JOURNAL OF CLINICAL MICROBIOLOGY, May 1992, p. 1205-1209 0095-1137/92/051205-05$02.00/0 Copyright ©) 1992, American Society for Microbiology
Vol. 30, No. 5
Detection of Bacterial Growth by Gas Absorption JOHN R. WATERS Abo, Inc., P.O. Box 271789, Tampa, Florida 33688 Received 25 November 1991/Accepted 13 February 1992
When 24 different aerobic organisms were grown in a shaken culture, all were found to first absorb gas from the headspace. In a rudimentary medium, such as tryptic soy broth, 16 of the 24 organisms did not produce gas following the initial gas absorption. We have developed a simple, noninvasive method for detecting both gas absorption and production in multiple culture vials. The time to positivity was compared with that obtained by the BACTEC 460 blood culture system. For nearly all of these organisms, there was no difference. For some of those organisms that did not produce gas, e.g. Staphylococcus epidermidis, MoraxeUla osloensis, and Neisseria meningitidis, detection by gas absorption was a few hours faster. Gas absorption appears to be a promising technique for a new automated blood culture system because of its simplicity and because medium without special additives can be used to detect organisms that do not produce gas.
Since no organisms are recovered from 85 to 90% of clinical blood cultures, much laboratory time and effort is wasted processing unproductive specimens. This has led to the development of instruments to monitor a large number of culture vials for growth, reporting the few suspicious ones for further investigation. Although several screening techniques have been tried, the only one in widespread use detects the production of carbon dioxide from organisms growing in culture medium. These are the BACTEC systems (4), originally from Johnston Laboratories, which detect either radioactive (10) (BACTEC 225, 301, and 460) or nonradioactive (5) (BACTEC 660 and 730) carbon dioxide. Recently, the BacT/Alert system from Organon Teknika (8), which also detects carbon dioxide, has come into use in several laboratories. In studying the broader mechanism of gas production by growing organisms, we noticed that all the aerobic organisms we tested, when grown in agitated fluid medium, first absorbed gas from their ambient atmosphere. Some later produced gas, but many, when grown in a rudimentary medium without special additives, did not. We developed a simple measuring system (11) that can monitor this gas absorption. If this phenomenon turns out to be as universal as our results indicate, detection of aerobic microorganisms by gas absorption promises to be a powerful technique for a new automated blood culture system.
dium (Carr-Scarborough) was used for initial culture and as a source tube. Test inocula. Serial hundredfold dilutions were made in five tubes, with each containing 5 ml of TSB. From four tubes, starting with the second, eight test bottles were set up with decimal dilutions of the inocula. No growth was usually seen in the last two or three bottles, the inoculum having been diluted to extinction. For anaerobes, dilutions were made in sealed bottles containing 30 ml of prereduced anaerobically sterilized medium. Test procedure. Screw-capped bottles (130-ml overflow volume) were filled with enough TSB to leave 16 to 20 ml of headspace above the liquid, capped, and autoclaved. When cool, the cap was removed and the test inoculum was added. For Haemophilus influenzae, 2% Fildes enrichment (Difco Laboratories) was added. After inoculation, the cap was replaced with another that had a limp, flexible, cone-shaped rubber nipple pointing down into the bottle as shown in Fig. 1. On incubation, the air in the bottle expanded, causing the nipple to rise to be about level with the bottle cap. A sensor head, containing a light-emitting diode (LED) and a light detector arranged to form a horizontal light beam about 0.5 in. (-1.3 cm) above the top of the bottle cap, was put on each bottle. Fifteen test bottles could be put into the test chamber which was mounted on a rotary agitator (180 rpm, 3/8-in. [0.95-cm] stroke) in the incubator set at 36 to
370C. Every 20 min, the shaking was stopped, the LEDs were turned on, and a small vacuum pump slowly reduced the air pressure in the test chamber from ambient (14.7 lb/in2 absolute) to produce a vacuum of about 2.2 lb/in2 gauge (i.e. to 12.5 lb/in2 absolute) in 3 to 4 min. Meanwhile, the computer attached to the sensor heads rapidly scanned the output from each of the light detectors to determine whether a light beam had been broken by a rising nipple, since these gradually became erect as the test chamber pressure was reduced. The computer recorded the vacuum reading in the chamber at that instant. We call this the reference vacuum (11). If the nipple had no stiffness, the gas pressure inside the bottle would equal that in the test chamber. At the end of the test cycle, the chamber was vented, the LEDs were turned off, and the shaker turned on. Changes in the volume of the gas in the bottle caused by gas production or absorption can be detected by monitoring the history of the reference vacua, plotted by the computer,
MATERUILS AND METHODS
Microorganisms. Test organisms were usually ATCC strains in the form of Q-Chek Cultiloops (Chrisope Technologies [formerly Scott Laboratories], Sulphur, La.). These are individually wrapped, disposable, inoculating loops containing stabilized viable microorganisms for direct inoculation. Other organisms were clinical isolates. Preference was given to test organisms which were nonfermenters, thus difficult to detect by gas production, and for their relevance in blood cultures. Table 1 lists the 24 aerobic and 6 anaerobic organisms tested. For aerobic organisms, a source tube was made by breaking off the loop in a tube containing 5 ml of tryptic soy broth (TSB) (Carr-Scarborough Microbiologicals, Stone Mountain, Ga.) which was then incubated for 18 to 42 h until heavy growth was seen. For anaerobes, a tube of prereduced anaerobically sterilized chopped meat me1205
1206
J. CLIN. MICROBIOL.
WATERS G
Vacuum Ipsil
P
1.5
.5
.1
,.. ........
