Coadministration of Allopurinol To Increase Antimycobacterial Efficacy of Pyrazinamide as Evaluated in a Whole-Blood Bactericidal Activity Model

ABSTRACT

Coadministering pyrazinamide (PZA) with the xanthine oxidase inhibitor allopurinol increases systemic levels of the active metabolite, pyrazinoic acid (POA), but the effects on bactericidal activity against tuberculosis are unknown. We randomized healthy volunteers to take a single dose of PZA (either 10 or 25 mg/kg of body weight) at the first visit and the same dose 7 days later, coadministered with allopurinol (100 mg daily; 2 days before to 1 day after the PZA dose). Blood was drawn at intervals until 48 h after each PZA dose, and drug levels were measured using liquid chromatography-tandem mass spectrometry. Whole-blood bactericidal activity (WBA) was measured by inoculating blood samples with Mycobacterium tuberculosis and estimating the change in bacterial CFU after 72 h of incubation. Allopurinol increased the POA area under the concentration-time curve from 0 to 8 h (AUC_0–8) (18.32 h · μg/ml versus 24.63 h · μg/ml for PZA alone versus PZA plus allopurinol) (P < 0.001) and its peak plasma concentration (C_max) (2.81 μg/ml versus 4.00 μg/ml) (P < 0.001). There was no effect of allopurinol on mean cumulative WBA (0.01 ± 0.02 ΔlogCFU versus 0.00 ± 0.02 ΔlogCFU for PZA alone versus PZA plus allopurinol) (P = 0.49). Higher systemic POA levels were associated with greater WBA levels (P < 0.001), but the relationship was evident only at low POA concentrations. The lack of an effect of allopurinol on WBA despite a significant increase in blood POA levels suggests that host-generated POA may be less effective than POA generated inside bacteria. Coadministration of allopurinol does not appear to be a useful strategy for increasing the efficacy of PZA in clinical practice. (This study has been registered at ClinicalTrials.gov under registration no. NCT02700347.)

INTRODUCTION

Pyrazinamide (PZA) is a key sterilizing drug used in first-line combination therapy to treat tuberculosis (TB) (1). PZA is a prodrug which is converted by pyrazinamidase inside bacteria to pyrazinoic acid (POA), the active form of the drug (1). PZA is also converted to POA in the liver by microsomal deamidase and is further metabolized by the enzyme xanthine oxidase (Fig. 1) (2, 3). Host-derived POA can penetrate lung lesions (4) and may play an important role in overall bactericidal activity of PZA in patients with TB.

FIG 1.

Pyrazinamide metabolism and the mechanism of action by which allopurinol increases the active metabolite (pyrazinoic acid). (Adapted from reference 24 with permission.)

Pharmacokinetic-pharmacodynamic (PK-PD) modeling suggests that boosting POA by increasing the current clinical dose of PZA may improve outcomes (5). An alternative approach might be to block liver POA metabolism by using a xanthine oxidase inhibitor, thereby augmenting systemic POA levels (Fig. 1). The most well-established xanthine oxidase inhibitor is allopurinol, which has a good safety profile and has been used for decades for management of gout and other conditions (6, 7). Studies with mice and healthy volunteers have shown that coadministration of allopurinol with PZA produces large increases in systemic POA levels (4, 8).

We hypothesized that coadministration of allopurinol might enhance the antimycobacterial efficacy of PZA, and we sought to test this in a human whole-blood bactericidal activity (WBA) model.

RESULTS

We screened 22 healthy volunteers between March and April 2016. Nine were excluded (6 were possibly or definitely HLA-B*5801 positive, 2 had abnormal blood results, and 1 had hepatitis B virus infection), and 1 withdrew before randomization. Twelve volunteers were randomized to two groups, i.e., 6 volunteers to each PZA dose (10 or 25 mg/kg of body weight). All were male and of Asian ethnicity, the median age was 31 years (range, 25 to 52 years), the median weight was 71.5 kg (range, 52.1 to 93.7 kg), all were HIV negative, and 11 were interferon gamma release assay (IGRA) negative (1 was indeterminate).

