INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The incidence of tuberculosis is rising in many parts of the world despite the availability of highly effective combinations of chemotherapeutic agents that can produce a 95% cure rate (1). Adherence to therapeutic regimens is poor, with an estimated noncompliance rate of 25% to 45% in the United States (2, 3) and up to 75% or more in certain regions of developing countries (4). The prolonged treatment period and resulting poor adherence to therapeutic regimens promotes transmission of disease and leads to the emergence of drug-resistant strains of Mycobacterium tuberculosis. New treatments for tuberculosis are urgently needed, but testing of new drugs and treatment regimens is expensive, particularly with the extended monitoring phase required for tuberculosis. A surrogate marker that would accurately reflect effective antimicrobial activity after only a few days of chemotherapy would provide a significant advance in assessing the response to standard and experimental antituberculosis drugs and protocols. Rapid evaluation of chemotherapy in order to provide a means for quickly determining whether a treatment regimen is appropriate is particularly important because of the increasing prevalence of multidrug-resistant strains of M. tuberculosis.

The most common methods for monitoring response to chemotherapy are a progressive reduction in the numbers of acid-fast bacilli (AFB) in stained preparations of sputum and conversion of sputum cultures to M. tuberculosis-negative status. These methods have certain disadvantages. One-half of all new cases of pulmonary tuberculosis are smear-negative upon initial diagnosis. Enumeration of AFB in sputum lacks sensitivity and is not specific for M. tuberculosis organisms. Clearance of AFB from sputum is often protracted, and staining cannot distinguish viable from nonviable bacilli. Conversion of a positive culture to a negative one at 1 to 2 mo after initiation of treatment correlates with the sterilizing activity (i.e., complete killing of M. tuberculosis bacilli in tissues) of the drugs administered, and is considered the best predictor of treatment success (5). However, because of the slow growth rate of M. tuberculosis, this measure can be determined only after weeks or months of treatment.

Effective treatment regimens result in a rapid decrease in the number of viable M. tuberculosis organisms in sputum, with a decrease in cultivable bacilli by ~ 10-fold within the first 1 to 2 wk (6). The early phase of killing, or early bactericidal activity (EBA), can be examined by measuring the decrease in bacilli cultivable from sputum within the first few days of treatment. Although EBA does not correlate with outcome, it is useful for comparing drug regimens (5) as well as for measuring the therapeutic margin of a drug (11).

The goal of the present study was to develop new surrogate methods to rapidly evaluate the efficacy of chemotherapy for tuberculosis and to compare them with the standard methods of culture conversion and EBA. In this study, subjects with smear-positive pulmonary tuberculosis were treated with an optimal short-course chemotherapy regimen, and microbiologic markers in the patients' sputa were quantified during the course of treatment. In addition to performing traditional enumeration of sputum AFB and viable bacilli, we measured specific levels of M. tuberculosis messenger RNA (mRNA), ribosomal RNA (rRNA), and DNA, representing molecules with three different inherent molecular stabilities. Prokaryotic mRNA has a short half-life (12) and would therefore be predicted to be found only in viable organisms. The specific mRNA target selected for this study is transcribed from the 85B (alpha antigen) gene of M. tuberculosis, which encodes a fibronectin-binding protein that is also involved in mycolic acid biosynthesis (13). The 85B protein is expressed at high levels in both broth- (16) and macrophage-grown M. tuberculosis organisms (17), and is expressed in vivo, with the majority of pulmonary tuberculosis patients showing a vigorous antibody response to this protein (18, 19). Ribosomal RNA is a stable RNA target with a substantially longer half-life than mRNA and greater abundance than mRNA, with estimated levels of 100 times that for the total pool of mRNA (20). M. tuberculosis DNA is the most stable of the three molecules we measured, and has been shown to persist in sputum in certain patients even after tuberculosis is cured (21).