2
...
. k.
.*
.,
.
-.*1~~~~~~W. A~
. 0
. . . AA
E F
FIG. 1. Schematic view of culture bottle (B) showing the nipple (N) and retaining cap (C). The LED light source (S) shines a beam to the detector (D). This is interrupted when the nipple becomes erect as a result of the action of the vacuum pump (P). The pump and an electronic vacuum gauge (G) are attached to the sealed outer incubator.
for each bottle. The detection time is taken when there is a significant excursion from the normal smooth curve for a sterile bottle. This detection time is marked with a circle in later figures showing typical results. For anaerobic cultures, sterile bottles were flushed with BACTEC anaerobic gas flowing through a cannula. While still flushing, anaerobic blood culture medium, e.g. Thiol broth (Difco), was aseptically poured into the bottle, the inoculum was added, and a sterile nipple and cap were applied. BACTEC 460 procedure. Growth and detection times for some of the slower-growing organisms were compared with those for the BACTEC 460 (Becton Dickinson Diagnostic Instrument Systems, Loveton, Md.). Aerobic cultures were tested in 6B soybean-casein digest broth-aerobic and anaerobes were tested in 7D tryptic soy broth-anaerobic. BACTEC bottles were inoculated in parallel with the test bottles, although not at all dilutions. Room air was used as the aerobic culture gas, and a mixture of 85% nitrogen, 10% hydrogen, and 5% carbon dioxide was used for anaerobes. Bottles were agitated on the same shaker and tested every 3 to 4 h during the work day only. Growth Index values were plotted and interpolated to determine the BACTEC detection time corresponding to a Growth Index of 30 (all aerobes, except 20 for one or two nonfermenters) or 20 (anaerobes). Sterile bottles containing blood. Blood from healthy volunteers was drawn into sterile yellow-top tubes (BD Vacutainer 4960) containing 1.7 ml of 0.85% sodium chloride with 0.35% sodium polyanethol sulfonate (SPS) anticoagulant, for a total volume of 10 ml. Test bottles with 30 ml of headspace were inoculated by pouring the tube's contents into the bottle. Test bottles, along with uninoculated normal TSB controls, were tested every 30 min for at least 7 days.
RESULTS Effect of agitation. There was a dramatic change in the shape of the reference vacuum curves when the cultures were agitated. Figure 2 shows the results for an uninoculated
.~
zi_
.
A
A
2C
.S i.
.S
FIG. 2. Effect of agitation of cultures of Serratia marcescens on the reference vacuum curves. Symbols: *, agitated culture; A, unagitated culture; *, control.
control and two cultures with equal inocula of Serratia with and without agitation. The abscissa (x axis) is the reference vacuum with zero vacuum (i.e., atmospheric pressure) at the left and increasing vacuum to the right. The ordinate is elapsed time running from top to bottom of the page. To better demonstrate changes, the three traces have been moved laterally to overlap which is permissible because the absolute value of the vacuum is irrelevant. Detection of bacterial growth is determined by observation of a deviation from the smooth curve generated by a culture with no growth. The unagitated culture shows a decrease in vacuum after 18 h, caused by gas production. In the agitated culture, in contrast, there is a rapid increase in the reference vacuum, caused by gas absorption, after 9 h, and 3 h later, a very rapid decrease caused by gas production. The data for agitated and unagitated controls are essentially identical for the first 48 h. All further work with aerobes was done with the cultures agitated. Decimal dilutions. A typical series of curves for decimal dilutions of the inocula for Pseudomonas aeruginosa grown in TSB is shown in Fig. 3, the first five curves demonstrating growth and gas absorption while in the sixth (control), the inoculum had been diluted to extinction. The circled points are where detection was assumed to occur. Table 1 shows whether there was gas absorption and production in agitated test cultures with the detection time for the smallest inoculum that gave growth, nominally 1 to 10 CFU. Cultures of anaerobic organisms were not agitated and all showed only gas production. marcescens
Comparison of detection times. For logistical reasons, it
was
possible to make this comparison for the eight slower-
VOL. 30, 1992
*1 ,-
--
DETECTION OF BACTERIAL GROWTH BY GAS ABSORPTION
-
.5
Vacuum Ipsi] 1
-
-
--
1.5
------------
2
TABLE 1. Organisms tested for their ability to absorb and produce gas
(h) Gas Produce Gas Time Absorb to posi-
Organism
F.. .. .. .. .. ... ,''* *~ ~~~~. .............
1207
tivity'
Aerobic (agitated) Acinetobacter calcoaceticus ATCC 19606 Aeromonas hydrophila ATCC 35654 Alcaligenesfaecalis ATCC 8750 Bordetella bronchiseptica ATCC 10580 Flavobacterium meningosepticum ATCC 13253 Haemophilus influenzae
+
-
16
+ + + +
-
12 22 26 25
+ + + +
-
29 24 18 17
0
[*
E 20
.
.. ..