Participants reported full adherence to allopurinol as prescribed, and PZA ingestion was observed by research staff. There were no clinical adverse events considered related to study drugs. The only laboratory adverse events were a rise in serum uric acid in 3 participants (grade 1 in 1 and grade 2 in 2 participants). One participant (10 mg/kg PZA) withdrew after the initial PZA dose due to reasons unrelated to the study; available data for this individual obtained to 24 h postdose are included in the results. Pharmacokinetic (PK) samples were collected at all scheduled time points to 48 h after both the PZA-only (PZA) visit and the PZA-allopurinol coadministration (PZA+ALLO) visit. WBA samples were collected at all scheduled time points to 8 h for all participants at both visits, at the 24-h time point for 8 participants, and at the 48-h time point for 4 participants (data not included in the main analysis).

For PZA doses of 10 mg/kg and 25 mg/kg combined, coadministration with allopurinol produced a large increase in the POA area under the concentration-time curve from 0 to 8 h (AUC_0–8) (18.32 h · μg/ml versus 24.63 h · μg/ml for PZA versus PZA+ALLO) (P < 0.001) (Table 1) and in its peak plasma concentration (C_max) (2.81 μg/ml versus 4.00 μg/ml) (P < 0.001) (Table 1). Coadministration with allopurinol produced a small increase in PZA AUC_0–8, did not change PZA C_max (Table 1), and produced a large decrease in AUC_0–8 for both 5-hydroxypyrazinamide (5-OH-PZA) and 5-hydroxypyrazinoic acid (5-OH-POA) (see Table S1 in the supplemental material). The effects of allopurinol coadministration were of similar relative magnitudes for the 10-mg/kg and 25-mg/kg PZA doses analyzed separately (Table 1; Fig. 2). The POA AUC and C_max obtained with 10 mg/kg PZA and allopurinol remained substantially lower than those obtained with the dose of 25 mg/kg PZA alone (Table 1). The effects of allopurinol on PZA and POA AUCs were similar for data points up to 24 h (Table 1).

TABLE 1.

