Intravesical Lactobacillus rhamnosus GG versus Saline Bladder Wash: Protocol for a Randomized, Controlled, Comparative Effectiveness Clinical Trial

Abstract

Background

Complicated urinary tract infection (cUTI) is pervasive and costly among people with spinal cord injury (SCI) and neurogenic lower urinary tract dysfunction (NLUTD).

Objectives

To describe the protocol for a comparative effectiveness randomized controlled trial of intravesical Lactobacillus rhamnosus GG (LGG) versus saline bladder wash (BW) for self-management of urinary symptoms.

Methods

Comparative effectiveness trial of self-administered LGG versus saline bladder wash among 120 participants with SCI+NLUTD at least 6 months post SCI. The study has both treatment and prophylaxis phases. After predictive enrichment at screening, randomized participants will enter the treatment phase (6 months) in which they instill either LGG or normal saline after trigger symptoms occur (more cloudy or more foul-smelling urine). In the prophylaxis phase (6 months), participants will instill their respective intervention every 3 days after the first occurrence of trigger symptoms.

Results

Study results will provide a comparison of effects on Urinary Symptom Questionnaire for Neurogenic Bladder (USQNB) bladder and urine symptoms and episodes of “presumed UTIs”; number of days antibiotics were used (both self-reported); days of work, school, rehabilitation, or other activity lost due to urinary symptoms; engagement with the health care system; number of instillations; satisfaction; and safety.

Conclusion

cUTI is a variable clinical entity. Unlike clinical trials that assume a single, simple entity (UTI) in inclusion or outcome criteria, this protocol targets the mechanisms underlying cUTI causes and phenotypes. Featuring reliable and valid outcome measures with analytic methods specifically appropriate for quantifying self-report, patient self-management, inclusion of both intervention and prophylactic phases, and predictive enrichment, these design elements may be adopted for future research.

Introduction

Urinary tract infection (UTI) is the most common outpatient infection worldwide, affecting an estimated 150 million people annually.^1,2 For people with spinal cord injury (SCI) and neurogenic lower urinary tract dysfunction (NLUTD), it is the most common infection, secondary condition, cause for emergency room visits, and infectious cause of hospitalization.^3–9 In two large-scale analyses of SCI Model Systems participants (>8600 participants in Cardenas et al.⁸ and >950 rehospitalizations among participants in DeJong et al.⁴), genitourinary causes (UTIs being the most frequent) were reported as the most common cause of unplanned rehospitalization. Further, Skelton et al.⁷ found a 0.9% to 2.5% annual increase in rehospitalizations among the SCI population in the Healthcare Cost and Utilization Project from 2006 to 2015 due to genitourinary complications (the most common of which was “infection”). In 2021, investigators with the Neurogenic Bladder Research Group (NBRG) queried 1479 people with SCI about their experience with “presumed” UTIs during the preceding year: 46% reported 1 to 3 infections, 15% reported 4 to 6 infections, 13% reported >6 infections, and a minority (26%) reported no UTIs. Further, 16.7% reported at least one urologic hospitalization or emergency room visit during that year, with UTI being the most common reason cited.¹⁰ These analyses reflect a worryingly high prevalence of presumed UTIs since at least the mid-1990s.

Occurrence of UTI during inpatient rehabilitation has been associated with reduced functional recovery at discharge (inclusive of functional independence motor, admission, discharge, gain, and efficiency scores).¹¹ In addition, indirect costs of UTI include longer length of rehabilitation stay,¹¹ time lost from work, social isolation, and others.¹² Lastly, greater numbers of UTIs have been reported to be associated with worse quality of life and greater limitation of daily activities.¹³

It is well known that repeated exposures of bacteria to antibiotics leads to multidrug-resistant organisms (MDROs). It is very likely that the frequent use of antibiotics for presumed (but not proven) UTI and pre-infectious non-UTI symptoms is a driver of the increasing prevalence of infections with MDROs in the SCI population.^14,15 This evolution of MDROs among people with SCI represents a microcosm of the worldwide public health crisis of antibiotic resistance.¹⁶ The Centers for Disease Control and Prevention (CDC) estimates that 2.8 million people annually develop infections with resistant organisms and that 35,000 die as a direct result of these infections.¹⁶ Infections with resistant organisms are associated with longer hospital lengths of stay,¹⁷ increased mortality,¹⁸ and higher health care costs.¹⁹ With diseases of the genitourinary system being the leading reason for rehospitalizations among people with SCI,^6–8infections with resistant organisms represent a significant contributor to further elevated health care utilization and costs as well as mortality risk.^7,20 Additionally, the cost to treat a UTI due to an extended-spectrum beta-lactamase (ESBL) organism (which is a type of MDRO) is estimated to be 1.5 times more expensive than treating a non-ESBL pathogen. Thus, not only are people with SCI and UTI at greater risk of mortality due to MDROs but they are also likely to have more engagement with the health care system and higher costs in terms of health care and quality of life as a result of UTI.²¹ These costs can only be expected to increase, as the rate of antibiotic-resistant UTIs is increasing at a faster rate in people with NLUTD compared to those without.²²