.l _... 1.< .
.
.
.
FIG. 3. Series of reference vacuum curves for decimal dilutions of Pseudomonas aenrginosa in agitated TSB cultures. Plots 1 to 5 showed gas absorption due to growth. A sterile control (0) is also shown. The detection time of each curve is shown by the circle.
growing aerobic organisms only when the BACTEC bottles did not give positive results during the night. Figure 4 compares the performance of the new system with that of the BACTEC 460 for four of these test organisms. The time from inoculation to detection is plotted for several decimal dilutions of the four organisms. A straight line should result when cultures are growing logarithmically; the lines shown are just to guide the eye. For the test system, these correspond to doubling times of 45, 28, 46, and 29 (longer line segment) min, respectively. Figure 4A shows line results when Candida albicans was the inoculum over six decades for the test system and three decades for BACTEC 460. These results are typical for an organism that shows no difference in detection times for the two systems. The other three panels of Fig. 4 compare results for Staphylococcus epidennidis (B), Moraxella osloensis (C), and Neissenia meningitidis (D) (one point missing for the test system in Fig. 4C and D because of nipple leakage or contamination). In each of these, detection by gas absorption occurred before detection by BACTEC 460. The other four slowly growing organisms were Alcaligenes faecalis, Bordetella bronchiseptica, Pseudomonas aeruginosa, and Flavobactenium meningosepticum, with the latter two showing slightly slower detection by the BACTEC. Sterile bottles containing blood. A total of 28 blood samples were tested of which two were contaminated, one showing growth after 24 h and one after 8 days (Propionibacterium acnes recovered). The reference vacuum curves were smooth and very similar to those of the uninoculated TSB controls; it was impossible to distinguish one from the other unless they were labelled. There were no sudden breaks or jumps that might be considered false positives. There was
Moraxella osloensis ATCC 10973 Neisseria meningitidis ATCC 13077 Pseudomonas aeruginosa ATCC 27853 Staphylococcus epidermidis ATCC 12228 Streptococcus agalactiae ATCC 12386 Streptococcus group D ATCC 27284
+
-
+
-
15 14
Streptococcus pneumoniae ATCC 27336 Streptococcus pyogenes ATCC 19615 Streptococcus sanguis ATCC 10556 Xanthomonas maltophilia ATCC 13637 Bacillus subtilis ATCC 6633 Candida albicans ATCC 60193
+
-
14
+
+ +
-
+ +
+ +
10 18 14 12 25
Enterobacter cloacae ATCC 23355 Escherichia coli ATCC 25922 Klebsiella pneumoniae ATCC 13883 Serratia marcescens ATCC 8100 Staphylococcus aureus ATCC 25923 Streptococcus group C ATCC 12388
+ + + + + +
+ + + + + +
14 10 11 12 14 15
-
+ +
-
+
-
+
28 32 17 40
-
+
114
-
+
28
Anaerobic (unagitated) Bacteroides fragilis Clostridium histolyticum ATCC 19401 Clostridium ramosum ATCC 25582 Fusobacterium nucleatum ATCC 25586 Peptostreptococcus anaerobius ATCC 27337 Peptococcus asaccharolyticus ATCC 29743
a Number of hours before the highest dilution bottle showing growth was detected.
slightly more gas absorption (curves drifted towards higher vacuum) over the whole 7 days, 0.5 vs 0.2 lb/in2 for controls. DISCUSSION In a sealed vial, oxygen from the headspace gas dissolves in the fluid culture medium until equilibrium is established. An organism growing in the culture medium can absorb oxygen only from its surrounding fluid so oxygen must diffuse from the headspace across the gas-liquid interface to replenish that used. This diffusion is a slow process limited by the surface area of the gas-liquid interface. Agitation, by continuously exposing a new surface for diffusion, accelerates oxygen absorption. Our results showed that in the absence of agitation, oxygen absorption from the headspace gas was so slow that it was overtaken by the production of gas, presumably carbon dioxide, from the culture. Hence,
J. CLIN. MICROBIOL.
WATERS
1208 1
EI
2-
.
L..J
\.
\.
0~~~~~U
4.
C
S-
d Os
i
'
20
.2
X
'a E 1-
3.
*\ .
\
E1 .
.
10
15
Di *
.l
4.
0
6' 30
I
do
I-
1o
I
2b
,
30
Hours
FIG. 4. Comparison of detection times by gas absorption (@) and by BACTEC 460 (-) for decimal dilutions of Candida albicans (A), Staphylococcus epidermidis (B), Moraxella osloiensis (C), and Neisseria meningitidis (D).