Pharmacokinetic characteristics of POA and PZA^b

Group and parameter	Value for visita		GMR or mean difference (95% CI)	P value
	PZA alone	PZA+ALLO
POA pharmacokinetic characteristics
PZA (10 and 25 mg/kg) groups (n)	12	11
AUC0–8 (h · μg/ml)	18.32 (7.1, 43.2)	24.63 (9.0, 54.1)	1.27 (1.17, 1.37)	<0.001
AUC0–24 (h · μg/ml)	46.44 (17.8, 102.7)	69.11 (25.9, 139.3)	1.42 (1.37, 1.48)	<0.001
Cmax (μg/ml)	2.81 (1.1, 6.2)	4.00 (1.5, 8.4)	1.35 (1.29, 1.41)	<0.001
Tmax (h)	5.1 (1.6)	6.7 (1.6)	1.5 (0.2, 2.9)	0.03
t1/2 (h)	12.1 (1.5)	13.0 (2.0)	0.9 (−0.1, 1.8)	0.07
PZA (10 mg/kg) group (n)	6	5
AUC0–8 (h · μg/ml)	11.12 (7.1, 14.3)	14.83 (9.0, 18.8)	1.30 (1.11, 1.50)	0.01
AUC0–24 (h · μg/ml)	28.54 (17.8, 35.9)	40.80 (25.9, 51.9)	1.43 (1.34, 1.52)	<0.001
Cmax (μg/ml)	1.74 (1.1, 2.2)	2.38 (1.5, 2.8)	1.34 (1.26, 1.43)	<0.001
Tmax (h)	4.8 (1.3)	6.8 (1.8)	1.8 (−1.8, 5.4)	0.23
t1/2 (h)	11.6 (0.6)	12.2 (1.2)	0.6 (−1.0, 2.1)	0.35
PZA (25 mg/kg) group (n)	6	6
AUC0–8 (h · μg/ml)	30.20 (21.9, 43.2)	37.59 (30.1, 54.1)	1.24 (1.10, 1.41)	0.01
AUC0–24 (h · μg/ml)	75.55 (54.4, 102.7)	107.23 (84.3, 139.3)	1.42 (1.33, 1.51)	<0.001
Cmax (μg/ml)	4.57 (3.2, 6.2)	6.15 (4.6, 8.4)	1.35 (1.24, 1.47)	<0.001
Tmax (h)	5.3 (2.0)	6.7 (1.6)	1.3 (0.1, 2.6)	0.04
t1/2 (h)	12.5 (2.0)	13.6 (2.3)	1.1 (−0.6, 2.7)	0.15
PZA pharmacokinetic characteristics
PZA (10 and 25 mg/kg) groups (n)	12	11
AUC0–8 (h · μg/ml)	138.81 (75.8, 216.4)	145.69 (86.1, 229.1)	1.02 (0.95, 1.09)	0.55
AUC0–24 (h · μg/ml)	280.67 (156.7, 474.1)	300.65 (175.5, 507.6)	1.04 (1.00, 1.09)	0.05
Cmax (μg/ml)	23.07 (12.5, 41.9)	23.84 (12.6, 36.4)	1.00 (0.90, 1.10)	0.92
Tmax (h)	1.6 (0.7)	1.8 (1.1)	0.3 (−0.4, 1.1)	0.37
t1/2 (h)	9.9 (1.2)	10.5 (1.2)	0.6 (−0.1, 1.2)	0.07
PZA (10 mg/kg) group (n)	6	5
AUC0–8 (h · μg/ml)	92.76 (75.8, 105.3)	93.64 (86.1, 98.8)	1.02 (0.85, 1.23)	0.73
AUC0–24 (h · μg/ml)	186.89 (156.7, 210.5)	192.14 (175.5, 200.5)	1.05 (0.94, 1.17)	0.28
Cmax (μg/ml)	15.29 (12.5, 20.7)	15.24 (12.6, 17.0)	1.00 (0.78, 1.28)	0.97
Tmax (h)	2.0 (0.8)	2.0 (1.2)	0.2 (−1.8, 2.2)	0.79
t1/2 (h)	9.4 (0.6)	10.1 (0.5)	0.68 (−0.51, 1.86)	0.19
PZA (25 mg/kg) group (n)	6	6
AUC0–8 (h · μg/ml)	207.73 (197.8, 216.4)	210.59 (186.5, 229.1)	1.01 (0.97, 1.06)	0.50
AUC0–24 (h · μg/ml)	421.50 (359.3, 474.1)	436.61 (357.3, 507.6)	1.04 (1.00, 1.08)	0.07
Cmax (μg/ml)	34.81 (31.9, 41.9)	34.63 (30.1, 36.4)	0.99 (0.89, 1.11)	0.90
Tmax (h)	1.3 (0.4)	1.7 (1.0)	0.4 (−0.3, 1.1)	0.19
t1/2 (h)	10.3 (1.5)	10.8 (1.6)	0.5 (−0.6, 1.5)	0.30

^aValues are geometric means and ranges (minimum, maximum) for AUC_0–8, AUC_0–24, and C_max and means (SD) for T_max and t_1/2. PK parameters were compared between PZA alone and PZA+ALLO by using a paired-sample t test (t_1/2, T_max, and log-transformed values of C_max, AUC_0–8, and AUC_0–24).

^bThe AUC_0–48 of POA was 62.91 (23.3, 128.1) h · μg/ml (geometric mean [minimum, maximum]) for PZA alone and 93.37 (35.5, 191.6) h · μg/ml for PZA+ALLO; the AUC_0–48 of PZA was 345.37 (182.0, 584.8) h · μg/ml for PZA alone and 365.09 (212.4, 638.6) h · μg/ml for PZA+ALLO (PZA dose levels were 10 and 25 mg/kg). ALLO, allopurinol; GMR, geometric mean ratio; CI, confidence interval.