These disturbing trends are long-standing, and our team has studied intravesical instillation of probiotics as an antibiotic-sparing therapeutic approach to the pervasive problem of UTI among people with NLUTD. Specifically, we have leveraged the modern recognition that healthy urine is not sterile²³ and have documented ways in which traditional UTI diagnosis can involve symptoms that are inaccessible in persons with SCI due to the neurologic impairment. Our recent work shows that bladder inflammation is present and persistent in most people with SCI due to the NLUTD.^24,25 The presence of uropathogens is also far more diverse than urine culture can reflect^26,27; moreover, there are both protective and uropathogenic bacteria in the urobiome that can impact the growth and/or activity of other urobiome bacteria.²⁸ Killing or targeting one specific member of the ecosystem with antibiotics likely inadvertently kills protective bacteria and/or promotes growth and or activity of other uropathogens. A more up-to-date, ecologically comprehensive approach is warranted, where protective bacteria and uropathogens are conceptualized within an ecosystem. Thus, we present the protocol of a comparative effectiveness randomized controlled trial of intravesical Lactobacillus rhamnosus GG (LGG) versus normal saline bladder wash (BW) in the self-management of urine quality and bladder symptoms. Vasudeva and Madersbacher²⁹ present a model of complicated UTI (cUTI) among people with NLUTD that describes the multifactorial causal contributors to cUTI in this population (inclusive of bacterial colonization, alteration of protective flora, direct inoculation by catheterization, immunodeficiency, bladder ischemia, defective glycosaminoglycan layer, defective apoptosis, deficient mechanical cleansing by voiding, etc.). Our active intervention (LGG) targets multiple contributing paths to cUTI (potentially impacting uropathogen access to urinary tract, ability to fight infection, and uropathogen adherence to the uroepithelium through competition or direct killing activity), whereas its comparator, normal saline bladder wash (BW), targets only one contributing path to cUTI (washout of uropathogenic bacteria).

Methods

SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) was used in the reporting of this clinical trial.³⁰ This trial is registered with ClinicalTrials.gov (NCT05230511).

The study design is a two-phase (treatment and prophylaxis), randomized, controlled, comparative effectiveness trial in which the efficacy of LGG+BW is compared to that of BW alone in response to the trigger symptoms of “more cloudy and/or more foul-smelling urine” (treatment phase). We follow this direct comparison of effectiveness with a second 6-month phase to compare the efficacies of LGG+BW and BW alone as a prophylactic intervention (prophylactic phase). In the treatment phase, participants will instill one or two doses over 24 hours when the trigger symptoms occur. In the prophylaxis phase, participants will initiate prophylaxis with one instillation every 3 days at the first occurrence of trigger symptoms. Figure 1 diagrammatically demonstrates the study design.

Figure 1.

Study summary. SMP = Self-Management Protocol.

Target enrollment is 120 subjects total at least 6 months post SCI. Inclusion criteria are (1) age ≥18 years, (2) SCI at least 6 months duration, (3) NLUTD (as determined by the participant’s SCI physician or urologist), (4) utilizing intermittent catheterization for bladder management, and (5) community dwelling (discharged from the acute care setting). Exclusion criteria are (1) ongoing genitourinary pathology beyond neuropathic bladder (i.e., the presence of vesicoureteral reflux, bladder or kidney stones, etc.), (2) use of prophylactic antibiotics, (3) instillation of intravesical agents (e.g., gentamicin, saline, or Lactobacillus), (4) immunodeficiency, (5) any oral antibiotics within the past 2 weeks, (6) psychologic or psychiatric conditions influencing the ability to follow instructions, (7) participation in another study in which results would be confounded, and (8) 2 weeks since prior exposure to intravesical LGG.

Predictive enrichment protocol

The first-in-human clinical trial demonstrated that roughly 30% of the population meeting the inclusion/exclusion criteria will not develop the intervention trigger symptoms (cloudier and/or more foul-smelling urine).³¹ To streamline the proposed study, participants will be screened using the inclusion and exclusion criteria, consented, and then will begin the final screening process: an enrichment protocol.