agitation of the fluid in the culture vials is inecessary for our system to function optimally. Since it has been well established (e.g. Weinstein et al. [12]) that algitation of blood cultures gives faster detection of positive cultures, it is a mandatory requirement anyway. Ahnell (1) studied the effect on gas pres sure in the headspace gas when oxygen-consuming bacteria were cultured in agitated medium and noted an initial abso rption of gas for several organisms. Ahnell also showed logarithmic growth curves similar to those in Fig. 4. He found a strain of Streptococcus bovis that did not show gas absorption, just copious gas production; we did not observe such behavior. Ahnell's result may have been an artifact caused by the very large initial inoculum of 0.5 ml from an overnight culture (more than ten million times greater than that seen in blood cultures); the comparison BACTEC 460 bottle with the same inoculum gave a positive reading in only 3 h. Ahnell noted that it has long been known that many bacteria consume oxygen, but it has been thought that no vacuum would be produced because the oxygen was replaced by evolved carbon dioxide. He quotes Arthur's (2) comment that "In a closed system, as bacteria utilize oxygen and create carbon dioxide, there is no effective change in pressure." Thus, a carbon dioxide absorber is always included in any system, e.g., a Warburg apparatus (9), that attempts to measure oxygen consumption. BACTEC 460 is still widely used, although no longer manufactured, and has been shown to have as good (5) or better (13) performance than the newer nonradiometric BACTEC systems. It is just less convenient to use. In our comparison with BACTEC 460, in no case was BACTEC faster than gas detection. Most organisms, however, were detected at about the same time in both systems. Of those
that were slower in BACTEC, some are known to present problems to the radiometric system. For example, Moracxella osloensis gave low Growth Index values; in fact, the culture bottles were visibly turbid 10 to 20 h before detection occurred. It is often difficult to induce growing organisms, particularly those that do not ferment glucose, to produce the carbon dioxide that these systems need for detection. The culture medium requires special formulation with the addition of complex amino acids and carbohydrate substrates tailored for specific problem organisms. Bopp and Ellner (3) illustrate this problem for the radiometric system. The newer nonradiometric systems have the same problem, although not the same limitation; even so, their overall performance has not improved (13). Other organisms, such as Staphylococcus epidennidis grow slowly in BACTEC (6) and other (8) systems. This may be related to the medium, since radiometric detection occurred at the same time as visible turbidity. Anaerobic organisms were grown in unshaken cultures, and all six were detected by gas production. Shaken cultures were not tested. In summary, detection of bacteria in blood cultures by gas absorption is attractive because of the apparent universality of this phenomenon for organisms growing aerobically. Optimization of medium components for specific nonfermentative organisms is not necessary. Moreover, should an organism be found that does not absorb gas it can be induced to produce gas and so be detected by our instrument, just as it is now by existing blood culture systems. The new detection system is simple, noninvasive, and inexpensive and can be used on large culture bottles that give a full 1:10 dilution of the blood sample. Small bottles inadequately dilute the blood sample. It has been shown (7) that using a dilution of 1:4 and not using resins (lla) resulted in failure to detect positive cultures from patients on antibiotic therapy. A limitation of the system described here is that the cap must be changed after inoculation because of the inability of the nipple to withstand autoclaving. A later development has replaced the nipple with a bellows-shaped diaphragm which protrudes above the cap and can be autoclaved without rupture. REFERENCES 1. Ahnell, J. E. May 1979. U.S. patent 4,152,213. 2. Arthur, R. M. June 1973. U.S. patent 3,740,320. 3. Bopp, H., and P. D. Ellner. 1988. Evaluation of substrates for radiometric detection of bacteria in blood cultures. J. Clin. Microbiol. 26:919-922. 4. DeLand, F. H., and H. N. Wagner, Jr. 1970. Automated radiometric detection of bacterial growth in blood cultures. J. Lab. Clin. Med. 75:529-534. 5. Jungkind, D., J. Millan, S. Allen, J. Dyke, and E. Hill. 1986. Clinical comparison of a new automated infrared blood culture system with the BACTEC 460 system. J. Clin. Microbiol. 23:262-266. 6. Kurlat, I., B. J. Stoll, and J. E. McGowan, Jr. 1989. Time to positivity for detection of bacteremia in neonates. J. Clin. Microbiol. 27:1068-1071. 7. Salventi, J. F., T. A. Davies, E. L. Randall, S. Whitaker, and J. R. Waters. 1979. Effect of blood dilution on recovery of organisms from clinical blood cultures in medium containing sodium polyanethol sulfonate. J. Clin. Microbiol. 9:248252. 8. Thorpe, T. C., M. L. Wilson, J. E. Turner, J. L. DiGuiseppi, M. Willert, S. Mirrett, and B. Reller. 1990. BacT/Alert: an automated colorimetric microbial detection system. J. Clin. Microbiol. 28:1608-1612. 9. Umbreit, W. W., R. H. Burris, and J. F. Stauffer. 1972.