FIG 2.

Plasma concentrations of pyrazinamide (PZA), pyrazinoic acid (POA), 5-hydroxypyrazinamide (5-OH-PZA), and 5-hydroxypyrazinoic acid (5-OH-POA), with and without allopurinol (ALLO) boosting.

Mean WBA values throughout the 8-h postdose measurement period were above zero (i.e., no bactericidal activity) at all individual time points with 10 mg/kg PZA alone and were marginally below zero with 25 mg/kg PZA alone (Fig. 3a). The mean maximal WBA activity was 0.01 ± 0.19 ΔlogCFU with 10 mg/kg PZA alone and −0.17 ± 0.19 ΔlogCFU with 25 mg/kg PZA alone (Table 2) (P = 0.57 and 0.04, respectively, versus zero change). The mean cumulative WBA at 8 h postdose was 0.02 ± 0.02 ΔlogCFU with 10 mg/kg PZA alone and 0.00 ± 0.02 ΔlogCFU with 25 mg/kg PZA alone (Table 2) (P = 0.95 and 0.39, respectively, versus zero change), indicating no bactericidal activity, although there appeared to be bacteriostatic activity compared to extrapolated values for no treatment (Fig. 3b).

FIG 3.

(a) Mean WBA at individual time points from predose (0 h) to 8 h postdose. (b) Mean cumulative WBA at intervals from predose (0 h) to 8 h postdose. The “without drug (extrapolated)” curve was obtained by assuming that the individual WBA values at the 0-h time point for PZA alone remain unchanged over the subsequent 8-h interval. Error bars indicate 1 standard deviation. PZA, pyrazinamide; ALLO, allopurinol.

TABLE 2.

WBA assay results^d

Group and parameter	Value for visitc		Mean difference (95% CI)	P value
	PZA alone	PZA+ALLO
PZA (10 and 25 mg/kg) groups (n)	12	11
WBAcum(0–8) (ΔlogCFU)	0.01 (0.02)	0.00 (0.02)	0.00 (−0.01, 0.02)	0.49
WBAcum(0–24) (ΔlogCFU)a	0.06 (0.05)	0.01 (0.06)	0.03 (−0.01, 0.08)	0.14
WBAmax (ΔlogCFU)b	−0.08 (0.21)	−0.14 (0.16)	−0.03 (−0.14, 0.09)	0.60
PZA (10 mg/kg) group (n)	6	5
WBAcum(0–8) (ΔlogCFU)	0.02 (0.02)	0.00 (0.02)	0.01 (−0.01, 0.04)	0.23
WBAcum(0–24) (ΔlogCFU)a	0.07 (0.04)	0.00 (0.06)	0.06 (−0.03, 0.15)	0.14
WBAmax (ΔlogCFU)b	0.01 (0.19)	−0.11 (0.20)	−0.08 (−0.32, 0.17)	0.44
PZA (25 mg/kg) group (n)	6	6
WBAcum(0–8) (ΔlogCFU)	0.00 (0.02)	0.00 (0.03)	0.00 (−0.02, 0.02)	0.73
WBAcum(0–24) (ΔlogCFU)a	0.03 (0.06)	0.03 (0.07)	0.00 (−0.09, 0.10)	0.85
WBAmax (ΔlogCFU)b	−0.17 (0.19)	−0.16 (0.14)	0.01 (−0.15, 0.17)	0.86

^aFor WBA_cum(0–24), n = 8 for both PZA alone and PZA+ALLO for both PZA (10 and 25 mg/kg) dosages; n = 5 and n = 4 for PZA alone and PZA+ALLO, respectively, for 10 mg/kg PZA; and n = 3 and n = 4 for PZA alone and PZA+ALLO, respectively, for 25 mg/kg PZA.

^bValues were obtained using all available WBA data time points.

^cValues are means (SD). WBA values were compared between PZA alone and PZA+ALLO by using a paired-sample t test.

