INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Pulmonary tuberculosis is usually considered to be definitively diagnosed when Mycobacterium tuberculosis is recovered from in vitro culture of respiratory specimens. Previous studies have demonstrated incrementally higher recovery rates of M. tuberculosis with each successive sputum specimen submitted (3, 9-11). It has been common cost-effective clinical practice, however, to limit the laboratory evaluation for tuberculosis by submitting only three expectorated sputa obtained on three successive days for in vitro culture and acid-fast smear microscopy.

The automated BACTEC 460 radiometric mycobacterial broth culture system was introduced into clinical practice in the 1980s and was in use at 37% of hospital-based clinical mycobacteriology laboratories surveyed in 1995 (19). Numerous comparative studies demonstrated that the BACTEC system, when used as the primary mycobacterial culture system (in conjunction with conventional solid media culture), detected mycobacterial growth sooner and more often than conventional methods (2, 6, 8, 13, 15-17, 20). The labor savings, increased sensitivity, and faster turnaround time of mycobacterial cultures using the BACTEC system led us to implement it as our primary mycobacterial broth culture system in 1992.

The increased sensitivity that the BACTEC system added to conventional methods for the in vitro recovery of M. tuberculosis led us to question whether a definitive laboratory diagnosis of tuberculosis could be established with fewer sputum specimens than the customary three. We also wanted to determine if fewer specimens were necessary to establish a definitive diagnosis of pulmonary tuberculosis in individuals infected with the human immunodeficiency virus (HIV), given that a 100% diagnostic yield (for both smear and culture) from just two respiratory specimens has been previously demonstrated (although it was not explicitly stated if a BACTEC radiometric culture system was used) (5). Finally, we wanted to compare the diagnostic yields of induced versus expectorated sputa for the diagnosis of pulmonary tuberculosis in our setting. To address these questions, we performed a retrospective analysis of culture results from sequential specimens obtained from 143 patients from which M. tuberculosis had been recovered during 1994 to 1996, a period during which the BACTEC system was firmly established within our mycobacteriology laboratory. We also evaluated the diagnostic yield of acid-fast smear microscopy in this same cohort.

(This work was presented in part at the 37th Annual Meeting of the Interscience Conference on Antimicrobial Agents and Chemotherapy, Toronto, Canada, 1997.)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Study design. A total of 143 patients were identified between 1994 and 1996 from whom M. tuberculosis had been recovered from culture of respiratory specimens. Laboratory records were reviewed to document the sequence of and results for all specimens that had been submitted for each of the patients. The "look back" period for additional specimens submitted from the same patient encompassed 2 months before and after and the date of the first positive culture.

Laboratory procedures. Mycobacterial cultures were performed by digesting and decontaminating respiratory specimens with N-acetyl-L-cysteine-NaOH, concentrating by centrifugation, and resuspending the sediment in 1.0 ml of 0.067 M phosphate buffer, pH 6.8, as previously described (7, 21). A 0.5-ml aliquot of the resuspended sediment was inoculated into a radiometric BACTEC 12B vial (Becton Dickinson Laboratory, Cockeysville, Md.), 2 drops were plated onto Middlebrook 7H11 agar, and cultures were incubated for a maximum of 8 weeks. A positive culture result was defined as recovery of M. tuberculosis from either the BACTEC 12B and/or 7H11 culture. Once one of the two cultures yielded M. tuberculosis, the remaining negative paired culture was not further evaluated. Nucleic acid probe hybridization (AccuProbe; Gen-Probe, San Diego, Calif.) was used to identify M. tuberculosis. Smears were prepared from the concentrated sediment, stained with auramine-rhodamine, and examined using fluorescence microscopy. Kinyoun staining was performed as necessary to confirm positive auramine-rhodamine smears.

Statistical analysis. Fisher's exact test (two tailed probability) was performed as previously described (1).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

A total of 430 respiratory specimens were submitted from the 143 patients. The mean and median number of specimens submitted per patient was three (range, 1 to 10). For 100 (70%) patients three or fewer specimens were submitted, and for 43 (30%) patients more than three specimens were submitted. A variety of respiratory specimens were submitted (Table 1). Of these various specimens, 50 patients submitted expectorated sputa only, 44 patients submitted induced sputa only, 8 patients submitted tracheal aspirates only and one patient submitted bronchoalveolar lavage (BAL) fluid only. The remaining 40 patients submitted more than one type of specimen among the first three specimens26 submitted both induced and expectorated sputa, 6 submitted both induced sputa and BAL fluids, 2 submitted both tracheal aspirates and induced sputa, 4 submitted both tracheal aspirates and expectorated sputa, 1 submitted both tracheal aspirate and BAL fluid, and 1 submitted both expectorated sputa and BAL fluid.

The overall cumulative culture- and smear-positive rates for all sequentially submitted respiratory specimens in this study are shown in Table 2.

