The effectiveness of sputum pH analysis in the prediction of response to therapy in patients with pulmonary tuberculosis

Introduction

Pulmonary tuberculosis (PTB) remains a major cause of death worldwide, accounting for 16.6% of newly diagnosed patients with PTB (an estimated 9.0 million) (World Health Organization, 2015a: Global Tuberculosis Report 2014) Antituberculous therapy using the combination of isoniazid (INH), rifampicin (RFP), ethambutol (EB), and pyrazinamide (PZA) has been established as an effective 6-month standard therapy, leading to a decrease in the global incidence of PTB (World Health Organization, 2010: Guidelines for treatment of tuberculosis). However, predictive factors of response to therapy have not been fully elucidated.

Among the standard therapy drugs, PZA has the potential to improve early bactericidal activity of other anti-tuberculosis combination drugs, such as INH-RFP or INH-streptomycin (Jindani et al., 1980). Hence, PZA has been considered a key drug, although the mechanism underlying this additive effect is unclear (Diacon et al., 2012; Gillespie et al., 2014; Jindani et al., 2014). In contrast, as PZA could raise concerns about liver dysfunction, especially when administered to elderly patients, the regimen without PZA would be initiated in clinical settings for safety reasons (Schaberg, Rebhan & Lode, 1996). However, patients treated without PZA receive a longer treatment period, which could raise other complications and poor adherence. Thus, the clarification of characteristics of patients who would be recommended for the PZA-including regimen would be important.

Based on the evidence showing that PZA had high bactericidal activity under acidic conditions (Salfinger & Heifets, 1988; Zhang et al., 1999), we hypothesized that monitoring airway pH could predict the response to therapy including PZA. The measurement of airway pH using exhaled breath condensate (EBC) has been reported to be useful for evaluating disease severity in patients with bronchial asthma and chronic obstructive pulmonary disease (COPD) (Antus et al., 2010; Kostikas et al., 2002; Papaioannou et al., 2011; Tseliou et al., 2010). Since collecting EBC in patients with PTB would raise safety concerns, we used fresh sputum samples. Our preliminary experiments demonstrated reliable and reproducible pH values in small amounts of sputum samples measured by a high-sensitive pH monitoring system.

This study examined the associations between sputum pH and length to sputum smear and/or culture conversion for the purpose of establishing markers predicting favorable outcomes in patients receiving standard 6-month antituberculous therapy.

Materials & Methods

Results

Analysis of factors affecting sputum pH

The sputum pH was uniformly-distributed between 5.50 and 8.37, with a median value of 7.00, from 39 included patients (Fig. 1 and Table 1). First, we examined whether the sputum pH could be reproducible under the conditions of (1) different time points after sample collection and (2) different temperatures of sample preservation. Our preliminary studies showed that the value of sputum pH did not differ up to 6 h after collection, regardless of storage temperature (Fig. 2). Of particular importance, there was no significant change in sputum pH before and 2 months after initiating treatment with HRZE (Fig. 3, p = 0.68, n = 19). The potential clinical characteristics associated with sputum pH were analyzed and summarized in Table 2. In univariate analysis, age ≥61 years was a statistically important factor predicting low pH value (p < 0.01 vs. aged <61 years). Of note, there was an inverse association between age and sputum pH value (r = − 0.56, p < 0.01, n = 39). Unexpectedly, smoking history and extensive lung lesions did not affect the sputum pH values. The use of an acid-suppressant (p = 0.048), being immunocompromised (p = 0.096), the presence of cavities (p = 0.105), and sputum bacterial load (p = 0.111) were potential predictors of sputum pH (Table 2). Multivariate analysis identified that only age ≥61 years was an independent predictor of low sputum pH (p = 0.005; odds ratio (OR) 24.535; 95% CI [2.685–224.213].

Table 2:

Univariate and multivariate analyses of odds ratio for an initial low sputum pH (<7.00).