^dPZA, pyrazinamide; ALLO, allopurinol; CI, confidence interval.

There was no significant effect of allopurinol coadministration on maximal WBA analyzed with both PZA doses combined or as separate doses (P ≥ 0.44) (Table 2; Fig. 3a). There was also no significant effect of allopurinol on cumulative WBA, the primary outcome parameter, with both PZA doses combined (0.01 ± 0.02 ΔlogCFU for PZA alone versus 0.00 ± 0.02 ΔlogCFU for PZA+ALLO; mean difference = 0.00; 95% confidence interval [CI], −0.01 to 0.02 ΔlogCFU; P = 0.49 based on comparison at 8 h postdose) (Table 2), with the 10-mg/kg and 25-mg/kg doses analyzed separately (Table 2; Fig. 3b), or in analyses including WBA results up to the 24-h time point (Table 2).

There was a significant association between higher POA levels and greater activity in the WBA assay (P < 0.001), although this relationship was evident only at low levels of POA (and with WBA activity in the static rather than cidal range), reaching a plateau at POA levels of 2 to 3 μg/ml (Fig. 4). There was no relationship between cumulative WBA and the AUCs of POA and PZA (Fig. S1).

FIG 4.

Relationship between plasma concentration of pyrazinoic acid (POA) and whole-blood bactericidal activity (WBA). Each point represents an individual blood sample in which both parameters were measured. The relationship between POA plasma concentration and WBA was assessed by use of a nonlinear mixed model with random intercepts, performed using a 4-parameter equation to describe a sigmoid curve, with the equation parameters fitted to observed values by iterative methods, using SAS. The significance of the relationship was determined using the F test. The 95% confidence intervals (CI) of the fitted curve are shown.

DISCUSSION

Overall, we found large increases in systemic POA levels when pyrazinamide was coadministered with allopurinol, but we did not find evidence supporting our hypothesis that this would increase the bactericidal activity achievable with standard doses of PZA. In part, this may be because the WBA assay is relatively insensitive to the effects of PZA (or POA). At the standard clinical dose of pyrazinamide (25 mg/kg), we found only very modest bactericidal activity, in keeping with previous WBA studies (limited by very small sample sizes) (9–11) and with early bactericidal activity (EBA) studies evaluating PZA efficacy (12, 13). We had postulated that WBA assays (in which bacteria are predominantly intracellular [10]) might better reflect the activity of PZA than standard EBA trials can (which mainly demonstrate extracellular bacillus killing), given that PZA has significant intracellular activity (14–16). However, both approaches appear to miss the potent sterilizing activity of PZA seen in long-term clinical trials (17, 18). This may be because neither approach reflects the killing in hypoxic/acidic lung lesions, which are considered to be the main site of PZA and POA activity (1, 19–21).

A more likely explanation for the observed lack of effect of higher systemic POA levels with allopurinol is the relatively poor efficacy of systemic POA compared to that of POA generated at the site of infection by bacterial conversion from PZA, as recently demonstrated in both mouse and rabbit TB models (22). We saw a significant relationship between plasma POA concentration and activity at low POA levels, reaching a plateau at levels above 2 μg/ml. We also found a trend toward enhanced WBA with allopurinol coadministration with the 10-mg/kg PZA dose, but this was not statistically significant (the dosing subgroups were small, and the study was not powered to show a difference within these subgroups). It is possible that when both systemic and bacterially generated POA levels are relatively low (with a reduced dose of PZA), there may be more opportunity for systemic POA to have an impact on efficacy, whereas at higher doses of PZA, the efficacy may already be at a maximum from the high local levels of bacterial POA alone. In the mouse model, the lower efficacy of host-generated POA may be related in part to lesion penetration, although the mechanisms by which PZA and POA enter and act within the bacterial cell are imperfectly understood (1). Given that bacteria in the WBA model are inside macrophages, as opposed to being inside structured lesions in the mouse model, it is possible that lesion penetration may play a less important role in this experimental paradigm.

