Potential Predictors for Elevated Post Void Urine Volume Following OnabotulinumtoxinA Treatment for Idiopathic Overactive Bladder

Franco I1, Schwartz M2, Patel A3

Research Type


Abstract Category

Overactive Bladder

Abstract 157
E-Poster 1
Scientific Open Discussion ePoster Session 7
Wednesday 4th September 2019
13:05 - 13:10 (ePoster Station 9)
Exhibition Hall
Overactive Bladder Mathematical or statistical modelling Voiding Dysfunction Urodynamics Techniques
1.Department of Urology, Yale School of Medicine, 2.MS Biostatistics, 3.Allergan plc

Israel Franco



Hypothesis / aims of study
OnabotulinumtoxinA (onabotA) has been demonstrated to be an effective treatment for patients with idiopathic overactive bladder and urinary incontinence. While generally well-tolerated, treatment with OnabotA has been shown to be associated with a transient increase in post void residual volume (PVR) that, in some cases, may require the use of clean intermittent catheterization (CIC). The ability to predict which patients could be at increased risk for developing elevated PVR following onabotA treatment would allow physicians to counsel patients more specifically, and also to tailor how closely they monitor patients who may be at an increased risk of developing a raised PVR. This exploratory analysis was designed to evaluate several possible predictors for post-treatment elevated PVR.
Study design, materials and methods
Data from patients from several sites in the United States who had been treated per local practice with onabotA for idiopathic overactive bladder were retrospectively captured and analyzed. Only data from patients receiving their first treatment with onabotA for overactive bladder were used in this analysis. The original intent of this preliminary analysis was to evaluate potential risk factors for developing a sufficiently high post-treatment PVR that would require CIC. In total, 211 patients had evaluable post-treatment PVR data; maximum values for post-treatment PVR; <200 mL (n=173), 200–299 mL (n=25), 300–399 mL (n=7), and ≥400 mL (n=6). As incidences of markedly elevated PVR were comparatively infrequent, a post-treatment PVR value of 200 mL was chosen as the threshold. This analysis sought to determine if it would be possible to predict which patients would have a PVR above this threshold, even though it is recognized that many patients with a PVR at or above 200 mL would not normally require CIC. This provided data from 173 patients with post-treatment PVR <200 mL and 38 patients with PVR ≥200 mL. Potential predictors evaluated in this analysis included pre-treatment peak and average urine flow (Qmax, Qavg) and Qmax and Qavg flow index (Qmax FI and Qavg FI) derived from the Liverpool flow formulas in isolation and in conjunction with patient age and baseline PVR. Descriptive analyses of baseline characteristics were performed. Receiver Operating Characteristic (ROC) curve analyses (Youden cutoff method) were conducted using logistic regression, with elevated PVR as the response.
Patients included in this analysis were predominantly female (73.7%) and Caucasian (86.8%). All patients in this cohort were naive to botulinum toxin treatment and had previously failed treatment for overactive bladder. The mean number of prior overactive bladder medications was 2.4, with a maximum of 7 medications.
When Qmax alone was evaluated as a predictor for post-treatment PVR ≥200 mL, there was a trend toward a potential relationship compared with patients with post-treatment PVR values of <200 mL; however this was not statistically significant (P=0.0855, Table 1). Logistic regression analysis of patients with post-treatment PVR values of ≥200 mL showed an odds ratio (95% confidence interval [CI]) of 1.04 (1.00–1.08) with a P-value of 0.0827. An ROC curve was generated that plotted the specificity and sensitivity of using Qmax to predict post-treatment PVR values ≥200 mL. The AUC was 0.5891, with a sensitivity value of 66% and 51% specificity, to correctly identify patients with post-treatment PVR ≥200 mL. Similar results were seen when evaluating Qavg (Table 1); the odds ratio (95% CI) was 1.07 (1.00–1.16) with a P-value of 0.0679, and the AUC of the ROC curve was 0.5916 (sensitivity, 68%; specificity, 50%).   
When Liverpool Qmax FI and Liverpool Qavg FI were evaluated as predictors of PVR ≥200 mL (Table 1), results were comparable. The logistic regression analysis of Qmax FI versus PVR ≥200 mL had an odds ratio (95% CI) of 0.30 (0.08–0.91) and was statistically significant (P-value 0.0488), suggesting that as pre-treatment Qmax FI increases, a patient’s chance of developing a PVR ≥200 mL drops. The AUC of the ROC curve (0.5880) was also similar to that seen for Qmax, with a sensitivity value of 95% and 23% specificity. Equivalent results were seen when assessing Liverpool Qavg FI (Table 1); the odds ratio (95% CI) was 0.07 (0.01–0.40) with a P-value of 0.0045, and the AUC of the ROC curve was 0.6603 (sensitivity, 71%; specificity, 61%).  
Further variables (age and baseline PVR) were then added to the analysis to determine whether the combination of variables would improve sensitivity and specificity. When Qmax and Qavg were evaluated with these variables, they appeared to be predictive for a post-treatment PVR ≥200 mL (Table 1). The Qmax ROC curve (with age and baseline PVR) had an AUC of 0.7378 (sensitivity, 61%; specificity, 77%) and for Qavg (with age and baseline PVR) the AUC was 0.7408 (sensitivity, 61%; specificity, 79%, Figure 1).
Similarly, when Liverpool Qmax FI and Liverpool Qavg FI were evaluated with age and baseline PVR, these variables together also appeared to be predictive for post-treatment PVR ≥200 mL (Table 1). The AUC of the Qmax FI ROC curve (with age and baseline PVR) was 0.7377 (sensitivity, 61%; specificity, 79%) and the AUC of the Qavg FI ROC curve (with age and baseline PVR) was 0.7362 (sensitivity, 74%; specificity, 68%, Figure 1).
Interpretation of results
While it should be reiterated that these analyses are preliminary, they do suggest that older patients, those with higher pre-treatment PVR values, and those with a lower pre-treatment urine flow (assessed by Qmax, Qavg or the Liverpool Qmax, Qavg FI) may be at higher risk of developing an elevated PVR following onabotA treatment.
Concluding message
The ability to predict which patients may be at higher risk of a higher post-treatment PVR and possibly requiring CIC following onabotA treatment would allow more tailored counselling and also more specific monitoring for those patients at risk. The data are supportive of a combination of age, pre-treatment PVR, and urine flow being predictive. The next step is to use the data to develop nomograms/formulae to help physicians treating patients with onabotA better quantify the risk for individual patients to assist with counselling and follow-up monitoring. Although it should be reiterated that the proportion of patients that develop a PVR over 200 mL is low and in this analysis, only 13 patients out of 211 developed a PVR >300mL. 
Further analyses in larger patient populations and use of a continuous statistical methodology will be needed to confirm and extend the findings from this exploratory analysis.
Figure 1 Table: Analysis of Qmax, Qavg, Liverpool Qmax FI, and Liverpool Qavg FI. FI = Flow index, IQR = interquartile range, PVR = post void residual, Qavg = average urine flow, Qmax = peak urine flow, SD = standard deviation, SE = standard error
Figure 2 ROC Analysis of Qavg (A) and Qavg FI (B) versus PVR ≥200 mL with the addition of age and baseline PVR volume. FI = flow index, PVR = post void residual, Qavg = average urine flow, Qmax = peak urine flow, ROC = Receiver Operating Characteristic
Funding Allergan plc Clinical Trial Yes Registration Number NCT03043287 RCT Yes Subjects Human Ethics Committee IEC Helsinki Yes Informed Consent Yes