Urinary ATP as an indicator of infection and inflammation of the urinary tract in patients with lower urinary tract symptoms

“Lower Urinary Tract Symptoms” (LUTS) is a collective term describing [1] urinary storage problems such as frequency, urgency and urge incontinence; [2] voiding difficulties such as hesitancy, reduced stream, intermittency and incomplete voiding; [3] sensory symptoms that include various experiences of pain; and [4] stress urinary incontinence. There is considerable overlap between these symptoms [1] so that diagnostic categorisation is difficult. Whatever the symptom mix, the exclusion of urinary tract infection (UTI) is a mandatory first step in the assessment of all LUTS [2]. Whilst acute UTI is not diagnostically challenging, in the case of LUTS without acute frequency and dysuria, exclusion of infection poses a diagnostic challenge.

There are good reasons to scrutinise urinary adenosine-5′-triphosphate (ATP) as a possible surrogate marker of UTI since the reliability of popular diagnostic methods used to exclude UTI have been questioned [3-6]. Published guidelines across Europe, USA and the UK reveal significant discrepancies in the choice of a quantitative threshold used to define significant bacteriuria. The clean-catch, midstream urine (MSU) sample culture in the UK and Europe commonly uses the Kass (1957) [7] criterion of 10⁵ colony forming units (cfu) ml^-1 of a single species of a known urinary pathogen. Kass drew these data from 74 women with acute pyelonephritis and 337 normal controls, a select sample unrepresentative of wider lower urinary tract symptoms (LUTS). Despite its limitations, this criterion has become a ubiquitous reference standard and has been challenged by several groups [6,8]. The European Associate of Urology (EUA) guidelines for urological infections emphasise that no single threshold can be applied in all clinical situations. The urinary dipstick tests for nitrite and leucocyte esterase are also commonly used as a bedside screening test for infection and as a measure of a positive urine culture. However the use of urinary dipstick have been validated against the urine culture to a threshold of 10⁵ colony forming units (cfu) ml^-1 and given the recent criticism of the Kass criterion, urinary dipstick have also recently been found to be unreliable [3,4,9].

Urinary tract ATP has attracted intense interest in the last 30 years for its pharmacological and pathophysiological associations. There are great hopes that purinergic receptor manipulation might influence detrusor motor function and urothelial afferents [10], achieving therapeutic benefit. ATP is an important urothelial cell distress signal [11] and is released by inflammatory cells and bacteria [12]. Urinary tract infection featuring bacterial invasion, urothelial distress and an innate immune response involving recruitment of inflammatory cells, should be associated with increased urinary ATP levels. Indeed, high levels of ATP have been detected in the urine of patients with interstitial cystitis and acute UTI with a positive urine culture [13]. Increased ATP has also been shown to be released from cultured urothelial cells infected with uropathogenic E.coli (UPEC) [14], and UPEC also produce ATP when cultured in vitro [15]. It has been postulated that ATP may reflect microbial biomass and hence ATP increases as the amount of bacteria present increases. Currently ATP levels are used widely in the food, water and sanitation industry as a measure of bacterial contamination [16].

Given the problems with current tests [3], developing alternative diagnostic assays is a high priority. We therefore sought to scrutinise the performance of urinary ATP to test for potential as a surrogate measure of inflammation and infection when assessing patients with chronic LUTS. The experiment was divided into two parts; [1] a clinical experiment that evaluated urinary ATP in patients with LUTS and controls, comparing urinary ATP with symptoms, microscopic pyuria and urine culture results; and [2] a laboratory experimental series that explored the factors that could influence sample collection, storage and preservation. As urine contains native ATPase activity, the time-decay curve of urinary ATP from collection to processing was evaluated. Boric acid crystals, which are commonly used as a urinary preservative, have been shown to prevent microbial swarming [17] and boric acid has a preservative influence on white cells [18]. We therefore studied the effects of the use of urinary preservative boric acid, storage temperature and the effect of centrifugation on urinary ATP concentration.

We compared urine samples from 75 healthy controls and 340 patients presenting with LUTS. The demographic data can be seen in Table 1. The patients cohort was grouped in the first model; with pyuria (≥1 wbc μl^-1) or without pyuria (0 wbc μl^-1). For the second model we used categorical scaling so that we compared pyuria 1-9 wbc μl^-1 (n = 120) and pyuria ≥10 wbc μl^-1 (n = 120) with a baseline factor of zero pyuria (0 wbc μl^-1). Of those with LUTS, 33.3% had only OAB symptoms, 4.1% had pain alone, 3.7% had only stress incontinence and 13.2% had only voiding dysfunction. Patients had a median of 3.5 LUTS (quartile range 1 to 6). The overlap of symptoms is illustrated in Figure 1.

Log₁₀ transformation of ATP changed the skewness from 4.2 to -0.3 and kurtosis from 27.5 to 1.1. The results of the two regression analyses are shown in Table 2. It can be seen that female gender was associated with higher predictions of the log₁₀ ATP. However given the small number of males and predominance of female patients, which is a reflection of this condition, limited discriminatory power precludes extrapolation. Voiding symptoms were also predictive of higher log₁₀ ATP. Interestingly, voiding symptoms in both sexes have been reported to be associated with inflammatory disease of the lower urinary tract [22]. A positive culture result predicted lower log₁₀ ATP. The first regression model shows that the presence of any pyuria was not predictive of higher log₁₀ ATP. The second regression model demonstrates that a pyuria ≥10 wbc μl^-1 was predictive of a higher log₁₀ ATP; however, lower levels of pyuria 1-9 wbc μl^-1 were not. These data do demonstrate that urinary ATP is elevated in association with inflammation but that urinary ATP lacks the discriminating properties at lower levels of pyuria, which are necessary for a useful clinical surrogate marker. This is illustrated by the ROC analysis which showed an area under the curve of 0.6 for a positive culture; 0.5 for pyuria > 0 wbc μl^-1 and 0.6 for pyuria ≥ 10 wbc μl^-1 (Figure 2). These data imply that there is no useful diagnostic role for this assay. The regression analyses showed that, age, average 24-hour frequency, average 24-hour incontinence, the number of urgency, stress incontinence, and pain symptoms provided no substantial explanation of the variance of urinary log₁₀ ATP.