Quantification of molecular targets was accomplished with an automated polymerase chain reaction (PCR) system, the ABI Prism 7700 Sequence Detection System, which provides accurate quantification of DNA molecules over a wide range of starting concentrations (23). The system allows a high throughput with no postamplification processing required, thus minimizing the chance of false-positive values as a result of contamination.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

The comparative microbial marker study was initiated at the Hospital Universitario Cassiano Antonio de Morase, Universidade Federal do Espírito Santo, Vitória, Brazil, using sputum collected from adults not infected with human immunodeficiency virus (HIV), but with initial episodes of newly diagnosed, AFB smear-positive pulmonary tuberculosis. All subjects underwent HIV testing after counseling before and after testing. Each subject gave informed consent by signing a statement in the local language. The study was approved by the institutional review boards of Case Western Reserve University, Universidade Federal do Espírito Santo, University of Arkansas for Medical Sciences, and Duke University Medical Center.

All subjects were treated daily for 2 mo with a standard four-drug regimen of isoniazid (300 mg/d), rifampin ( 50 kg at 450 mg/d, > 50 kg at 600 mg/d), ethambutol ( 50 kg at 800 mg/d, 50 kg to 69 kg at 1,000 mg/d,  70 kg at 1,200 mg/d), and pyrazinamide ( 50 kg at 1.5 g/d, > 50 kg at 2 g/d), followed for an additional 4 mo by daily isoniazid and rifampin. Subjects were hospitalized during the first 2 wk of treatment for directly observed therapy. Subsequent treatment was self-administered and adherence was monitored through clinic visits, drug-dispensing records, and testing of urine for isoniazid metabolites. All of the subjects responded favorably to therapy, and were 100% culture negative after 3 mo of treatment.

Sputum Collection and Processing

Spontaneously produced, spot sputum specimens were collected in triplicate before the start of chemotherapy (Day 0) and in duplicate on Days 2, 4, 7, 14, 30, 60, 120, and 180 after the start of treatment. Sputum was stored at 4° C for no longer than 24 h and was then homogenized by adding 1 ml of 2.5% N-acetyl-L-cysteine (NALC) dissolved in phosphate buffer (67 mM, pH 6.8) and vortex mixing with 4-mm glass beads (26). Specimens were divided into aliquots of 0.5 ml of homogenate, and were frozen at 70° C until nucleic acid extraction. A separate aliquot of homogenized sputum was further processed with 2% NaOH-sodium citrate for quantitative microscopy and culture (27).

Patient Control Group

To assess the specificity of the assays, sputum was collected from each of nine subjects with nontuberculosis suppurative lung disorders (e.g., chronic bronchitis, paracoccidioidomycosis). These specimens were processed identically to those obtained from the subjects in the tuberculosis study group. No IS6110 DNA, 85B mRNA, 16S rRNA, AFB, or viable M. tuberculosis bacilli were detected in any of these samples.

Quantitative AFB Microscopy and Culture

For quantitation of AFB, 20 µl of each aliquot of homogenized and decontaminated sputum was fixed onto a 78-mm² area of a glass slide, heat-fixed, and stained with auramine-rhodamine, and the number of AFB was counted microscopically (28). This was done with a ×40 objective without oil. A minimum of five fields (for specimens with more than 100 bacilli per field) and a maximum of 300 fields (for specimens with five or fewer bacilli per field) were counted per slide in order to count no fewer than 300 bacilli per specimen. A ×100 objective was used in cases in which a count was difficult to obtain with the ×40 objective (i.e., if the bacilli were too numerous). The number of bacilli counted was corrected to numbers of AFB per milliliter of sputum by correcting for the number of fields counted, the area counted (i.e., the number of fields and magnification factor of the objective), and the dilution factors for the original specimen during its processing.

Quantitative culture was achieved by plating replicate serial dilutions of homogenized and decontaminated sputum from each aliquot on 7H10 and 7H10S agar plates (7H10S medium contains 100 U/ml polymyxin B, 5 µg/ml amphotericin B, 25 µg/ml carbenicillin, and 10 µg/ml trimethoprim) (29). The number of colonies was counted on four occasions during incubation for 3 to 4 wk at 37° C in a 5% CO₂ atmosphere.