VOL. 30, 1992
DETECTION OF BACTERIAL GROWTH BY GAS ABSORPTION
Manometric and biochemical techniques, 5th ed., p. 9-13. Burgess Publishing Co., Minneapolis. 10. Waters, J. R. July 1972. U.S. patent 3,676,679. 11. Waters, J. R. September 1991. U.S. patent 5,051,360. 11a.Waters, J. R., and R. Brotman. December 1986. U.S. patent 4,632,902. 12. Weinstein, M. P., S. Mirrett, L. G. Reimer, and L. B. Reller. 1989. Effect of agitation and terminal subcultures on yield and
1209
speed of detection of the Oxoid Signal blood culture system the BACTEC radiometric system. J. Clin. Microbiol. 27:427-430. 13. Zimmerman, S. J., N. Sofat, N. Gunnersen, and D. Amsterdam. 1989. Microbial kinetics of seeded blood cultures detected by radiometric and infrared spectroscopic systems, abstr. C-222, p. 430. Abstr. 89th Annu. Meet. Am. Soc. Microbiol. 1989. American Society for Microbiology, Washington, D.C. versus
Vol. 30, No. 5
Detection of Bacterial Growth by Gas Absorption JOHN R. WATERS Abo, Inc., P.O. Box 271789, Tampa, Florida 33688 Received 25 November 1991/Accepted 13 February 1992
When 24 different aerobic organisms were grown in a shaken culture, all were found to first absorb gas from the headspace. In a rudimentary medium, such as tryptic soy broth, 16 of the 24 organisms did not produce gas following the initial gas absorption. We have developed a simple, noninvasive method for detecting both gas absorption and production in multiple culture vials. The time to positivity was compared with that obtained by the BACTEC 460 blood culture system. For nearly all of these organisms, there was no difference. For some of those organisms that did not produce gas, e.g. Staphylococcus epidermidis, MoraxeUla osloensis, and Neisseria meningitidis, detection by gas absorption was a few hours faster. Gas absorption appears to be a promising technique for a new automated blood culture system because of its simplicity and because medium without special additives can be used to detect organisms that do not produce gas.
Since no organisms are recovered from 85 to 90% of clinical blood cultures, much laboratory time and effort is wasted processing unproductive specimens. This has led to the development of instruments to monitor a large number of culture vials for growth, reporting the few suspicious ones for further investigation. Although several screening techniques have been tried, the only one in widespread use detects the production of carbon dioxide from organisms growing in culture medium. These are the BACTEC systems (4), originally from Johnston Laboratories, which detect either radioactive (10) (BACTEC 225, 301, and 460) or nonradioactive (5) (BACTEC 660 and 730) carbon dioxide. Recently, the BacT/Alert system from Organon Teknika (8), which also detects carbon dioxide, has come into use in several laboratories. In studying the broader mechanism of gas production by growing organisms, we noticed that all the aerobic organisms we tested, when grown in agitated fluid medium, first absorbed gas from their ambient atmosphere. Some later produced gas, but many, when grown in a rudimentary medium without special additives, did not. We developed a simple measuring system (11) that can monitor this gas absorption. If this phenomenon turns out to be as universal as our results indicate, detection of aerobic microorganisms by gas absorption promises to be a powerful technique for a new automated blood culture system.
dium (Carr-Scarborough) was used for initial culture and as a source tube. Test inocula. Serial hundredfold dilutions were made in five tubes, with each containing 5 ml of TSB. From four tubes, starting with the second, eight test bottles were set up with decimal dilutions of the inocula. No growth was usually seen in the last two or three bottles, the inoculum having been diluted to extinction. For anaerobes, dilutions were made in sealed bottles containing 30 ml of prereduced anaerobically sterilized medium. Test procedure. Screw-capped bottles (130-ml overflow volume) were filled with enough TSB to leave 16 to 20 ml of headspace above the liquid, capped, and autoclaved. When cool, the cap was removed and the test inoculum was added. For Haemophilus influenzae, 2% Fildes enrichment (Difco Laboratories) was added. After inoculation, the cap was replaced with another that had a limp, flexible, cone-shaped rubber nipple pointing down into the bottle as shown in Fig. 1. On incubation, the air in the bottle expanded, causing the nipple to rise to be about level with the bottle cap. A sensor head, containing a light-emitting diode (LED) and a light detector arranged to form a horizontal light beam about 0.5 in. (-1.3 cm) above the top of the bottle cap, was put on each bottle. Fifteen test bottles could be put into the test chamber which was mounted on a rotary agitator (180 rpm, 3/8-in. [0.95-cm] stroke) in the incubator set at 36 to
370C. Every 20 min, the shaking was stopped, the LEDs were turned on, and a small vacuum pump slowly reduced the air pressure in the test chamber from ambient (14.7 lb/in2 absolute) to produce a vacuum of about 2.2 lb/in2 gauge (i.e. to 12.5 lb/in2 absolute) in 3 to 4 min. Meanwhile, the computer attached to the sensor heads rapidly scanned the output from each of the light detectors to determine whether a light beam had been broken by a rising nipple, since these gradually became erect as the test chamber pressure was reduced. The computer recorded the vacuum reading in the chamber at that instant. We call this the reference vacuum (11). If the nipple had no stiffness, the gas pressure inside the bottle would equal that in the test chamber. At the end of the test cycle, the chamber was vented, the LEDs were turned off, and the shaker turned on. Changes in the volume of the gas in the bottle caused by gas production or absorption can be detected by monitoring the history of the reference vacua, plotted by the computer,
MATERUILS AND METHODS
Microorganisms. Test organisms were usually ATCC strains in the form of Q-Chek Cultiloops (Chrisope Technologies [formerly Scott Laboratories], Sulphur, La.). These are individually wrapped, disposable, inoculating loops containing stabilized viable microorganisms for direct inoculation. Other organisms were clinical isolates. Preference was given to test organisms which were nonfermenters, thus difficult to detect by gas production, and for their relevance in blood cultures. Table 1 lists the 24 aerobic and 6 anaerobic organisms tested. For aerobic organisms, a source tube was made by breaking off the loop in a tube containing 5 ml of tryptic soy broth (TSB) (Carr-Scarborough Microbiologicals, Stone Mountain, Ga.) which was then incubated for 18 to 42 h until heavy growth was seen. For anaerobes, a tube of prereduced anaerobically sterilized chopped meat me1205
1206
J. CLIN. MICROBIOL.
WATERS G
Vacuum Ipsil
P
1.5
.5
.1
,.. ........