ATP has been proposed as a potential clinical marker of infection for both acute and chronic LUTS [10,27]. To avoid premature use, when assessing the diagnostic potential of test, it is important to assess first whether the measure explains the symptoms and other manifestations of the disease of interest [28]. The data published here demonstrate that urinary ATP does not offer any additional benefit to tests currently used in screening for infection in patients presenting with LUTS, and therefore does not show promise for future development of a diagnostic test for this particular group. This observation is confirmed in the ROC curves that explore sensitivity and specificity properties.

Our results demonstrate that patients with LUTS and any pyuria do not manifest a significantly raised urine ATP concentration. It is only when there is pyuria of greater than 10 wbc μl^-1, which is a currently used marker of urinary infection, that there is a significantly raised urinary ATP. However, this signal was not a discriminating marker for low levels of lower urinary tract inflammation (pyuria 1-9 wbc μl^-1) as shown by our second regression model, and it is in these clinical circumstances that there is greatest need for novel clinical surrogate markers. It has been shown that low-level pyuria [26], voiding symptoms [22], overactive bladder symptoms [24] and pain symptoms [23] all correlate with urinary infection. Only voiding symptoms explained a small amount of the variance in ATP. These findings are therefore discouraging.

Counter-intuitively, urinary ATP was lower, given a positive urine culture. This unexpected result may reflect the fact that there were so few positive cultures (16.9%) that there weren’t enough data to draw meaningful conclusions. Increased microbial dephosphorylation of ATP to adenosine, which we did not assay, may also be a contributing factor. In addition, the urinary ATP reflects the urinary microbial biomass directly post void and there may be significant variation in the urinary biomass at the time of culture secondary to transportation and processing delay.

There was an unbalanced sample size with fewer control participants with a lower average age, and this reflects the common difficulty of finding older subjects without any urinary tract symptoms. The controls in this experiment were all asymptomatic. We did control for these differences in the analysis but would nevertheless wish to avoid wide inferential generalisations. The essence of this study was to scrutinise the discriminating properties amongst patients. The recruitment field was also predominantly female which may explain the gender difference. The small numbers of males were nevertheless included in the analysis as the model was sufficiently powered for this variable, which required ≥20 for each independent variable. The sample was powerful enough to detect the influence of voiding symptoms, pyuria and culture, despite the wide variance. Urinary ATP concentrations proved independent of age and the number and nature of LUTS symptoms, other than voiding symptoms. The association with voiding symptoms was not surprising because in patients with chronic LUTS, these have been reported to be correlated with pyuria in both genders, in clear contrast to OAB and pain symptoms [22].

We found that urinary ATP could not pick out low levels of inflammation (pyuria 1-9 wbc μl^-1) when used alone, a situation where clinicians need most help. Current dipstick methods and direct urinary microscopy are able to identify pyuria when it is abundant [26], so urinary ATP would offer no additional advantage. There can be little doubt that ATP plays a very significant role in the pathophysiology of urinary tract disease [29,30] but it would seem that this does not translate into a useful role as a clinical test when used alone.

The laboratory experimental series showed that urine samples should be processed immediately or frozen and stored at below -20°C. We demonstrated significant urinary ATP decay with time when samples were left at room temperature, with the rate being substrate dependent. We also found that storage of urine at -20°C is adequate and this therefore allows for wider use of standard freezers as opposed to the specialised -80°C devices. The centrifuge studies examined the effect of the removing cellular material. The supernatant showed marginally lower levels of ATP. This could be attributed to increased ATPase activity from burst cells or it may have resulted from delayed freezing, allowing additional time for enzyme activity. Additionally, centrifugation results in the removal of biomass and hence this may explain the lower levels of urinary ATP. This marginal difference, though statistically significant, may not be clinically significant; nevertheless, centrifuging of urine prior to freezing or analysis is an additional labour worth avoiding. Boric acid proved counterproductive and so should not be used as a preservative.

In summary these data discourage the idea that urinary ATP should be developed as a clinical surrogate test for UTI in patients presenting with LUTS. This assay does not appear more effective than markers used in current clinical practice [3,26]. However, abundant urinary ATP is certainly evident amongst patients with lower urinary tract symptoms and dysfunction and these data do encourage continued interest in the pharmacology and pathophysiology of purinergic functions in the bladder. For those committed to these avenues of discovery, we provide valuable data on human sample processing, storage and assay.

In conclusion, urinary ATP does not improve on the use of microscopy as a surrogate marker of urinary tract infection in patients presenting with LUTS and is therefore not a promising clinical diagnostic marker. We need to continue to explore other potential markers that can be used to screen LUTS patients for UTI, particularly applicable to those that have lower levels of pyuria, where significant disease may currently be overlooked [26]. The relevance of measuring ATP to study the pathophysiology of the lower urinary tract is still very evident and in that context urine is a useful biological sample.