The overall culture-positive rate for all respiratory specimens is shown in Table 3. The overall culture-positive rates for all respiratory specimens except expectorated sputa were comparable to that of expectorated sputa.

A direct comparison of M. tuberculosis recovery rates from BACTEC 12B broth versus 7H11 agar was not performed as part of this study.

Acid-fast smear microscopy results. Although all 143 patients had M. tuberculosis cultured from one or more specimens, only 46 (32%) of the 143 patients had acid-fast bacilli detected in one or more concentrated smear preparations (Table 2). Within this group, there was complete concordance between detection of acid-fast bacilli and recovery of M. tuberculosis in in vitro culture.

Culture results for expectorated sputa versus induced sputa. Fifty patients submitted expectorated sputa only, and 44 patients submitted induced sputa only. The cumulative culture-positive rates for the two groups were identical (Table 4).

Diagnostic yield of specimens obtained from AIDS patients. Twenty-three (16%) of the 143 patients had a prior diagnosis of AIDS. Seventy-eight specimens had been submitted from these 23 AIDS patients, with a mean number of 3.4, a median of 4, and a range of 1 to 8 specimens submitted per patient. With the exception of seven AIDS patients for whom expectorated sputa (four patients) or induced sputa (three patients) only had been submitted, the remaining 16 AIDS patients had a variety of respiratory specimens submitted for laboratory evaluation. Specimens obtained from AIDS patients comprised 26 (16%) of the 167 expectorated sputa, 33 (17%) of the 195 induced sputa, 12 (22%) of the 55 tracheal aspirates, and six (55%) of the BAL fluids.

Nondiagnostic specimens. The 43 patients (including 11 AIDS patients) from whom more than three specimens were submitted had a total of 210 nondiagnostic specimens submitted, of which 74 were the fourth or greater sequential specimen. These 74 nondiagnostic specimens represented 17% of the 430 total specimens in this study.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Despite the well-documented increased sensitivity for the BACTEC 460 radiometric mycobacterial broth culture system for recovery of M. tuberculosis, our study failed to demonstrate that this increased laboratory sensitivity could be extrapolated to a clinical recommendation that fewer respiratory specimens would be necessary for a definitive diagnosis of tuberculosis. In fact, the incremental diagnostic yield of culture for sequential specimens in our study was of similar magnitude to those in previously published studies conducted either with (14) or without (3, 4, 10) a BACTEC radiometric culture system (Table 6).

It had been questioned locally whether induced sputum has a higher diagnostic yield than expectorated sputum for respiratory tuberculosis. Our results failed to demonstrate a difference in diagnostic yield between these two specimen types, and they confirm those of Merrick and colleagues, who previously demonstrated comparable diagnostic culture yields for expectorated and induced sputa (12).

Our study did not demonstrate a higher diagnostic yield for specimens obtained from AIDS patients compared with non-AIDS patients. In contrast, Finch and colleagues were able to recover M. tuberculosis from culture of the first sputum (either expectorated or induced) for 20 (100%) of 20 HIV type 1-infected individuals (5) and detected acid-fast bacilli in the first sputum of 11 (79%) and in the second sputum of the remaining 3 (21%; cumulative rate, 100%) of their 14 smear-positive HIV-infected patients (5). Our differing results may have been related to a bias unknowingly introduced into our study by our inability to identify all HIV-infected patients who had not yet received a diagnosis of AIDS. An alternative explanation, however, may be related to the extraordinary diagnostic yield Finch and colleagues achieved in generalthey recovered M. tuberculosis from culture of the first specimen from 142 (99%) of 143 non-HIV-infected patients (5), a much higher diagnostic yield from the first specimen than achieved by us or others (Table 6).

Seventy-four (17%) of the 430 specimens in this analysis represented the fourth or greater sequential specimen for individual patients. None of these 74 specimens yielded diagnostic results. Despite the observations that in vitro recovery of M. tuberculosis is proportional to the total number of specimens submitted (3, 9, 10), the minimal increase in recovery from the fourth or greater sequential specimen has led to the recommendation that only three sputum specimens are necessary for a definitive yet cost-effective diagnosis of pulmonary tuberculosis (4, 5, 14). Given our observation of fairly comparable isolation rates of M. tuberculosis from different types of respiratory specimens, we further recommend that only three sequential respiratory specimensregardless of typeare necessary for a definitive and cost-effective laboratory diagnosis of pulmonary tuberculosis.

Our overall diagnostic yield for acid-fast smear microscopy for sequential specimens was comparable to that of previous studies (Table 7). Of note and for all patients from whose respiratory specimens M. tuberculosis was recovered in in vitro culture, there was complete concordance between detection of acid-fast bacilli and recovery of M. tuberculosis in culture. Our high diagnostic culture yield from smear-positive specimens also supports the recommendation that laboratory evaluation of only two smear-positive specimens is necessary to definitively diagnose pulmonary tuberculosis (5, 18).