2
...
. k.
.*
.,
.
-.*1~~~~~~W. A~
. 0
. . . AA
E F
FIG. 1. Schematic view of culture bottle (B) showing the nipple (N) and retaining cap (C). The LED light source (S) shines a beam to the detector (D). This is interrupted when the nipple becomes erect as a result of the action of the vacuum pump (P). The pump and an electronic vacuum gauge (G) are attached to the sealed outer incubator.
for each bottle. The detection time is taken when there is a significant excursion from the normal smooth curve for a sterile bottle. This detection time is marked with a circle in later figures showing typical results. For anaerobic cultures, sterile bottles were flushed with BACTEC anaerobic gas flowing through a cannula. While still flushing, anaerobic blood culture medium, e.g. Thiol broth (Difco), was aseptically poured into the bottle, the inoculum was added, and a sterile nipple and cap were applied. BACTEC 460 procedure. Growth and detection times for some of the slower-growing organisms were compared with those for the BACTEC 460 (Becton Dickinson Diagnostic Instrument Systems, Loveton, Md.). Aerobic cultures were tested in 6B soybean-casein digest broth-aerobic and anaerobes were tested in 7D tryptic soy broth-anaerobic. BACTEC bottles were inoculated in parallel with the test bottles, although not at all dilutions. Room air was used as the aerobic culture gas, and a mixture of 85% nitrogen, 10% hydrogen, and 5% carbon dioxide was used for anaerobes. Bottles were agitated on the same shaker and tested every 3 to 4 h during the work day only. Growth Index values were plotted and interpolated to determine the BACTEC detection time corresponding to a Growth Index of 30 (all aerobes, except 20 for one or two nonfermenters) or 20 (anaerobes). Sterile bottles containing blood. Blood from healthy volunteers was drawn into sterile yellow-top tubes (BD Vacutainer 4960) containing 1.7 ml of 0.85% sodium chloride with 0.35% sodium polyanethol sulfonate (SPS) anticoagulant, for a total volume of 10 ml. Test bottles with 30 ml of headspace were inoculated by pouring the tube's contents into the bottle. Test bottles, along with uninoculated normal TSB controls, were tested every 30 min for at least 7 days.
RESULTS Effect of agitation. There was a dramatic change in the shape of the reference vacuum curves when the cultures were agitated. Figure 2 shows the results for an uninoculated
.~
zi_
.
A
A
2C
.S i.
.S
FIG. 2. Effect of agitation of cultures of Serratia marcescens on the reference vacuum curves. Symbols: *, agitated culture; A, unagitated culture; *, control.
control and two cultures with equal inocula of Serratia with and without agitation. The abscissa (x axis) is the reference vacuum with zero vacuum (i.e., atmospheric pressure) at the left and increasing vacuum to the right. The ordinate is elapsed time running from top to bottom of the page. To better demonstrate changes, the three traces have been moved laterally to overlap which is permissible because the absolute value of the vacuum is irrelevant. Detection of bacterial growth is determined by observation of a deviation from the smooth curve generated by a culture with no growth. The unagitated culture shows a decrease in vacuum after 18 h, caused by gas production. In the agitated culture, in contrast, there is a rapid increase in the reference vacuum, caused by gas absorption, after 9 h, and 3 h later, a very rapid decrease caused by gas production. The data for agitated and unagitated controls are essentially identical for the first 48 h. All further work with aerobes was done with the cultures agitated. Decimal dilutions. A typical series of curves for decimal dilutions of the inocula for Pseudomonas aeruginosa grown in TSB is shown in Fig. 3, the first five curves demonstrating growth and gas absorption while in the sixth (control), the inoculum had been diluted to extinction. The circled points are where detection was assumed to occur. Table 1 shows whether there was gas absorption and production in agitated test cultures with the detection time for the smallest inoculum that gave growth, nominally 1 to 10 CFU. Cultures of anaerobic organisms were not agitated and all showed only gas production. marcescens
Comparison of detection times. For logistical reasons, it
was
possible to make this comparison for the eight slower-
VOL. 30, 1992
*1 ,-
--
DETECTION OF BACTERIAL GROWTH BY GAS ABSORPTION
-
.5
Vacuum Ipsi] 1
-
-
--
1.5
------------
2
TABLE 1. Organisms tested for their ability to absorb and produce gas
(h) Gas Produce Gas Time Absorb to posi-
Organism
F.. .. .. .. .. ... ,''* *~ ~~~~. .............
1207
tivity'
Aerobic (agitated) Acinetobacter calcoaceticus ATCC 19606 Aeromonas hydrophila ATCC 35654 Alcaligenesfaecalis ATCC 8750 Bordetella bronchiseptica ATCC 10580 Flavobacterium meningosepticum ATCC 13253 Haemophilus influenzae
+
-
16
+ + + +
-
12 22 26 25
+ + + +
-
29 24 18 17
0
[*
E 20
.
.. ..
.l _... 1.< .
.
.
.