RNA and DNA Isolation

RNA was isolated from sputum through a modification of a procedure described previously (26). Briefly, 1,000 µl TRIzol-LS (Gibco BRL Life Technologies, Grand Island, NY) was added to 0.5 ml of homogenized sputum, and the mixture was transferred to a glass matrix tube for cell lysis (FastRNA Blue; BIO 101, Vista, CA). The mixture was processed in a spin/rotation instrument for cell lysis (FastPrep-120; BIO 101), with a speed setting of 6.5 and a time setting of 2 × 45 s. After processing, 200 µl chloroform was added to the mixture and it was vortex-mixed for 2 min. The aqueous and organic layers were separated by microcentrifugation for 15 min at room temperature at 16,000 × g. The aqueous phase, containing the RNA, was removed and 100 µl Cleanascite (CPG, Lincoln Park, NJ) was added, after which the samples were mixed for 10 min on a rocker table and the Cleanascite was removed by centrifugation for 1 min at 16,000 × g. The aqueous phase was extracted with 500 µl chloroform:isoamyl alcohol (24:1 vol/vol). This step was followed by addition of 4 µl glycogen (Ambion, Austin, TX), 0.1 volume of 5 M ammonium acetate, and an equal volume of isopropanol, and the RNA was precipitated overnight at 20° C. The resulting RNA pellet was washed twice with 75% ethanol, resuspended in 80 µl diethyl pyrocarbonate-treated distilled H₂O (dH₂O) (Ambion), and treated with deoxyribonuclease (DNase) by adding 10 µl 10X DNase I buffer (0.5 M Tris, pH 7.5; 0.1 m MgCl₂; 1 mM dithiothreitol; 50 µl/ml bovine serum albumin), 5 µl Prime RNase inhibitor (5 Prime  3 Prime, Inc., Boulder, CO), and 5 µl DNase I (2 U/µl, Ambion) and incubating for 30 min at 37° C. RNA was extracted with phenol and CHCl₃, precipitated with isopropanol, and resuspended in a final volume of 100 µl dH₂O.

DNA was isolated from the organic phase and interface layers remaining from the foregoing procedure through a back-extraction produced described previously (23)

Quantification of M. tuberculosis Nucleic Acids

16S rRNA. Reverse transcription-polymerase chain reaction (RT- PCR) for mycobacterial 16S rRNA was performed as described (26). A limiting dilution RT-PCR was used for quantifying the number of 16S rRNA molecules. Individual RT reactions were performed on serially diluted RNA samples isolated from sputum. The RT reaction was followed by PCR with oligonucleotide primers that amplify a region of the 16S rRNA gene in all members of the genus Mycobacterium with the exception of M. genevense (26). PCR amplification was done in the presence of ³²P-labeled oligonucleotide primers; the amplified products were electrophoresed through 8% polyacrylamide gel and visualized with autoradiography. Control reactions to monitor DNA contamination were conducted with extracted RNA in a mock RT reaction without the presence of avian myeloblastosis virus (AMV; Boehringer-Mannheim, Indianapolis, IN) enzyme.

The number of molecules of 16S rRNA per specimen was calculated by determining the titer of a sample (i.e., the greatest dilution at which a band for the 16S rRNA amplicon was observed autoradiographically) and correcting for dilution factors for RT and PCR, and for the efficiency of these reactions. As a control for the efficiency of amplification, PCR for the 16S rRNA gene was done on known quantities of genomic DNA purified from M. bovis ATCC 19210 (American Type Culture Collection, Rockville, MD).

To determine the limit of detection for this assay, sputa from M. tuberculosis-negative subjects was spiked with M. tuberculosis strain H37Rv and processed as a test specimen. A concentration of 10 cfu/ml in the original specimen was easily and consistently quantified. The in vivo sensitivity of the assay could not be determined, since all pretreatment (Day 0) specimens had very high levels of 16S rRNA, but would be expected to be similar to that of the spiked sputum studies.

