Annals of Clinical & Laboratory Science, vol. 33, no. 2, 2003 179 Microscopic Morphology in Smears Prepared from MGIT Broth Medium for Rapid Presumptive Identification of Mycobacterium tuberculosis complex, Mycobacterium avium complex and Mycobacterium kansasii Hui-Zin Tu, 1 Shu-Huei Chang, 1 Tsi-Shu Huaug, 1,3 Wen-Kuei Huaug, 1,3 Yung-Ching Liu, 1,2 and Susan Shin-Jung Lee 1,2 Section of Microbiology 1 and Infectious Diseases, 2 Kaohsiung Veterans General Hospital, and Department of Medical Technology, 3 Foo-Yin Institute of Technology, Kaohsiung, Taiwan Abstract. Mycobacterium species has a specific morphology when grown in liquid medium. Mycobacterium tuberculosis complex (MTB) often exhibits serpentine cording, which is different from the dot and crossbarring morphology observed in Mycobacterium avium complex (MAC) and Mycobacterium kansasii (MK), respectively. These characteristic morphologies can be used as a cost-effective method for rapid, presumptive identification of mycobacterial isolates cultured from the MGIT 960 system. By using Kinyoun acid-fast stain, serpentine cording was found in 840 of 904 (92.1%) samples positive for MTB; dot or loose aggregation was observed in 112 of 136 (82.3%) samples positive for MAC; and the cross-barring, ladder-like, morphology was observed in 45 of 56 (80.5%) samples positive for MK. The sensitivity and specificity were 92.9% and 96.4% for MTB; 82.4% and 94.5% for MAC; and 80.4% and 94.6 % for MK, respectively. Using growth rate selection to exclude rapid growers, the positive and negative predictive values were 98% and 87.6% for MTB; 78.3% and 98% for MAC; and 78.9% and 99.1% for MK, respectively. Twenty-eight (93.3%) of 30 strains with ball morphology were rapid growers. Microscopic morphology can be used for rapid, presumptive identification of M. tuberculosis complex, M. kansasii, and M. avium complex and act as a guide for appropriate selection of initial probes to reduce costs. (received 15 November 2002; accepted 26 January 2003) Keywords: mycobacteria, microscopic morphology Introduction Address correspondence to Susan Shin-Jung Lee, M.D., Section of Infectious Diseases, Kaohsiung Veterans General Hospital, 386 Ta-Chung 1st Rd, Kaohsiung, Taiwan, ROC; tel 886 7 346 8169;fax 886 7 346 8296; e-mail ssjlee@isca.vghks.gov.tw. Rapid, specific, and inexpensive detection of Mycobacterium tuberculosis complex, Mycobacterium avium complex, and Mycobacterium kansasii from clinical specimens is an important concern, especially in developing countries with a high incidence of mycobacterial disease and scarce economic resources. The distinctive morphology that some mycobacteria may exhibit during growth in liquid culture has been previously noted [1-5]. M. tuberculosis complex (MTB) grows as tight, rope-like aggregates of acidfast bacilli in which the long axes of the bacteria parallel the long axes of the cord, due to the presence of a cord factor [6]. Several papers have shown that M.tuberculosis complex (MTB) has the ability to form serpentine cording in liquid medium with a sensitivity of 23 to 95% and a specificity of 95 to 100% [1-5]. MAC has been described as dot-needlelike and loose aggregates, while MK has been described as broad rods that exhibit marked crossbarring likened to ladders. These characteristics are not very specific, but, with experience, permit one to distinguish MTB, MAC, and MK from the most common clinical mycobacterial species. This study assessed the feasibility of a simple method, based on morphology in smears prepared from culture broth, for presumptive identification of mycobacteria commonly encountered in clinical specimens. 0091-7370/03/0200/0177 $1.25 2003 by the Association of Clinical Scientists, Inc.
