Hypothesis / aims of study
A pressure flow study (PFS), part of the ICS standard urodynamic test, is the gold standard for the classification and quantification of bladder outflow obstruction (BOO). For men with benign prostatic hyperplasia (BPH), the minimal urethral opening pressure (pmuo) is considered the most objective parameter describing BOO. pmuo, is, consequential to the distensible collapsible tube uro(hydro)dynamic theory or paradigm, found at the termination of (male) voiding. Therefore, pmuo is, as seen from the (patho)physiology perspective, the most reliable parameter and the most independent of detrusor voiding contraction strength[1]. However, in clinical practice pmuo is difficult to determine because of misalignement of intravesical pressure and the flow entering the flowmeter with a relatively large delay. For that reason, several methods were developed in the past to estimate pmuo, and thereby BOO. The maximal flow (Qmax) and the corresponding pressure (pdetQmax), both more synchronous and unambiguously (automatically) detectable than pmuo, were used for extrapolation. These methods include linPURR with nomogram, URA, and BOOI. In addition, a method was developed including not only pdetQmax but also all ‘lowest pressure’-data points after Qmax to further increase the accuracy of the estimation of pmuo; the three-parameter model (3PM). The four different methods have never been mutually compared. The aim of this study was to compare the accuracy of these four methods in determining pmuo, using a large database of pressure flow measurements.
Study design, materials and methods
2349 PFS of men referred with lower urinary tract symptoms (LUTS) without earlier lower urinary tract surgery between 2003 and 2020 were initially included. Measurements with a Qmax of > 35 ml/s or < 2 ml/s (4.3%) were excluded. Also, PFS with a voided volume < 100 ml (9.6%), and PFS with a pdetmax during voiding < 20 cmH2O or > 200 cmH2O (0.8%) were excluded from further analysis. Lastly, studies with catheter artifacts (e.g. kinking or slipping out) (12.9%) were also removed from the dataset.
From the 1717 resulting studies, pmuo was calculated according to Schäfer linPURR (PmuolinPURR), URA (PmuoURA), and BOOI (PmuoBOOI). In addition, a 3PM fit was performed through the lowest pressure flank of the pressure-flow plot, using the formula, pdet = pmuo + A*Qk, with 2/3≤k≤2 and A being a patient-specific factor, including all possible theoretical relations between Qmax/pdetQmax and pmuo, as derived by Spangberg (Pmuo3PM) [2]. The resulting pmuo’s were compared with the actual pmuo for every PFS. For this study, we defined the actual pmuo (PmuoAct) as the average pdet associated with the flowrate between 1 and 0.5 ml/s at the end of voiding. We considered this the best way to reduce the pressure (with flow delay) -artifacts of the pmuo. For the comparison in this study, the four estimated pmuo’s were adjusted to represent the pressure at 0.75 ml/s as well. Differences between the predicted pmuo’s and the PmuoAct’s per PFS were analyzed.
PFS with a flow variation smaller than 1 ml/s and a pressure variation < 5 cmH2O after Qmax were expected to show the most realistic pmuo values, not confounded by abnormal dynamics or interruptions of voiding. A subselection of 376 (21.9% of all) PFS with this minimal variation throughout the voiding phase was selected and used in a secondary analysis.
Results
Of the 1717 studies, 55 studies (3.2%) were excluded because all pressures after Qmax were found higher than pdetQmax and therefore 3PM analysis was impossible. Pmuo3PM was found most accurate in predicting pmuo, with 75.9% of the estimations found within a range of +10/-10 cmH2O around PmuoAct. PmuoURA (52.0%) and PmuolinPURR (53.6%) showed similar performance, while PmuoBOOI was found significantly less accurate (40.0%) to predict PmuoAct (N-1 Chi-squared test p<0.025 when compared to all others). Within the minimal variation subset, the overall accuracy increased, with 93.0% of Pmuo3PM estimations within a bandwidth of 10 cmH2O. PmuoURA (65.1%) and PmuolinPURR (57.0%) were found not significantly different, while PmuoBOOI (45.1%) again showed a smaller correctly estimated proportion.
Bland-Altman plots were created in which the average of one of the estimation methods with PmuoAct was compared with their differences, see figure 1. A significant, negative linear regression was seen for PmuoBOOI, (1B) indicating that PmuoBOOI tended to be higher when the average Pmuo is larger. The opposite, although less prominent, was seen for PmuoURA (1C). No significant regression was found for Pmuo3PM and PmuolinPURR, but the 95% confidence interval was smaller for Pmuo3PM (1A). The subset analysis showed similar results, with a much smaller confidence interval (+- 15cmH2O, compared to +-30 for the complete set) for Pmuo3PM and a non-significant linear regression for PmuoURA.
Interpretation of results
Although URA, BOOI, linPURR, as well as 3PM, are mentioned in the ICS standard for PFS a ‘quality ranking’ was not included, and a comparison of their ability to predict pmuo has never been reported. We found that the 3PM was superior in predicting Pmuo to three well-known one-parameter methods, although not applicable in 3.2% of the cases and requiring more complex calculations. From the one-parameter methods, Schäfers linPURR was found to be most accurate over the whole range of pmuo. BOOI, the current, but still provisional ICS standard, was found least accurate, with a tendency of overestimating pmuo in men with a higher grade of BOO. The clinical applicability should be studied further and cut-off values for BOO should be reformulated for the 3PM method.