85B mRNA. We measured 85B mRNA by using a 5' exonuclease TaqMan assay (30) and the ABI 7700 Sequence Detection System (Perkin-Elmer/Applied BioSystems, Foster City, CA). RNA (5 µl of the total volume of 100 µl RNA) isolated from sputum was reverse transcribed using AMV reverse transcriptase in a 20 µl reaction volume, as previously described (26). The efficiency of RT was determined for each assay by using in vitro RNA transcripts of a cloned M. tuberculosis 85B gene. Dilutions of control transcripts ranging from 10² to 10⁴ molecules/µl in 10 ng/µl yeast carrier RNA (Ambion) were included in each RT-PCR assay. The efficiency of RT was further determined by calculating the ratio of the number of observed to expected 85B molecules per reaction.

Following RT, PCR-based enumeration of 85B complementary DNA (cDNA) molecules from both samples and in vitro control RNA was done with the ABI Prism 7700. PCR primers were specific for the M. tuberculosis complex 85B gene (26). The TaqMan detector probe is a dual-labeled oligonucleotide, hybridizing internally to the flanking PCR primers, with the sequence 5'-(5-carboxyfluoroscein [FAM])- TCGAGTGACCCGGCATGGGAGCGT-(N,N,N,'N'-tetramethyl-6-carboxyrhodamine [TAMRA]). The TAMRA emission spectrum was used to standardize for background fluorescence (24). The threshold value of florescence (cycle threshold, Ct) indicating the presence of a specific PCR-generated product was set at 10 times the standard deviation (SD) of the mean baseline emission. The quantity of specific target DNA in each reaction was determined from the Ct value with reference to a standard curve generated by amplification of known amounts of genomic DNA obtained from M. bovis ATCC 19210. Standard reactions containing from five to 78, 125 copies of M. bovis genomic DNA (containing one copy of the 85B gene per genome), were included in triplicate with each assay. Graphs of starting DNA concentration versus Ct value are consistently linear over this range. Ct values for samples and standards fell between 18 and 35 cycles of amplification.

PCR amplification and quantification of each RT reaction was done twice, and the mean value obtained from the replicate determinations was used in subsequent calculations. In each case, to correct for possible DNA contamination of the extracted RNA, the amount of 85B target measured in the absence of reverse transcriptase was subtracted from the value obtained when the enzyme was included in the reaction mixture. Samples yielding values in the absence of reverse transcriptase that were 20% or more than the value obtained in the presence of the enzyme were considered contaminated with DNA and the results were not used. This occurred in five specimens of 355 analyzed for this study.

The number of molecules of 85B mRNA/ml sputum was calculated by correcting for the efficiency of RT, as determined with the control RNA transcripts, the dilution factors involved in RT and PCR amplification, and the dilution factor for the original processing of sputum. In this way all values were corrected to the same original volume of sputum and could be directly compared.

For this assay, the limit of detection of M. tuberculosis H37Rv, spiked into negative sputum and processed as described earlier, was 1,000 cfu/ml. The results were similar for the test subjects' sputa, in which a bacillary load of 1,000 to 10,000 cfu/ml sputum was required for a positive signal.

IS6110 DNA. Levels of microbial DNA present in sputum specimens were determined by quantification of the IS6110 insertion element. The assay conditions showed no cross-reaction with nontuberculous mycobacteria or bacteria from related genera, and were specific for the M. tuberculosis complex (23). DNA was measured with the ABI Prism 7700 as described (23). The limits of detection were greater for the DNA assay than for the RNA assays, since positive values were obtained from specimens in which no positive values were measured with the RNA assays. The in vitro studies showed that DNA could be measured when as few as 100 cfu/ml M. tuberculosis H37Rv were spiked into negative sputum and processed described earlier.

Statistical Analyses

All values were log transformed to base 10. The variability within samples was assessed by calculating the coefficient of variation (CV) between values obtained for individual markers in replicate spot sputum samples. The CV was then averaged across patients at each time point. Paired t tests were used to evaluate changes in a specific marker from one time point to baseline, and a standard t test was used to compare the same marker in different subsets of patients at a given time point. Fisher's exact test was used for all categorical variables, and Pearson's product moment correlation coefficient (r) was used as a measure of correlation between markers at a particular time interval. An alpha value of 0.05 was used for all analyses.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Thirty-six patients were enrolled in the study and 19 were evaluable upon its completion. Subjects were withdrawn from the study if they were shown to harbor a drug-resistant strain of M. tuberculosis (n = 4), failed to produce sputum after enrollment in the study (n = 3), developed drug-induced hepatitis (n = 2), failed to return to clinic during the trial period (n = 2), had baseline cultures that developed contamination (n = 5), or had laboratory errors during processing of their specimens (n = 1). Those remaining in the study group were 74% male (14 of 19), with a mean age of 34 yr (range: 19 to 55 yr); 95% (18 of 19) had cavitary disease.