180 Annals of Clinical & Laboratory Science Materials and Methods Specimens. A total of 11,328 specimens were collected and processed in the Mycobacteriology Laboratory of Kaohsiung Veterans General Hospital from Feb to Dec 2001. The specimens included 7604 sputum samples, 280 bronchial washing samples, 385 gastric lavage samples, 2198 body fluid samples, 358 urine samples, 208 tissue samples, and 295 pus/wound samples. All of the specimens were digested and decontaminated with N-acetyl-Lcysteine and 3%NaOH, except urine specimens, which were decontaminated with 4% sulfuric acid, at 35 C for 20 min. The reaction was stopped by addition of phosphate buffer (ph 6.8) and the specimens were centrifuged at 3000 x g for 15 min. Specimens from normally sterile sites were inoculated directly into the media. Each specimen was inoculated into a MGIT tube; pus or wound specimens were also inoculated onto a L-J agar slant. Culture. During incubation at 37 C in a BACTEC MGIT 960 instrument, every tube was monitored continuously for positive growth and reported as negative after 42 days. The BACTEC MGIT 960 instrument automatically monitors for an increase of fluorescence every 60 min. Algorithms enable the detection of a presumptive positive result and alert the operator to the locations of positive tubes by means of indicator lights on the front of the instrument. The status of all tubes is displayed on the liquid crystal display (LCD). Each sample that was identified as positive was removed from the BACTEC MGIT 960 instrument and examined by the Kinyoun acid-fast staining technique. Microscopic morphology. Seven morphological patterns in acid-fast smears were commonly observed and defined by microscopy at a magnification of x1000 (except x 100 for ball morphology) (Fig. 1). Cording (Fig. 1a): 2-3 µm long, forming serpentine cords defined as rope-like aggregates in which the long axis of the bacteria is parallel to the long axis of the cord [3]. Dot (Fig. 1b): irregular short bacilli, 1-2 µm, coexisting with coccoid forms, arranged as single, dispersed cells or in small clumps. Needle (Fig. 1c): 2-3 µm bacilli, gram-negative bacilli-like, dispersed or forming irregular clumps. Ladder (Fig. 1d): broad, long (3-5 µm) bacilli, staining in characteristic cross-banding appearance, forming clumps resembling Chinese characters; sometimes incomplete cords [1]. Ball (Fig. 1e): long (3-5 µm) bacilli, aggregated into balls (magnification, x100). Short (Fig. 1f): regular long (3-5 µm) bacilli, Corynebacteria-like, forming clumps occurring in dispersed or palisade arrangements. Mixed: more than one morphological pattern. Identification. The BDProTecET system (Becton Dickinson Diagnostic Systems, Baltimore, MD) was used to identify mycobacterial isolates. Nontuberculous mycobacteria were grouped by the growth rate test. Appropriate kits (BDProTec MAC or KAN) were selected for slow growers based on microscopic morphology to detect M. kansasii and M. avium complex. If results were negative, further identification was performed using another kit. Results There were 1,372 (12.3%) specimens that were positive for mycobacteria from 736 patients, including 1,259 respiratory secretions (91.8%), 70 body fluids (5.0%), 11 urine specimens (0.8%), 26 tissue specimens (1.9%), and 6 pus specimens (0.4%). There were 904 (65.9%) MTB and 468 (34.1%) NTM (Table 1). Of 286 slow growers, 192 strains were MK and MAC, which were rapidly detected by BD ProbeTec kits. The microscopic morphology of acid-fast smears prepared from MGIT positive tubes were grouped into 7 types and the distribution of each type in Table 1. The positive culture rate of mycobacteria. Culture No. (%) MTB 904 (65.9%) NTM 468 (34.1%) MAC 136 (9.9%) MK 56 (4.1%) Rapid grower 182 (13.2%) Other slow grower 94 (6.9%)
Microscopic morphology of mycobacteria 181 Fig. 1. Morphological patterns, defined as a) cording, b) dot, c) needle, d) ladder, e) ball, and f) short patterns, (Kinyoun acid-fast stain; magnification x1000, except for ball pattern, x100.)
182 Annals of Clinical & Laboratory Science Table 2. The microscopic morphology and identification results. No. (%) of isolates of: TB MAC MK Rapid growers Other slow growers Morphology no. (%) 904 (65.9%) 136 (9.9%) 56 (4.1%) 182 (13.3%) 94 (6.9%) Cording 857 (62.4%) 840 (92.9%) 0 1 (1.8%) 12 (6.6%) 4 (4.3%) Dot 180 (3.1%) 3 (0.3%) 112 (82.4%) 5 (8.9%) 34 (18.7%) 26 (27.7%) Needle 120 (8.8%) 18 (2.0%) 12 (15.4%) 3 (5.4%) 40 (22.0%) 38 (40.4%) Ladder 116 (8.4%) 35 (3.9%) 1 (0.8%) 45 (80.5%) 24 (13.2%) 11 (11.7%) Ball 30 (2.2%) 0 1 (0.8%) 0 28 (15.4%) 1 (1.1%) Short 31 (2.3%) 2 (0.2%) 0 1 (1.8%) 19 (10.4%) 9 (9.6%) Mixed 38 (2.8%) 6 (0.7%) 1 (0.8%) 1 (1.8%) 25 (13.7%) 5 (5.3%) Table 3. Overall performance of presumptive identification of the mycobacteria by characteristic microscopic morphology. Morphology Presumptive Sensitivity Specificity Positive predictive Negative predictive identification (%) (%) value (%) value (%) Cording MTB 92.9 96.4 98.0 87.6 Dot MAC 82.4 94.5 78.3 98.0 Ladder MK 80.4 94.6 78.9 99.1 Table 4. Morphologic assessments of MTB, MAC, and MK. Number of isolates of: Morphology MTB (n = 904) MAC (n = 136) MK (n = 560) 1st 2nd agreement (%) 1st 2ndagreement (%) 1st 2nd agreement (%) Cording 841 840 99.9 0 0 100 1 1 100 Dot 3 3 100 105 112 93.8 5 5 100 Needle 34 35 97.1 28 21 75 4 3 100 Ladder 18 18 94.4 1 1 100 44 45 97.8 Ball 0 0 100 1 1 100 44 45 97.8 Short 2 2 100 0 0 100 1 1 100 Mixed 6 6 100 1 1 100 1 1 100 Average 98.7 95.2 96.1