There were several potential clinical applications that motivated our study. Primarily, we wanted to see if allopurinol might improve the activity of the standard PZA dose, leading to more effective sterilization and potential shortening of the standard TB regimen, a key objective of current treatment research for drug-sensitive TB (23). Our findings do not support the use of allopurinol for this strategic goal. A second clinical concept was that coadministration with allopurinol might allow the use of lower doses of PZA, with increased systemic POA levels supplementing local intrabacterial POA levels to achieve drug activity similar to that with the standard dose while possibly reducing toxicity. WBA activity levels (albeit limited) indeed appeared to be similar between 10 mg/kg PZA plus allopurinol and 25 mg/kg PZA alone. Furthermore, POA (and PZA) levels were substantially lower with this combination than with the standard PZA dose. As PZA side effects, particularly arthralgia, are more closely linked to POA than to PZA, there may be considerable potential for toxicity sparing arising from the reduction in levels of POA (as well as PZA) with this combination. 5-OH-POA has also been identified as one metabolite that may be responsible for PZA-induced hepatotoxicity (24). Coadministration of allopurinol may have the additional benefit of lowering 5-OH-POA concentrations, thereby reducing potential hepatotoxicity.

A third clinical motivation was that boosting systemic POA might present a way of mitigating the decrease in clinical efficacy accompanying pyrazinamide resistance (commonly caused by a mutation in pncA inhibiting bacterial conversion of PZA to POA). We did not test a pyrazinamide-resistant strain in the WBA model, so we cannot draw any direct conclusions about this. However, drawing from our findings with 10 mg/kg PZA, we may reasonably infer a possibility for a direct effect of systemic POA in situations where POA levels generated within bacteria are relatively low. An alternative therapeutic strategy for these clinical scenarios might be to deliver POA or POA salts to the lung by aerosol, as suggested by some investigators (25, 26). There is some recent evidence for the efficacy of this approach in a guinea pig model (27).

In conclusion, we found that allopurinol significantly increased POA levels but did not enhance WBA. This may reflect limitations of the WBA model, which detected only minimal activity of PZA at the standard clinical dose, or the possibility of host-generated POA being less effective than POA generated inside bacteria. The approach of combining low-dose PZA with allopurinol may merit further exploration in a phase IIb clinical trial (with toxicity and culture sterilization endpoints) as a strategy for toxicity sparing or overcoming pyrazinamide resistance.

MATERIALS AND METHODS

We recruited adults between 21 and 70 years of age, identified from a database of healthy research volunteers. Exclusion criteria were pregnancy or breastfeeding, body weight below 50 kg, clinical evidence of active TB, known hypersensitivity to study drugs, use of medication known to interact with study drugs, serum creatinine level or liver enzyme activity above the upper limit of normal, history of hepatitis, gout, alcohol abuse, or hyperuricemia, and a positive test for the HLA-B*5801 allele.

There were two study visits, and participants fasted for at least 6 h prior to each visit. At the first visit, the participant was given a single dose of PZA alone (10 mg/kg or 25 mg/kg, allocated by randomization using preprepared opaque envelopes containing treatment codes). At the second visit, 7 days (window of 6 to 21 days) later, the same dose of PZA was given with allopurinol (4 daily doses, from 2 days prior to the day after PZA dosing). Allopurinol was given as 100 mg daily (Drug Houses of Australia P/L, Singapore). Pyrazinamide (Novartis, Bangladesh) was given according to weight-banded dosage group (8.3 to 12.3 mg/kg for the 10-mg/kg PZA group and 23.1 to 27.3 mg/kg for the 25-mg/kg PZA group). Treatment was open label, and all study drugs were administered orally.

HIV antibody testing and IGRA (QuantiFERON-TB Gold) were done at baseline. Full blood count (FBC), renal function, alanine aminotransferase (ALT), and uric acid levels were measured predose and 48 h after each PZA dose. Participants were asked about symptoms 24 and 48 h following PZA doses. Blood samples were collected predose and at 1, 1.5, 2, 3, 4, 6, 8, 24, and 48 h post-PZA dose at each visit. PK analyses were performed on all blood samples obtained; WBA assays were done for all time points to 8 h and, where feasible, for the 24- and 48-h time points.