To correlate a molecular marker with either EBA or 2-mo culture conversion, sputum specimens were serially collected from patients treated with an optimal drug regimen. At each time point, sputum was homogenized, total RNA and DNA were isolated, and numbers of cfu and AFB were determined. Replicate sputum specimens were processed on each day of collection, and the mean value was recorded for each marker.

Evaluation and Validation of Baseline Data

Levels of 85B mRNA, IS6110 DNA, and 16S rRNA, and numbers of AFB and cfu at selected time points before and after initiation of chemotherapy are shown in Table 1. The mean number of M. tuberculosis organisms grown from sputum before treatment was log₁₀ 5.92 cfu/ml. The corresponding pretreatment number of AFB was log₁₀ 6.20 AFB/ml. The close agreement between the number of AFB enumerated by microscopy and the number of bacilli detected by growth on plates (r = 0.83, p = 0.0001) indicates efficient recovery of viable organisms before the onset of treatment, with only limited numbers of M. tuberculosis organisms killed during the decontamination process. The observed numbers of molecules for the various M. tuberculosis molecular markers were in good agreement with predicted values. A mean of ~ 800 copies of 16S rRNA per cfu was found, closely agreeing with the estimated number of 4,000 ribosomes per cell in M. leprae (31) and our laboratory's estimate of ~ 5,000 copies of 16S rRNA per cell in log-phase cultures of M. tuberculosis H37Rv (unpublished observation). The level of 85B mRNA averaged ~ 0.1 copy per cfu. The actual amount of any individual mRNA in a cell is not known, but we have estimated that from one to 10 copies of 85B mRNA are present per cell in cultured log-phase M. tuberculosis H37Rv cells. The initial mean IS6110 DNA value was 5-fold greater than the corresponding viable count, as would be expected for an insertion element found in multiple copies within the M. tuberculosis genome.

Colony-Forming Units and AFB Levels during Treatment

The regimen used in this study has been shown to result in 95% of patients having sputum culture conversion to negative after 2 mo of treatment (1). In the study, 89% (17 of 19 subjects) became culture negative by 2 mo (Table 1). The two patients who were culture positive at 2 mo became culture negative at 3 mo. The mean decrease in cfu was 0.6log₁₀ from Day 0 to Day 2, and 1.71log₁₀ from Day 0 to Day 7; both values represent a significant reduction from baseline (p = 0.0014, p = 0.00001) and are within the ranges of values reported in previous EBA studies (3, 6).

The number of AFB present during the course of therapy remained high for prolonged periods, showing no significant decrease from the baseline value until 14 d after the initiation of chemotherapy (p = 0.003). By the end of 2 mo of treatment, 47% of the patients were smear negative. The mean pretreatment cfu count was 10-fold lower among the patients who became AFB smear negative after 2 mo of treatment than among those who remained AFB smear positive at this point (log₁₀ 5.4 ± 0.7 and log₁₀ 6.4 ± 0.7, respectively; p = 0.006). However, there were no significant differences between these two groups in the other markers' baseline values.

Quantification of Molecular Markers during Treatment

The levels of 85B mRNA for individual patient specimens during the course of treatment are shown in Figures 1A and 1B. Although amounts of 85B mRNA varied from time point to time point, all patients had no detectable mRNA by Day 30 with the exception of one patient (Figure 1B). Rapid mRNA loss correlated with rapid culture clearance. A sputum sample that was consistently negative for 85B mRNA starting at Day 2 of therapy was significantly more likely to be culture negative at 1 mo (p = 0.03). By Day 14 of treatment, 14 of 19 subjects had no detectable mRNA in their sputum (Figure 1A).