Microscopic morphology of mycobacteria 183 variant mycobacteria are shown in Table 2. Cording morphology was found in 92.9% of the MTB cultures; dot and loose aggregation morphology in 82.4% of MAC cultures, and ladder morphology in 80.5% of MK cultures. The sensitivity and specificity were 92.9% and 96.4% for MTB; 82.4% and 94.5% for MAC; and 80.4% and 94.6 % for MK, respectively. The positive and negative predictive values were 98% and 87.6% for MTB; 62.2% and 98% for MAC; and 38.8% and 99.1% for MK, respectively. The positive predictive values were low for MAC and MK, but if slow growers were selected, the positive predictive values was improved to 78.3% and 78.9% for MAC and MK, respectively (Table 3). Twenty-eight of 30 strains (93.3%) with ball morphology were rapid growers. In order to avoid subjective bias in the recording of the microscopic morphology, all of the 1096 smears were assessed twice independently by one technician. The agreement between these assessments was good (98.7%, 95.2 % and 96.1% for MTB, MAC, and MK, respectively) (Table 4). Discussion Morphologic assessment has been used as a rapid, presumptive diagnostic test and a cost-effective guide to the selection of appropriate nucleic acid probes [2,6]. Our study showed high sensitivity (92.9%), specificity (96.4%), and positive predictive value (98%) when using the presence of serpentine cording on microscopy as presumptive identification of MTB positive cultures. Our results showed higher sensitivity and specificity than previous reports. This may reflect the greater number of samples in our study. Dot morphology was present in 82.4% of MAC, also higher than previously reported, and may be due to the heterogeneity and different distribution pattern of serotypes of MAC isolates in Taiwan. The ladder morphology was observed in 80.4% of MK, similar to other reports. So when BDProTecET strips are used to identify MAC and MK, the morphologic characteristics can help reduce costs by selecting 1 appropriate strip rather than 2. The possible reasons for absence of cording morphology in 7.1% of MTB may be (1) the time-to-detection was too early to form serpentine cording and (2) about 50% of non-cording MTB was recorded as ladder morphology. These results suggest difficulty in making a distinction between the 2 morphologies, especially for inexperienced medical technologists. In conclusion, MTB, MAC, and MK have characteristic morphologies that enable their differentiation from other mycobacteria when recovered from broth medium. The high positive predictive value of cording morphology for M. tuberculosis complex allows early presumptive reporting to clinicians. The initial selection of a probe for testing of primary BACTEC MGIT 960 cultures on the basis of microscopic morphology in smears can offer a rapid, reliable and cost-effective approach to the laboratory diagnosis of M. tuberculosis complex, M. avium complex, and M. kansasii. Acknowledgement This study was supported by Kaoshiung Veterans General Hospital grant VGHKS 89-31. References 1. Gonzalez J, Tudo G, Gomez J, et al. Use of microscopic morphology in smears prepared from radiometric cultures for presumptive identification of Mycobacterium tuberculosis complex, Mycobacterium avium complex, Mycobacterium kansasii and Mycobacterium xenopi. Eur J Clin Microbiol Infect Dis 1998;17:493-500. 2. Kaminski DA, Hardy DJ. Selective utilization of DNA probes for identification of Mycobacterium species on the basis of cord formation in primary BACTEC 12B cultures. J Clin Microbiol 1995;33:1548-1550. 3. McCarter YS, Ratkiewicz IN, Robinson A. Cord formation in BACTEC medium is a reliable, rapid method for presumptive identification of Mycobacterium tuberculosis complex. J Clin Microbiol 1998;36:2769-2771. 4. Morris AJ, Reller LB. Reliability of cord formation in BACTEC media for presumptive identification of mycobacteria. J Clin Microbiol 1993;31:2533-2534. 5. Silvia A, Sherry D, Jille C III. Assessment of morphology rapid for presumptive identification of Mycobacterium tuberculosis and Mycobacterium kansasii. J Clin Microbiol 2000;38:1426-1429. 6. Nelson SM, Cartwright CP. Comparison of algorithms for selective use of nucleic-acid probes for identification of Mycobacterium tuberculosis from BACTEC 12B bottles. Diagn Microbiol Infect Dis 1998;31:537-541.