Detection of transrenal DNA for the diagnosis of pulmonary tuberculosis and treatment monitoring

Supplementary info

Performance characteristics of trDNA detection in vitro

We used DNA spiked urine samples to determine the capture and amplification efficiency of our PCR-based trDNA assay in vitro. Urine samples were spiked with two different DNA templates, a 75 base pair (bp) DNA fragment containing the highly specific MTB motif IS6110 and genomic DNA from the MTB reference strain H37Rv. To estimate the capture efficiency we compared the calibration curves from DNA isolated from spiked urines with pure MTB DNA (75bp fragment or H37Rv) or pure DNA directly spiked into the PCR reagent mix. Our results indicate that the calibration curves for both template types were linear over a concentration range spanning from 10 to 100,000 copies/ml (Suppl. fig. 1; p < 0.0001; r2 > 0.99). A global analysis of linear regression parameters indicates that differences between slopes were statistically non-significant between different templates (ANCOVA test p > 0.1; see supplemental information). Below 10 copies/ml ct-values showed higher variability in agreement with the detection-limit of the assay (see below). TrDNA assay calibration curves were on average 1.6 ct-values smaller than those observed in RT SYBR Green PCR reactions. These differences were marginally significant (ANCOVA test p >0.03) and might reflect that trDNA processing leads to partial loss of spiked input or that elution might carry unknown PCR inhibitors. Since we observed that PCR amplification efficiencies ranged from 96-99% for trDNA assay to 98-101% for RT SYBR Green PCR, we can rule out that eluates carry PCR inhibitors. These results indicate that DNA spiked in urine is efficiently captured and amplified within the mentioned linear range. Using a conservative approach, we set to use the 75bp DNA fragment linear regression parameters as calibration curve for trDNA content in patient urine samples within this study. In order to define the cut-off value for TB negative samples we set the upper 95% confident limit from the no template dilution step whose value corresponded to a ct of 38.00 (i.e. trDNA positive TB cases reflect ct-values < 38; this result set the detection level to 3 copies/mL). This cut-off value is in agreement with results from an independent analysis of 42 healthy urine samples (data not shown).

Urine dip stick, renal function, CRP and trDNA correlations

The assay seems to act independently from renal function parameters and general inflammatory status as we find no significant correlation between maximal trDNA with creatinine, urea and CRP levels (data not shown). Moreover the correlations between measured urine parameters (specific gravity, pH, leucocytes, nitrate, protein, glucose, urobilinogen, bilirubin, erythrocytes) and measured trDNA values were not statistically significant. In addition the maximal trDNA levels observed on each patient did not show a significant correlation with the number of IS6110 repeats determined for infectious TB strains from the culture isolate (p>0.4; r2=0.076, data not shown).

Beside these physiological and epidemiological parameters, study subject data as gender, age, smoker status and skin test result (Table 1) were examined and found not to be correlated with the maximal trDNA values (data not shown).

Supplementary table 1. Time to first positive to negative conversion for four diagnostic methods (MGIT, AFB, LJ, trDNA). The ID numbers represent the identification numbers of the enrolled patients.

Suppl. Figure 1. Performance of trDNA detection using in-vitro samples. Calibration curves obtained from spiked urine sample processed for trDNA extraction and analyzed by either trDNA assay (upper panel) or RT SYBR Green PCR (lower panel) are shown. Spiked material was either genomic TB (H37rv strain) or a 75 bp long clone. TrDNA detection relies on the amplification of the TB- specific IS6110 motif. Lines indicates the fitted linear regression line (continuous) and the observed 95% confident limits (dotted).

Suppl. Figure 2. Spearman’s correlation scatterplots between Ralph’s score and maximal trDNA leves (A), TTP of MGIT cultures (B) and TTP of LJ cultures (C) at the onset of treatment. D correlation between IS6110 motif copy number and the max trDNA per patient.

Suppl. Figure 3. Time series (days, 0 = time point of therapy initiation) trDNA levels of all 22 TB-positive patients. An asterisk (*) beside the patient ID number highlights the treatment naïve subgroup. Time series are restricted to 60 days after treatment initiation because frequency of trDNA becomes low as well as the levels of trDNA drops below 0,5 log cp/ml.

Suppl. Figure 4. Course of TB negative and TB positive results (percentage) from AFB microscopy after treatment initiation

1