View larger version (23K): [in this window] [in a new window]   Figure 1.   Quantification of 85B mRNA in sputum of pulmonary tuberculosis patients before and after initiation of treatment. Values (mean value per day, log10) for individual patients are represented by the various symbols. (A) The 14 patients whose 85B mRNA fell to an undetectable level by Day 14 of treatment. (B) The 5 patients whose 85B mRNA became negative after 14 d of therapy. Solid line with asterisks denotes the relapsed patient (Subject 29).

On average, levels of 16S rRNA decreased by less than 10-fold during the first week of treatment and remained high in specimens from the majority of subjects for the first 60 d (Figure 2). Four of 19 subjects had a greater than average decrease in rRNA, having undetectable levels by 60 d of treatment. These same four subjects had correspondingly abrupt decreases in M. tuberculosis DNA, and 85B mRNA was undetectable in their sputum by Day 7. This group of four subjects also had significantly lower baseline 16S rRNA levels than the remaining 15 subjects (p = 0.0004). Baseline values for all other markers were not different for this subset of patients.

View larger version (24K): [in this window] [in a new window]   Figure 2.   Quantification of 16S rRNA in sputum of pulmonary tuberculosis patients before and after initiation of treatment. Values (mean value per day, log10) for individual patients are represented by circles and dotted lines. The mean value for all subjects is depicted by the solid heavy line.

Levels of IS6110 DNA in sputum remained high for the first 2 mo of treatment for 13 of 19 subjects (Figure 3), but six subjects had substantial declines in this marker during this time. The mean pretreatment DNA level for the latter group of six subjects was log₁₀ 5.1 ± 0.9 IS6110 molecules/ml (range: 4.2 to 6.3 molecules/ml) as compared with the other 13, who showed a mean of log₁₀ 7.2 ± 0.9 molecules/ml (range: 5.8 to 8.2 molecules/ml). Similarly, the average pretreatment number of cfu/ml for this group of six subjects was also lower (mean: log₁₀ 5.1 ± 0.4, versus log₁₀ 6.3 ± 0.7), with both differences highly significant (p = 0.001 for DNA and p = 0.0005 for cfu). All six subjects had no detectable 85B mRNA in their sputum by Day 7, and three of the six subjects also showed rapid clearance of rRNA. Thus, comparison of molecules of IS6110 DNA enumerated during the first week of treatment with that after 2 mo of therapy is of limited value as a means for monitoring response to therapy, particularly when the initial bacillary load is great.

View larger version (25K): [in this window] [in a new window]   Figure 3.   Quantification of IS6110 DNA in sputum of pulmonary tuberculosis patients before and after initiation of treatment. Values (mean value per day, log10) for individual patients are represented by circles and dotted lines. The mean value for all subjects is depicted by the solid heavy line.

Evaluation of EBA

As mentioned earlier, sterilization of sputum at 2 mo could be measured either through cfu or levels of mRNA. We next wished to determine which markers would reflect EBA. Differences among the levels of reduction for the various markers examined are more easily visualized when the data are plotted as a percentage of pretreatment (Day 0) values (Figure 4). The mean decline from Day 0 to Day 2 for cfu was 75% (0.6log₁₀), and that for 85B mRNA was 99% (2.5log₁₀), both of which were significant decreases (p = 0.001 and p = 0.0001, respectively). Neither the AFB nor DNA markers showed any significant decrease in levels from baseline at Days 2, 4, and 7. However, by 14 d of treatment, both AFB and DNA showed significant reductions in levels as compared with baseline values (p = 0.003 and p = 0.005, respectively). Enumeration of AFB showed a relatively rapid decline between 30 and 60 d of treatment, perhaps reflecting the limit of sensitivity of the assay technique of ~ 10⁴ bacilli per milliliter of sputum. Levels of 16S rRNA fell between those observed for DNA and mRNA, with a significant reduction from baseline observed by Day 7 (p = 0.02). However, only four of 19 subjects became negative for this marker after 2 mo of treatment. In summary, for the surrogate markers examined, EBA was detectable during the first few days of treatment only from the decrease in cfu, as previously described (3, 6), or from the decrease in 85B mRNA.