FIG. 3. Series of reference vacuum curves for decimal dilutions of Pseudomonas aenrginosa in agitated TSB cultures. Plots 1 to 5 showed gas absorption due to growth. A sterile control (0) is also shown. The detection time of each curve is shown by the circle.
growing aerobic organisms only when the BACTEC bottles did not give positive results during the night. Figure 4 compares the performance of the new system with that of the BACTEC 460 for four of these test organisms. The time from inoculation to detection is plotted for several decimal dilutions of the four organisms. A straight line should result when cultures are growing logarithmically; the lines shown are just to guide the eye. For the test system, these correspond to doubling times of 45, 28, 46, and 29 (longer line segment) min, respectively. Figure 4A shows line results when Candida albicans was the inoculum over six decades for the test system and three decades for BACTEC 460. These results are typical for an organism that shows no difference in detection times for the two systems. The other three panels of Fig. 4 compare results for Staphylococcus epidennidis (B), Moraxella osloensis (C), and Neissenia meningitidis (D) (one point missing for the test system in Fig. 4C and D because of nipple leakage or contamination). In each of these, detection by gas absorption occurred before detection by BACTEC 460. The other four slowly growing organisms were Alcaligenes faecalis, Bordetella bronchiseptica, Pseudomonas aeruginosa, and Flavobactenium meningosepticum, with the latter two showing slightly slower detection by the BACTEC. Sterile bottles containing blood. A total of 28 blood samples were tested of which two were contaminated, one showing growth after 24 h and one after 8 days (Propionibacterium acnes recovered). The reference vacuum curves were smooth and very similar to those of the uninoculated TSB controls; it was impossible to distinguish one from the other unless they were labelled. There were no sudden breaks or jumps that might be considered false positives. There was
Moraxella osloensis ATCC 10973 Neisseria meningitidis ATCC 13077 Pseudomonas aeruginosa ATCC 27853 Staphylococcus epidermidis ATCC 12228 Streptococcus agalactiae ATCC 12386 Streptococcus group D ATCC 27284
+
-
+
-
15 14
Streptococcus pneumoniae ATCC 27336 Streptococcus pyogenes ATCC 19615 Streptococcus sanguis ATCC 10556 Xanthomonas maltophilia ATCC 13637 Bacillus subtilis ATCC 6633 Candida albicans ATCC 60193
+
-
14
+
+ +
-
+ +
+ +
10 18 14 12 25
Enterobacter cloacae ATCC 23355 Escherichia coli ATCC 25922 Klebsiella pneumoniae ATCC 13883 Serratia marcescens ATCC 8100 Staphylococcus aureus ATCC 25923 Streptococcus group C ATCC 12388
+ + + + + +
+ + + + + +
14 10 11 12 14 15
-
+ +
-
+
-
+
28 32 17 40
-
+
114
-
+
28
Anaerobic (unagitated) Bacteroides fragilis Clostridium histolyticum ATCC 19401 Clostridium ramosum ATCC 25582 Fusobacterium nucleatum ATCC 25586 Peptostreptococcus anaerobius ATCC 27337 Peptococcus asaccharolyticus ATCC 29743
a Number of hours before the highest dilution bottle showing growth was detected.
slightly more gas absorption (curves drifted towards higher vacuum) over the whole 7 days, 0.5 vs 0.2 lb/in2 for controls. DISCUSSION In a sealed vial, oxygen from the headspace gas dissolves in the fluid culture medium until equilibrium is established. An organism growing in the culture medium can absorb oxygen only from its surrounding fluid so oxygen must diffuse from the headspace across the gas-liquid interface to replenish that used. This diffusion is a slow process limited by the surface area of the gas-liquid interface. Agitation, by continuously exposing a new surface for diffusion, accelerates oxygen absorption. Our results showed that in the absence of agitation, oxygen absorption from the headspace gas was so slow that it was overtaken by the production of gas, presumably carbon dioxide, from the culture. Hence,
J. CLIN. MICROBIOL.
WATERS
1208 1
EI
2-
.
L..J
\.
\.
0~~~~~U
4.
C
S-
d Os
i
'
20
.2
X
'a E 1-
3.
*\ .
\
E1 .
.
10
15
Di *
.l
4.
0
6' 30
I
do
I-
1o
I
2b
,
30
Hours
FIG. 4. Comparison of detection times by gas absorption (@) and by BACTEC 460 (-) for decimal dilutions of Candida albicans (A), Staphylococcus epidermidis (B), Moraxella osloiensis (C), and Neisseria meningitidis (D).