Blood samples drawn for WBA assays were kept at room temperature for a maximum of 8 h before being transported to the National University of Singapore biosafety level 3 (BSL-3) laboratory. Blood for PK analysis was centrifuged within 30 min of collection, and separated plasma was frozen at −20°C.

Pharmacokinetic assays.

PZA, POA, 5-OH-PZA, and 5-OH-POA levels in plasma were measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS) on a Sciex Qtrap 4000 triple-quadrupole mass spectrometer using Analyst software combined with an Agilent 1260 high-pressure liquid chromatograph (HPLC). Chromatography was performed with an Agilent Zorbax SB-C₈ column (4.6 by 75 mm; particle size, 3.5 μm), using a reverse-phase gradient elution. Formic acid (0.1%) in Milli-Q deionized water was used for the aqueous mobile phase, and 0.1% formic acid in acetonitrile (ACN) was used for the organic mobile phase. Multiple-reaction monitoring (MRM) of parent/daughter transitions in electrospray positive-ionization mode (ESI⁺) was used to quantify PZA and POA, and negative-ionization mode (ESI⁻) was used for 5-OH-PZA and 5-OH-POA.

Undiluted dimethyl sulfoxide (DMSO) stocks were serially diluted in 50-50 ACN-water to create standard curves and quality control spiking solutions. Twenty-microliter aliquots of undiluted spiking solutions were added to 20-μl aliquots of drug-free plasma, and extraction was performed by adding 180 μl of 50-50 acetonitrile-methanol protein precipitation solvent containing the internal standards. Internal standards were PZA-N15-d3, POA-d3, and pravastatin (5-OH-POA and 5-OH-PZA). Human plasma from Bioreclamation (K2-EDTA) was used to build standard curves. Extracts were vortexed for 5 min and centrifuged (4,000 rpm) for 5 min. The supernatants were analyzed by LC-MS. All reference standards were sourced from Toronto Research Chemicals. The following MRM transitions were used: 124/81.1 for PZA, 125.2/81.1 for POA, 138.0/95.0 for 5-OH-PZA, 139.0/95.0 for 5-OH-POA, 128.2/84.1 for PZA-N15-d3, 128.2/84.1 for POA-d3, and 423.2/101.0 for pravastatin. Sample analysis was accepted if the concentrations of the quality control samples were within 20% of the nominal concentrations.

PK parameters were estimated by a model-independent method. The areas under the concentration-time curve from 0 to 8 h (AUC_0–8), 0 to 24 h (AUC_0–24), and 0 to 48 h (AUC_0–48) postdosing were calculated according to the linear up/log down trapezoidal rule. The terminal half-life (t_1/2) in plasma was calculated from the elimination rate constant (k_el), estimated as the slope of the log-linear terminal portion of the mean plasma concentration-time curve, by linear regression analysis.

WBA assay.

The WBA assay methodology was adapted from previously published methods (10, 11). In brief, a standard stock for all experiments was made using Mycobacterium tuberculosis H37Rv grown in 7H9 medium to mid-log phase and then frozen in 10% glycerol aliquots at an optical density (OD) of 1.0 at −80°C. A standard curve relating the time to positivity (TTP) in culture to the volume of the stock was generated by performing serial 5-fold dilutions of stock (25 to 25 × 10⁻⁸ μl) in 7H9 medium in duplicate, inoculating the dilutions into MGIT tubes, and incubating the tubes in an MGIT960 detection system (Becton Dickinson, Franklin Lakes, USA).

