remaining 14 patients. previous studies. 3,8 However, in the study Another important factor in the evaluaby Hoeprich, 8 a Gram stain was consid- tion of any potential screening test is the ered predictive of bacteriuria of 10,000 cost of performing the procedure. The organisms per ml. or more (and therefore cost of a Limulus assay as performed in positive) "if bacteria were seen in nearly this study would be approximately 75^! every field." Even with these less stringent per specimen. While this seems reasonacriteria, Hoeprich experienced a false- bly inexpensive, it cannot compare to the positive rate of 11.5%. Brupacher and almost negligible cost of preparing a associates 3 considered a Gram-stained Gram-stained smear. smear positive if "any bacteria were T h e lack of predictive value for bacnoted." The latter authors also considered teriuria obtained by visualization of a colony count of 10,000 bacteria per ml. leukocytes in unsedimented urine is in to be significant. Therefore, the more agreement with findings of the previous stringent criteria for positivity (at least study of Pryles and Eliot.14 T h e data in one organism per oil-immersion field and Table 5 emphasize that bacteriuria may be at least 100,000 organisms per ml.) used present in the absence of pyuria as in the present investigation probably ex- defined, and pyuria may be present withplain the lower predictive value of the out bacteriuria. Therefore, the visualizaGram stain in our hands. tion of significant numbers of leukocytes Specimens in this study which produced in urinary sediment may be suggestive of growth of