View larger version (23K): [in this window] [in a new window]   Figure 4.   Percent of pretreatment (Day 0) levels of markers in sputum for first 14 d after beginning of chemotherapy.

Relationship of Surrogate Markers to Relapse

One subject (Subject 29) initially responded well to treatment, becoming culture-negative by Day 60 of treatment (Table 2). This subject's sputum remained negative for AFB and cfu through the 6-mo visit, at which time he finished chemotherapy. This subject returned to the clinic for the 13-mo visit with new clinical symptoms of tuberculosis, and was found to be both sputum smear and culture positive. The relapse isolate was fully sensitive to all antimicrobial agents tested, and the subject was successfully retreated. He demonstrated the highest levels of 85B mRNA of all the subjects in the study during the first 2 wk of treatment (Figure 1B). An examination of the three molecular markers showed that this subject (Subject 29) was unique in the study (Table 2). Although he had no detectable mRNA at Day 30, he was the only subject in the study with a positive Day-60 value of 158 molecules/ml. Moreover, his sputum levels of 16S rRNA increased by 100-fold from Day 30 to Day 60 (from 1.6 × 10⁵ to 1.6 × 10⁷ molecules/ml), and his sputum levels of IS6110 DNA increased from 523 to 4,600 molecules/ml (9-fold) from Day 120 to Day 180. A similar increase in the subject's sputum DNA levels in the later weeks of therapy was not observed in any other patient.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The traditional measure of efficacy of new tuberculosis drugs and treatment regimens is 2-mo culture conversion of sputum, which has been shown to correlate with relapse rate after 2 yr from termination of therapy. EBA, as measured by a decrease in cfu, does not correlate with the sterilizing activity of a drug or predict final outcome, but it is useful for comparing a new drug or dosage with current antituberculosis therapy. In the present study, molecular targets were measured in sputum collected from tuberculosis patients receiving optimal short-course therapy, in order to evaluate their usefulness as surrogate markers. Such a marker would shorten the duration of tests of new antituberculosis drugs or more potent treatment regimens, particularly when applied to ultra-short-course trials. The ideal surrogate marker would be measured within the first week of chemotherapy, and would reflect the EBA and predict sterilization of sputum after 2 mo of continued therapy.

We tested three molecular markers, which represent the relative stabilities of various nucleic acid species found in mycobacteria. We measured M. tuberculosis DNA levels, which represented the most stable target and one not necessarily limited to the presence of viable organisms. Previous studies have shown that M. tuberculosis DNA persists in the sputum of pulmonary tuberculosis patients for as long as a year after completion of treatment (21). Although quantitative measurements of DNA have been suggested as a possible method for following response to therapy, studies with mice (22) and humans (23) have indicated that it is not an appropriate target. In the present study, quantitative DNA levels did not serve as a marker for successful chemotherapy in the majority of patients, although some subjects (six of 19) did show appreciable decreases in sputum DNA levels during the first 2 mo of therapy. The automated assay described here provides for relatively simple and rapid quantification of M. tuberculosis DNA. Because levels of IS6110 DNA obtained either before treatment or during the first week of therapy correlated with initial bacillary load (r = 0.72, p = 0.00006), quantification of DNA is an attractive alternative to enumeration of cfu by culture. This method is faster than traditional means, and could provide pretreatment baseline data in cases in which therapy was started without a baseline sample available.

Levels of a stable and abundant structural RNA, 16S rRNA, showed an intermediate clearance rate from sputum. This target did not appear to reflect bactericidal effect in 15 of 19 cases, an observation reported by others, using a commercial assay system (32). The assay we used for measurement of 16S rRNA levels amplifies this target for all members of the Mycobacterium genus, and it is possible that some subjects in our study were either coinfected or subsequently became infected or colonized with nontuberculous mycobacteria. If so, prolonged increased levels of 16S rRNA could be due to cross-reaction with another species. It is also possible that environmental mycobacteria contaminated our samples. Only M. tuberculosis was cultured from our subjects during the first 2 mo of the study period. However, subsequent cultures revealed that several subjects had at least one positive culture for nontuberculous mycobacteria. A species-specific assay would allow the further evaluation of this target as a viability marker.