	Univariate analysis			Multivariate analysis
Characteristics	OR	95% CI	P e	OR	95% CI	P f
Age ≥61 yr	8.750	2.100–36.251	0.004	24.535	2.685–224.213	0.005
Male sex	2.291	0.613–8.498	0.320
Smoking history (current/ex-smoker vs. never)	0.938	0.214–4.108	1.000
Use of acid-suppressanta	4.444	1.115–17.497	0.048
Immunocompromisedb	3.600	0.905–14.132	0.096
Extensive lung lesionc	1.697	0.466–6.165	0.514
Presence of cavities	0.300	0.080–1.130	0.105
Sputum appearanced(M vs P)	0.542	0.134–2.219	0.480	0.104	0.010–1.128	0.063
Sputum bacterial load (<2+ vs ≥2+)	3.208	0.877–11.719	0.111

DOI: 10.7717/peerj.1448/table-2

Notes:

CI

confidence interval

OR

odds ratio

a Acid-suppressant including histamine-2 receptor antagonist and proton pump inhibitor.

b Immunocompromised; Patients having diabetes mellitus and/or corticosteroid user.

c Extensive lung lesion; Radiological extent of parenchymal disease over one whole lung.

d Sputum appearance classified by Miller and Jones’ classification.

e Fisher’s exact test.

f ^¶ Logistic regression.

Analysis of sputum pH and outcome

We further analyzed the association between initial sputum pH and clinically important outcome, including the time to, (1) smear conversion and (2) culture conversion of sputum, as useful indicators of response to treatment with HRZE. Among the 39 patients included, 5 patients showing persistent smear-positive results fulfilled the discharge criteria. Therefore, the remaining 34 patients were examined as evaluable smear-conversion subjects. There was a moderate positive relationship between initial sputum pH and time to smear conversion (r = 0.342, p = 0.048, n = 34). Similarly, initial sputum pH was modestly correlated with time to culture conversion (r = 0.304, p = 0.060, n = 39). Further, when patients were divided into initial sputum pH <7.00 vs. ≥7.00 (median for whole group) groups, the low pH group showed significantly shorter hospital stay as determined by time to smear conversion (29.6 ± 31.0 vs. 61.5 ± 32.5 days [mean ± SD], p = 0.028, log-rank test) or alternatively, time to culture conversion (30.5 ± 17.9 vs. 51.4 ± 19.4 days [mean ± SD], p = 0.007, log-rank test) as shown in Fig. 4. Since age was shown to be significantly correlated with sputum pH (Table 2), we evaluated the effect of age on outcome analysis. Also, known factors such as sputum bacterial load, extent of lung lesion, and cavity formation have been thought of as biomarkers to identify PTB patients at risk of longer hospital stays and relapse (Fortun et al., 2007; Hesseling et al., 2010; Horne et al., 2010), and thus be included in multivariate Cox regression analysis. Interestingly, although age ≥61 years is also a factor affecting initial low sputum pH, the lack of significance between age and response to therapy is identified (Tables 2 and 3). This is convincing because elder PTB patients exhibit higher mortality (Feng et al., 2011). Possible reasons of this dissociation are thought to be treatment interruption and/or discontinuation due to multiple organ dysfunctions caused by initiating therapy in elder patients. However, 14 (30%) of the 47 patients enrolled in the current study were aged 75 years or older, and the majority (n = 12, 86%) of these patients could continue HRZE, and could be discharged from hospital. Thus, older age may not always associate with higher mortality in PTB. Importantly, other potential factors such as sputum bacterial load, extensive lung involvement, and cavity formation did not affect the outcome analysis using a multivariate Cox regression model in our cohort (Table 3). Accordingly, sputum pH was found to be the most powerful independent predictor of the time to both sputum smear and culture conversion in patients receiving the standard 2-month HRZE followed by a 4-month HR therapy (Table 3, p = 0.034 and 0.019, respectively).

Table 3:

Cox regression analysis for baseline predictors of sputum smear and culture conversion in patients.

	Predictor variable	Hazard ratio	95% CI	P
Sputum smear conversion (days)	Age ≥61 yr	1.910	0.724–5.036	0.193
	Sputum bacterial loada	1.465	0.644–3.334	0.366
	Extensive lung lesion b	0.543	0.221–1.334	0.185
	Presence of cavities c	1.825	0.845–3.941	0.128
	Sputum pH ≥7.0	3.094	1.093–8.760	0.034
Sputum culture conversion (days)	Age ≥61 yr	1.556	0.724–3.347	0.260
	Sputum bacterial load a	1.390	0.650–2.974	0.398
	Extensive lung lesionb	0.723	0.322–1.623	0.434
	Presence of cavities c	2.044	0.894–4.673	0.092
	Sputum pH ≥ 7.0	2.717	1.183–6.240	0.019

DOI: 10.7717/peerj.1448/table-3

Notes:

CI

confidence interval

a Dummy variables for sputum bacterial load: 0 for bacterial load <2+, 1 for bacterial load ≥2+.

b Dummy variables for extensive lung lesion: 0 for lung lesion < one whole lung, 1 for lung lesion ≥ one whole lung.

c Dummy variables for presence of cavities: 0 for no cavity, 1 for presence of ≥ one cavity.