agitation of the fluid in the culture vials is inecessary for our system to function optimally. Since it has been well established (e.g. Weinstein et al. [12]) that algitation of blood cultures gives faster detection of positive cultures, it is a mandatory requirement anyway. Ahnell (1) studied the effect on gas pres sure in the headspace gas when oxygen-consuming bacteria were cultured in agitated medium and noted an initial abso rption of gas for several organisms. Ahnell also showed logarithmic growth curves similar to those in Fig. 4. He found a strain of Streptococcus bovis that did not show gas absorption, just copious gas production; we did not observe such behavior. Ahnell's result may have been an artifact caused by the very large initial inoculum of 0.5 ml from an overnight culture (more than ten million times greater than that seen in blood cultures); the comparison BACTEC 460 bottle with the same inoculum gave a positive reading in only 3 h. Ahnell noted that it has long been known that many bacteria consume oxygen, but it has been thought that no vacuum would be produced because the oxygen was replaced by evolved carbon dioxide. He quotes Arthur's (2) comment that "In a closed system, as bacteria utilize oxygen and create carbon dioxide, there is no effective change in pressure." Thus, a carbon dioxide absorber is always included in any system, e.g., a Warburg apparatus (9), that attempts to measure oxygen consumption. BACTEC 460 is still widely used, although no longer manufactured, and has been shown to have as good (5) or better (13) performance than the newer nonradiometric BACTEC systems. It is just less convenient to use. In our comparison with BACTEC 460, in no case was BACTEC faster than gas detection. Most organisms, however, were detected at about the same time in both systems. Of those
that were slower in BACTEC, some are known to present problems to the radiometric system. For example, Moracxella osloensis gave low Growth Index values; in fact, the culture bottles were visibly turbid 10 to 20 h before detection occurred. It is often difficult to induce growing organisms, particularly those that do not ferment glucose, to produce the carbon dioxide that these systems need for detection. The culture medium requires special formulation with the addition of complex amino acids and carbohydrate substrates tailored for specific problem organisms. Bopp and Ellner (3) illustrate this problem for the radiometric system. The newer nonradiometric systems have the same problem, although not the same limitation; even so, their overall performance has not improved (13). Other organisms, such as Staphylococcus epidennidis grow slowly in BACTEC (6) and other (8) systems. This may be related to the medium, since radiometric detection occurred at the same time as visible turbidity. Anaerobic organisms were grown in unshaken cultures, and all six were detected by gas production. Shaken cultures were not tested. In summary, detection of bacteria in blood cultures by gas absorption is attractive because of the apparent universality of this phenomenon for organisms growing aerobically. Optimization of medium components for specific nonfermentative organisms is not necessary. Moreover, should an organism be found that does not absorb gas it can be induced to produce gas and so be detected by our instrument, just as it is now by existing blood culture systems. The new detection system is simple, noninvasive, and inexpensive and can be used on large culture bottles that give a full 1:10 dilution of the blood sample. Small bottles inadequately dilute the blood sample. It has been shown (7) that using a dilution of 1:4 and not using resins (lla) resulted in failure to detect positive cultures from patients on antibiotic therapy. A limitation of the system described here is that the cap must be changed after inoculation because of the inability of the nipple to withstand autoclaving. A later development has replaced the nipple with a bellows-shaped diaphragm which protrudes above the cap and can be autoclaved without rupture. REFERENCES 1. Ahnell, J. E. May 1979. U.S. patent 4,152,213. 2. Arthur, R. M. June 1973. U.S. patent 3,740,320. 3. Bopp, H., and P. D. Ellner. 1988. Evaluation of substrates for radiometric detection of bacteria in blood cultures. J. Clin. Microbiol. 26:919-922. 4. DeLand, F. H., and H. N. Wagner, Jr. 1970. Automated radiometric detection of bacterial growth in blood cultures. J. Lab. Clin. Med. 75:529-534. 5. Jungkind, D., J. Millan, S. Allen, J. Dyke, and E. Hill. 1986. Clinical comparison of a new automated infrared blood culture system with the BACTEC 460 system. J. Clin. Microbiol. 23:262-266. 6. Kurlat, I., B. J. Stoll, and J. E. McGowan, Jr. 1989. Time to positivity for detection of bacteremia in neonates. J. Clin. Microbiol. 27:1068-1071. 7. Salventi, J. F., T. A. Davies, E. L. Randall, S. Whitaker, and J. R. Waters. 1979. Effect of blood dilution on recovery of organisms from clinical blood cultures in medium containing sodium polyanethol sulfonate. J. Clin. Microbiol. 9:248252. 8. Thorpe, T. C., M. L. Wilson, J. E. Turner, J. L. DiGuiseppi, M. Willert, S. Mirrett, and B. Reller. 1990. BacT/Alert: an automated colorimetric microbial detection system. J. Clin. Microbiol. 28:1608-1612. 9. Umbreit, W. W., R. H. Burris, and J. F. Stauffer. 1972.
VOL. 30, 1992
DETECTION OF BACTERIAL GROWTH BY GAS ABSORPTION
Manometric and biochemical techniques, 5th ed., p. 9-13. Burgess Publishing Co., Minneapolis. 10. Waters, J. R. July 1972. U.S. patent 3,676,679. 11. Waters, J. R. September 1991. U.S. patent 5,051,360. 11a.Waters, J. R., and R. Brotman. December 1986. U.S. patent 4,632,902. 12. Weinstein, M. P., S. Mirrett, L. G. Reimer, and L. B. Reller. 1989. Effect of agitation and terminal subcultures on yield and
1209
speed of detection of the Oxoid Signal blood culture system the BACTEC radiometric system. J. Clin. Microbiol. 27:427-430. 13. Zimmerman, S. J., N. Sofat, N. Gunnersen, and D. Amsterdam. 1989. Microbial kinetics of seeded blood cultures detected by radiometric and infrared spectroscopic systems, abstr. C-222, p. 430. Abstr. 89th Annu. Meet. Am. Soc. Microbiol. 1989. American Society for Microbiology, Washington, D.C. versus