According to the US Food and Drug Administration (FDA), “Enrichment is the prospective use of any patient characteristic to select a study population in which detection of a drug effect (if one is in fact present) is more likely than it would be in an unselected population.”³² During the predictive enrichment period (up to 6 months), participants will self-monitor urinary symptoms using the USQNB-IC (Urinary Symptoms Questionnaire for Neurogenic Bladder-Intermittent Catheter version) weekly. As soon as self-reported trigger symptoms (cloudier and/or more foul-smelling urine) occur two distinct times within the 6-month observation period, they will be considered “eligible” for randomization, as this indicates a greater likelihood to experience triggers to engage with the intervention during the study phases. Each qualifying event (i.e., trigger symptoms experienced) must be followed by a 7-day period of full resolution of the trigger symptoms, so the minimum period of participation in the enrichment protocol before randomization is possible is 2 weeks. If the trigger symptoms are not experienced two times within the 6-month screening period, then the individual will be classified a “screen failure” and not randomized. Participants who do meet the enrichment criteria will be randomized to LGG+BW or BW and will enter the treatment phase.

Randomization in a 1:1 ratio will occur only after participants experience trigger symptoms twice during the predictive enrichment period.

Intervention training

Participants randomized to the LGG+BW intervention group will be trained on and follow the Self-Management Protocol with Probiotics (SMP-Pro), whereas those randomized to the BW group will follow the BW-specific Self-Management Protocol (SMP-BW). These materials and protocols differ only in terms of instillation components. Participants in each group will participate in a one-on-one training session immediately prior to starting the treatment phase and will receive their group’s respective manual (LGG or BW). Each group’s manual will include the instructions for both the treatment and prophylaxis phases.

Intervention and control materials

Instillation of intervention and control materials requires the following supplies: sterile gloves, new catheter, 0.9% normal saline, sterile cup, 60-cc syringe, and bed pad. Addition of the LGG capsules is the only difference between treatment arms. Participants will be provided with instillation kits, and supplies will be monitored.

Intervention

For the LGG instillation, participants will be instructed to mix the contents of one LGG capsule into 45 cc sterile 0.9% saline.³³ After mixing, participants will draw up the 45 cc liquid LGG mixture into a 60-cc syringe and instill via the intermittent catheter after the last catheterization prior to going to bed. Those randomized to BW only will follow the same instructions for drawing up 45 cc of normal saline in a 60-cc syringe and instilling via the intermittent catheter.

Self-management protocols

The SMP-Pro^33,34 will be used to direct participants how and when to initiate intravesical LGG instillations. A SMP-BW has been adapted from the SMP-Pro for the BW group. The trigger symptoms for initiation of instillation in both protocols are “cloudier urine” and/or “more foul-smelling urine.”

Measures

Symptoms are the primary outcome and will be measured using the USQNB-IC, which will be completed during both treatment and prophylaxis phases at the time of SMP trigger symptoms (treatment phase) and twice weekly (prophylaxis phase). Other outcomes include: episodes of “presumed UTIs”; number of days antibiotics used (both self-reported); days of work, school, rehabilitation, or other activity lost due to urinary symptoms; engagement with the health care system (calls, telehealth, visits, emergency); number of instillations; satisfaction (using a simple three-item survey at the end of the study participation); and safety data (adverse event/serious adverse event counts and attribution to the intervention, as determined by the independent data safety monitoring board).

Study variables

“Instillers” are defined as those who instill (in their respective study arm) at least once, and anyone who completes a phase without any instillation at all is characterized as a “non-instiller.” The dependent variables and co-primary outcomes for the two phases are symptom burden³⁵ based on the USQNB-determined bladder (B1 = changes in quality of pain when passing urine; change in quality of pain during catheterization; incontinence/urine leakage; urinary urgency; increase in catheterization frequency; decreased urine volume during catheterization; and blood in urine) and urine quality (B2 = worse-smelling urine; darker urine; and cloudier/sediment in urine) symptom burden estimates. Burden is computed for each individual as a function of the number of observed symptoms of a given type divided by the number of possible symptoms of that type.^31,35

To summarize B1 and B2 burden over all weeks of participation in each phase, we will count the number of reported symptoms in that type across the number of weekly USQNB-ICs that a given participant completed during that phase (max = 24 weeks). This overall experience of burden of the specific symptom type for the intervention phase is then divided by a denominator formed by the product of (number of possible symptoms in that type in any given week) and (number of weeks of participation).