The rapid decline of M. tuberculosis 85B mRNA in sputum during the first few days of effective treatment, and clearance of 85B mRNA after 2 mo of chemotherapy, indicate that this marker correlates with microbial viability. The use of RT- PCR for mRNA as a measure of viability of Mycobacterium has been suggested in the past (21, 26, 33, 34). However, the present study was the first in which RT-PCR was applied to sputum from subjects undergoing chemotherapy in a controlled setting. The success of our strategy included optimization of specimen collection and processing, and standardization of molecular assays. The use of a high-throughput automated system (ABI 7700 [TaqMan] System) allowed for accurate comparison of patient specimens by permitting rapid quantification of molecular targets in replicate, and quantification over a wide dynamic range, as compared with standard quantitative PCR assays (23). The assay system's incorporation of an internal hybridizing oligonucleotide probe provided additional specificity for the target message, which was necessary in the context of the heterogeneous RNA found in sputum.

Our results suggest that measurement of 85B mRNA may provide an alternate approach to the assessment of EBA through enumeration of cfu. The decline in 85B mRNA was greater than that observed for cfu for the Day 0 to Day 2 interval, and was similar to that observed in cultures of M. tuberculosis H37Rv treated with rifampin (data not shown). Although EBA as defined by cfu did not correlate with sputum conversion, EBA as defined by 85B mRNA levels did correspond to faster clearance of M. tuberculosis from sputum. The relationship to early culture sterilization by 1 mo was significant for clearance of 85B mRNA by Day 2 of treatment (p = 0.03). This difference could have been due to the assessment of slightly different parameters by these two methods, and to the presumed greater impact of rifampin on mRNA levels than on cfu.

The single patient to relapse in this study also had the highest levels of sputum 85B mRNA during the first 14 d of treatment, suggesting that sustained high levels of 85B mRNA measured during the first 2 wk of treatment may be an early indicator of an eventual poor treatment outcome. There are many possible explanations for treatment failure, but in this subject who relapsed, the organism was sensitive to all treatment drugs, and medications were given under direct supervision while the subject was in the hospital. Factors not evaluated included host susceptibility to tuberculosis, virulence of the infecting strain, and adherence of the patient to the drug regimen after hospital discharge. Further studies are required to determine whether sustained 85B mRNA levels during the first 2 wk of chemotherapy predict treatment failure.

The average number of molecules of 85B mRNA measured in M. tuberculosis is ~ 5 molecules/organism in vitro and ~ 0.1 molecule/organism in vivo. Similarly, 16S rRNA levels averaged 5,000 molecules/organism in vitro and ~ 800 molecules/ organism in vivo. The difference between the amounts of 16S rRNA in broth-grown organisms and those recovered from sputum may be due to incomplete recovery of RNA or loss of RNA between the time of sputum collection and processing. It is also likely that the numbers of ribosomes and certain mRNAs are reduced in bacilli growing under less than favorable culture conditions, such as those in an immunocompetent host. Expression of 85B mRNA in sputum fluctuated between positive and negative values until Day 7 of treatment. The recovery of 85B mRNA from sputum, in addition to the limit of sensitivity for 85B mRNA detection of 10³ to 10⁴ cfu/ml, most likely contributed to this variation. These results suggest that a different RNA surrogate marker will be required for monitoring treatment of paucibacillary disease.

With the rising global incidence of tuberculosis and increasing numbers of drug-resistant M. tuberculosis isolates, clinical trials of new drugs and regimens are urgently needed (35). We have shown that it is possible to isolate labile mycobacterial RNA from specimens collected at a remote site, and to subsequently quantify selected molecular targets with an automated system. The rapid disappearance of M. tuberculosis mRNA from sputum during the first few days of therapy suggests that it is a good marker of microbial viability and that quantification of mRNA will be useful for monitoring response to chemotherapy. This methodology could be applied to assessing the efficacy of new drugs and regimens, and to the rapid evaluation of treatment in subjects with suspected multidrug-resistant tuberculosis.