The volume of mycobacterial stock calculated from the standard curve to give a TTP of 5.5 days (0.5 μl) was added to heparinized blood (300 μl) and topped up with sufficient tissue culture medium (RPMI-GlutaMAX) to bring the total culture volume to 600 μl. Cultures were set up in sealed screw-cap tubes and incubated at 37°C with slow constant mixing for 72 h. Following incubation, cultures were centrifuged at 12,000 rpm for 5 min, and the liquid phase was removed. Blood cells were lysed by adding 1 ml sterile water to the pellet and vortexing for 10 min. Samples were centrifuged again (12,000 rpm) for 10 min, and the supernatant was discarded. The pellet was resuspended in 500 μl of 7H9 medium and inoculated into MGIT tubes, and the TTP was recorded (to the nearest minute). WBA cultures were set up in duplicate at every time point, and the mean TTP was calculated. Control cultures (in duplicate) were set up on the same day by inoculating the standard volume of stock directly into MGIT tubes. The WBA at each of the individual sample time points was obtained from the difference between the log of the volume on the standard curve that corresponded to the TTP for that time point and the log of the volume corresponding to the TTP of the control culture. This is equivalent to the difference in log bacterial CFU between the sample and the control, reported as ΔlogCFU.

Sample size and statistics.

The original study design comprised 3 PZA dose cohorts (10 mg/kg, 25 mg/kg, and 35 mg/kg). The first two cohorts were to be recruited by randomized allocation; the third, nonrandomized cohort (35 mg/kg) was optional and would proceed after review of safety data from the initial two cohorts. For the purposes of sample size calculation, we assumed that the effect of allopurinol would not differ according to the dose of pyrazinamide, and we therefore planned to combine all dose cohorts in the analysis. We estimated the sample size to detect a difference in cumulative WBA of 0.1 ΔlogCFU between the PZA and PZA+ALLO visits by using a paired-sample t test (two-tailed), assuming a standard deviation (SD) of 0.1 ΔlogCFU for the difference, based on previously published data (28). With 6 participants in each cohort, we would have a power of >90% to detect the specified change if all 3 cohorts were enrolled (total sample size of 18) and would retain >80% power if 2 cohorts were enrolled (total sample size of 12).

Following the recently published results of an animal model study suggesting that host-derived POA does not contribute to bacillary killing (22), the planned review of safety data (after completion of the initial 2 cohorts) was extended to review preliminary efficacy data. A decision was made not to proceed with enrolling the final cohort based on the efficacy review.

PK parameters were compared between the PZA and PZA+ALLO visits by using a two-sided t test for paired data (C_max and AUC data were log transformed). Individual WBA values were expressed as the total bacterial killing over the total 72 h of the assay. The maximum WBA (WBA_max) for each participant was the lowest of the individual WBA values (lower values indicate greater antibacterial activity). The cumulative WBA (WBA_cum) was calculated for each participant from the area under the curve of the individual WBA values measured up to that time point, determined using the trapezoidal method (29). An estimate of the cumulative WBA without drug was obtained by extrapolation from the WBA value at the 0-h time point prior to the first PZA dose for each participant, with the assumption that this value remained unchanged over the subsequent 8-h interval. For the purposes of this calculation, the individual WBA values were converted to a rate of kill per hour (assuming that this rate was constant over the 72-h period of each WBA culture). The time factors cancel in the calculation of area under the curve, so WBA_cum is expressed as the absolute kill by the specified time point after dosing. For analysis of bactericidal activity of PZA alone, WBA_max and WBA_cum were compared to zero by using a one-sided one-sample t test. For comparison of bactericidal activities between the PZA and PZA+ALLO visits, WBA_max and WBA_cum were compared by using a two-sided paired-sample t test.

The relationship between WBA levels and POA concentrations at all sample time points at which both parameters were measured was assessed by use of a nonlinear mixed model with random intercepts, performed using a 4-parameter equation to describe a sigmoid curve, with the equation parameters fitted to observed values by iterative methods, using SAS, version 9.4.

Ethical and regulatory approvals.

This study is registered under ClinicalTrials.gov registration no. NCT02700347. The National Healthcare Group's Domain Specific Review Board (NHG-DSRB) and Health Sciences Authority (HSA), Singapore, approved the study, and all participants provided written informed consent.