Analyses

Intervention phase

Comparative effectiveness of LGG+BW versus BW will be based on analyses of bladder (B1) and urine (B2) symptom burden (described below) during the 6-month intervention period. Linear mixed-effects modeling will estimate the effect of the intervention on bladder (B1) or urine (B2) symptom burden as continuous dependent variables in two separate models with a random intercept (representing the individual’s overall symptomaticity). Candidate covariates (sex, age, time since injury, and days to initiate first intervention [LGG+BW or BW]) will be assessed in a two-step process. Each candidate will be tested univariately for balance across the two treatment groups, where “balanced” is defined as p value > .10. Candidates that are out of balance (i.e., p < .10) will then be univariately tested for association with each outcome (separately for B1 and B2 symptom burden), where “no association” is defined as p > .15. Variables that are both out of balance across treatment (p < .10) and associated with an outcome (p < .15) will be included in that outcome’s model. We will split alpha for these two inference tests (B1 burden model, B2 burden model). Sensitivity analyses will include running models for those who never instilled (“non-instillers”) and those who did instill at least one time (“instillers”) separately.

Prophylaxis phase

The prophylaxis phase requires that participants initiate their group’s intervention the first time they experience trigger symptoms after the previous phase ends. Instead of treating with one to two doses as in the previous phase, participants continue prophylaxis every 3 days until the end of the 6-month phase. Analyses will include only those who did initiate the prophylactic treatment. The same two-step covariate evaluation procedure and mixed-effects regression modeling will be repeated separately for B1 and B2 burden during this phase. These exploratory tests will not be corrected for overall multiple comparisons.

General exploratory

Not all those who are randomized will instill in either phase. Among instillers in each phase, we will conduct simple t tests, not corrected for multiple comparisons, to compare the average values of each of the following outcomes across arms within phase: number of days of work, school, rehabilitation, or other activity lost due to urinary symptoms; number of calls, video- or tele-visits with a health care provider; number of “presumed UTIs” treated with antibiotics; and number of days taking antibiotics. At the end of the study or at the dropout/final visit, participants who used their respective interventions at least once will receive a final survey asking them to rate their satisfaction with the intervention; likelihood of paying for the intervention if it was only available at some cost; likelihood of using/recommending their respective intervention in response to symptoms, prophylactically, or both; and, for dropouts (who used the intervention at least once), whether they dropped out because intervention was not working for them. Simple t tests will compare the average values of each survey item across arms with no corrections for multiple comparisons.

Power calculations

Only the intervention phase analyses are powered. We plan to use mixed-effects regression modeling for the most precision in our estimated effects of LGG+BW versus BW on bladder (B1) and urine quality (B2) symptom burden. For the most conservative power estimates, we use the most basic analytic method to estimate the worst-case scenario power calculation, a t test for each burden type. It is difficult to predict the exact form of the final models given the potential for any of our four potential covariates to qualify for inclusion and their effects on the model’s power (covariates will increase statistical power in linear regression). In this pair of primary outcomes (bladder/B1 burden, urine quality/B2 burden), alpha is set at 0.025 for each power calculation (and for each inference test). Our power considerations were informed by our maximum capacity of 120 participants over 5 years. In our prior study, we observed a 12% drop out over the baseline and intervention phases of the study, compounded by 30% of participants who never experienced the trigger symptoms during the intervention phase and so did not contribute informative data to the analysis of the intervention. We estimate that the enrichment protocol will decrease the likelihood that those randomized will fail to use the intervention in the arm to which they are randomized from 30% in our prior unenriched sample to roughly 10% (i.e., by two-thirds). Thus, with an anticipated 10% dropout rate, we anticipate a total loss of 20% dropout plus non-instillation during the 6-month period from randomization to the end of the intervention phase of the study.

With 1:1 randomization, enrolling 120 participants and assuming the 20% drop out/non-initiation of the intervention, will result in 100 participants who instill at least once and complete the intervention phase of the study, or n = 50 in each of two groups. With one-tailed alpha = 0.025 (per co-primary endpoint), we have 80% power to detect an effect size of 0.57, corresponding to a roughly “medium” effect³⁶ for each of the symptom burden estimates in a simple t test.

We plan unpowered post hoc analyses of potential interaction effects arising from significant models with covariates only. We will also plan to run post hoc sensitivity tests of the models to identify sex effects (e.g., different models for men and women) if warranted and potential impacts on symptoms for those who end up receiving antibiotics treatment and to test assumptions about linearity and covariate inclusion.