Discussion

The sputum sample provides important information in both infectious and non-infectious pulmonary diseases (Dimakou, Hillas & Bakakos, 2009; Kodric et al., 2007; Ugarte-Gil et al., 2013). Previous reports have documented that the analysis of pH in sputum or EBC could monitor the inflammatory status in various lung diseases, and might reflect the success of subsequent therapy (Antus et al., 2010; Hunt et al., 2000; Kostikas et al., 2002; Papaioannou et al., 2011). Although low pH in sputum or EBC indicates airway acidity, which disadvantageously affects host defense and immune activation (Sutto, Conner & Salathe, 2004; Trevani et al., 1999), several antibiotics, such as PZA, have more bactericidal activity in acidic conditions (Salfinger & Heifets, 1988; Zhang et al., 1999). Based on these facts, we hypothesized that measuring pH in sputum or EBC would be useful for predicting response to therapy in PTB, because they require long-term treatment with careful management of several side effects.

One serious concern was accidental exposure to medical staff, especially when collecting the EBC samples. In contrast, sputum samples can be easily collected and mandatory assessment of isolation in patients with smear-positive active PTB. In addition, measuring pH is quite easy using an electrode or even by dipstick test. Since there have been few reports on the analysis of sputum pH, we needed to establish an appropriate procedure of measurement and analysis.

Our preliminary experiment revealed that the pH value of sputum was uniformly distributed, and was reproducible up to 6 h under either cold conditions or room temperature (Figs. 1 and 2). Unfortunately, we could not establish the control value because collecting sputum samples from healthy subjects was quite difficult. However, EBC or induced sputum from healthy subjects revealed the normal pH value around 7.5–7.7. (Kodric et al., 2007; Kostikas et al., 2002; Vaughan et al., 2003). Our PTB patients showed a relatively acidic airway environment (pH = 7.02 ± 0.89). Low pH values of EBC and induced sputum were also addressed in patients with asthma, COPD, and acute lung injury, where such values were associated with resistance to therapy (Antus et al., 2010; Gessner et al., 2003; Papaioannou et al., 2011). However, current findings show, for the first time, that an initial low pH value of sputum in PTB patients was the most powerful indicator predicting a favorable response to standard combination therapy (Table 3 and Fig. 4).

Next, we considered the reason for good prognosis in PTB patients with low pH sputum. A potential benefit of low pH in the lung lesion is that PZA has much more bactericidal activity in acidic conditions (Salfinger & Heifets, 1988). When administered in an acidic lesion, PZA is easily degraded into an activated form of pyrazinoic acid and accumulated in bacterial cytoplasm, which facilitates bactericidal activity (Zhang et al., 1999). Second, bacteria, even tuberculosis, is thought to be intolerable to acidic conditions (Piddington, Kashkouli & Buchmeier, 2000). Acidic conditions and the administration of PZA could act synergistically to kill tuberculosis, and thus lead to favorable shorter hospital stays in PTB patients presenting with low sputum pH. Our findings could support the establishment of a future new shorter regimen including PZA and the selection of eligible patients for a PZA-including regimen.

The current study has several limitations. First, the number of patients is limited, and the patients were recruited from a single hospital, limiting the generalizability of treatment regimen and results. Actually, we have tried to apply intensified treatment with HRZE even in patients older than 80 years, and 91.4% (n = 43) of eligible patients initiated therapy with HRZE, and 90.7% (n = 39) of them successfully completed 2 months HRZE without serious adverse events. The remaining patients (17.0%, n = 8) could not initiate therapy including PZA, or discontinued PZA due to liver dysfunction. We could not compare the outcome with or without PZA, in this relatively small group. Second, our cohort consists of single ethnic Japanese patients with favorable susceptibility to INH, RFP, EB, and PZA. That is, relatively low prevalence of drug resistance PTB (approximately 3.2% to any drug including INH, RFP, and EB among newly diagnosed patients in Japan) should be considered (Tuberculosis Research Committee , 2015). Thus, our findings may not be applicable to other unique populations. Third, we could follow and collect samples until patients fulfilled the discharge criteria. Therefore, we could not determine if their sputum pH reversed to normal, around 7.7–8.0, after completing treatment. Since this should be clarified, we are planning to monitor long-term sputum pH with a large number of PTB patients as a validation cohort from another center.

Conclusions

Airway acidity is easily monitored using sputum samples, which might enable us to predict response to standard intensified therapy of HRZE in patients with PTB.

Supplemental Information