Discussion

There are several design features of this clinical trial that are relevant to SCI clinical research and care and potentially reproducible for other clinical trials. Firstly, we have argued that “UTI” is a poor outcome measure to be used in people with SCI and NLUTD due to the lack of a consistently applied and evidence-based definition^30–35 (which consists of varying combinations of bladder inflammation, symptoms, and bacterial load). As a result, diagnosis is often subjective and moreover, urinary tract infection (UTI) cannot be considered a single clinical entity. Specifically, UTI falls on a spectrum and can be uncomplicated, recurrent (2+ in 6 months or 3+ annually),^37–42 catheter-associated (CA-UTI),^37,38 and/or complicated (cUTI).^38,43,44 cUTI is defined as a urinary infection occurring in people with a structural or functional abnormality of the genitourinary tract or in people with underlying medical conditions or infection due to atypical organisms or occurring in conjunction with medical equipment (such as urinary catheters). cUTI carries a higher risk of treatment failure than either recurrent UTI or CA-UTI.⁴⁴ Because nearly all people with SCI have coexistent NLUTD,⁴⁵ often require the use of urinary catheters, and often experience UTIs with atypical organisms, the UTIs experienced by people with SCI should be considered “complicated” and may or may not also be catheter-associated. We propose more consistent use of “cUTI” to represent UTIs among people with SCI and NLUTD to clearly differentiate it from the UTI that occurs in people with normally functioning bladders.

Because of the inconsistencies associated with the UTI construct, we have developed and instead rely on “urinary symptoms” as a valid and reliable outcome measure. Whereas our clinical trial co-primary endpoints are based on a patient-reported outcome instrument, our summarization method of computing the burden of each symptom type focuses analyses on countable events rather than on subjective assessments of severity. The burden estimation yields interpretable interval level continuous variables for the regression modeling we propose in the protocol. Not only are our co-primary endpoints based on valid and reliable instruments, they are specific to the participants’ bladder management method,⁴⁶ represent symptoms with high prevalence among this patient cohort,⁴⁶ and are summarized in a reproducible and rigorous way that is amenable to parametric analytic methods.

Secondly, we implement predictive enrichment⁴³ prior to randomization into the first intervention phase, but we also utilize a modified version of this method again in the prophylaxis phase. The predictive enrichment protocol prior to randomization optimizes power for our intervention phase while simultaneously ensuring that all participants who are randomized have extensive practice with the content of the outcome instrument, the USQNB-IC. Our pilot work has suggested that participants may initially endorse symptoms more frequently than they do subsequently, leading to an appearance of high initial levels of symptom burden that plateau as the assessment becomes more familiar. Thus, this pre-randomization phase is an essential feature for both power and interpretability of the intervention phase. By implementing a similar approach in the prophylaxis phase, we ensure that that there is no “carry-over effect”⁴⁷(p8) from the intervention phase. For those who continue to experience the trigger symptoms, this study design offers the opportunity to study the comparative effectiveness of the interventions in both a symptomatic and a prophylactic administration.

Thirdly, in addition to comparative effectiveness under two conditions (symptomatic treatment and prophylactic), this design offers some of the benefits of a crossover trial⁴⁷ – including the opportunity to determine if there is a carry-over effect of either intervention – but recognizing that both interventions have the potential to change the environment of the bladder urobiome. Some data on the time course of such changes will also become available given the study design.

Finally, our analyses consider instillers and non-instillers separately. Even with the enrichment protocol, participants who experience, or do not experience, the trigger symptoms within a 6-month observation period cannot be compared statistically beyond descriptions of background characteristics.³¹ Non-instillers cause zero inflation (which the enrichment protocol seeks to limit) that can cause poor statistical model fit without adding useful information. However, the analyses that include non-instillers act as sensitivity analyses to enable fuller understanding of the heterogeneity of the effects of the intervention. Our sample enrichment protocol was designed to limit this particular source of heterogeneity, while sensitivity analyses ensure that we understand both the effects of our control (BW only) and also the effectiveness of each intervention on those who experience the trigger symptoms.

In sum, this protocol offers a modern, evidence-informed approach to treating the underlying causes of cUTI among people with SCI and NLUTD. The protocol design features enhance power and efficiency but also target outcomes that are objective rather than the highly subjective “UTI.” Because cUTI is a complex and variable clinical entity, this protocol targets the mechanisms underlying cUTI phenotypes. Taken together, the design, outcomes, and analyses are intended to ensure a more accurate understanding of the effects of this intervention while being adaptable for other studies in this and other patient populations.

Funding Statement

Financial Support The contents of this article were developed under a grant from the National Institute on Disability, Independent Living, and Rehabilitation Research (NIDILRR grant no. 90SIMS009-01). NIDILRR is a Center within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The contents of this article do not necessarily represent the policy of NIDILRR, ACL, or HHS, and you should not assume endorsement by the Federal Government.