MRI‐Derived Transition Zone Index Is Highly Predictive of Urodynamic Bladder Outlet Obstruction Prior to Holmium Laser Enucleation of the Prostate

ABSTRACT

Introduction and Objective

Urodynamic study (UDS) is required to diagnose bladder outlet obstruction (BOO) during evaluation of benign prostatic hyperplasia (BPH) but is seldom performed due to cost and invasiveness. Therefore, anatomic and clinical parameters to predict BOO have been proposed, including the prostate transition zone index (TZI) which is the ratio of prostate transition zone volume (TZV) to whole gland volume (WGV). Historically computed with ellipsoid volume estimation of prostate WGV and TZV from transrectal ultrasound measurements, controversy exists regarding the utility of TZI to predict likelihood of BOO on UDS and clinical outcomes following BPH surgery. Here, we aim to assess the association between MRI‐measured TZI and BOO on preoperative UDS in a modern BPH cohort before holmium laser enucleation of the prostate (HoLEP).

Methods

A prospectively maintained institutional database of 944 consecutive HoLEP patients between 2018 and 2022 was reviewed to identify those with preoperative UDS and MRI within 1 year of surgery. UDS was used to measure bladder outlet obstruction index (BOOI). 3D WGV and TZV were estimated using clinically available software (DynaCAD). We used linear regression to assess the relationship between TZI, WGV, and BOOI and logistic regression to determine the association between TZI, WGV, and BOO (defined as BOOI > 40).

Results

45/944 (4.8%) patients had both preoperative UDS and MRI within 1 year of HoLEP. Of these, 27 patients were obstructed (BOOI > 40) and 18 patients were not obstructed (BOOI ≤ 40) on preoperative UDS. Obstructed patients had larger prostate WGV, TZV, and TZI compared to non‐obstructed patients but were otherwise similar with respect to preoperative characteristics. Univariate analysis showed a positive association between TZI and BOO (R² = 0.373, p < 0.001) and WGV and BOO (R² = 0.214, p < 0.001). Multivariable logistic regression showed that TZI was independently associated with BOO (OR 1.08, 95% CI 1.02–1.14, p = 0.013) while accounting for WGV. WGV was not independently associated with BOO while accounting for TZI (OR 1.00, 95% CI 0.98–1.01, p = 0.614). The Youden index was utilized to create an optimal cutpoint for TZI (0.528) above which urodynamic BOO was very likely on multivariate logistic regression while accounting for WGV (OR 25.0, 95% CI 3.40–183.58, p = 0.002). The generated cutpoint for WGV (61.5 mL) was not significantly associated with urodynamic BOO on multivariate logistic regression while accounting for TZI (OR 0.993, 95% CI 0.98–1.01, p = 0.452).

Conclusions

Noninvasive MRI measurement of prostate TZI was highly and independently predictive of BOO before HoLEP and superior to WGV alone. This suggests that MRI obtained in evaluation of BPH/LUTS patients may be used to calculate TZI and inform patient selection for invasive urodynamic study and surgical treatment.

1. Introduction

Benign prostatic hyperplasia (BPH) is common in aging men, with histologic BPH present in an estimated 50% of men at 50 years [1]. Resulting prostate enlargement and fibrosis leads to bladder outlet obstruction (BOO) and causes moderate lower urinary tract symptoms (BPH/LUTS) in 40% of men by their seventh decade [2]. Men with bothersome BPH/LUTS are usually treated with medications followed by surgical resection of the prostate when medical treatment is ineffective. The most common surgical intervention for BPH is transurethral resection of the prostate (TURP) [3]. However newer methods, such as holmium laser enucleation of the prostate (HoLEP), offer a size independent surgical treatment in which the entire prostate transition zone is enucleated and is increasingly utilized as an alternative to TURP with improved initial clinical outcomes and lower long‐term re‐treatment rates [4].

The presence and severity of BOO is consistently cited as both an indication for BPH surgery and predictor of surgical success [5, 6]. Definitive diagnosis of BOO requires multichannel urodynamic study (UDS) and presence of BOO is determined via calculation of the bladder outlet obstruction index (BOOI) [7]. However, the expense, invasive nature, and limited availability of UDS in routine clinical practice results in a minority of patients undergoing UDS before BPH surgery [8]. Therefore, there is a need for noninvasive alternatives to UDS to diagnose BOO in BPH/LUTS patients.

The transition zone index (TZI), the ratio of prostate transition zone volume (TZV) to prostate whole gland volume (WGV), was first introduced in 1995 as an imaging‐based metric that correlated to worsening urodynamic metrics of peak urine flow (Qmax) and detrusor pressure at peak urine flow (PDetQmax) [9]. TZI can easily be calculated from common prostate imagining, including transrectal ultrasound (TRUS) and magnetic resonance imaging (MRI). Prior literature correlating TRUS measured TZI with UDS parameters and clinical outcomes in BPH/LUTS patients has yielded mixed results [9, 10]. This is perhaps due to known underestimation of prostate volume with TRUS compared to MRI and/or greater interobserver variability with ellipsoid method of volume estimation [11, 12]. To address these limitations, we aim to evaluate the relationship of MRI‐derived 3D estimation of TZI and urodynamic BOO in BPH/LUTS patients treated with HoLEP.

2. Methods

Following Institutional Review Board (IRB) approval, we performed a retrospective cohort study using our institutional database. We included consecutive adult patients (age > 18 years) who underwent HoLEP at our institution between June 2018 and January 2022 with UDS and MRI within 1 year before HoLEP. Patients with inaccessible, unavailable, or incomplete UDS data were excluded. Patients with MRI obtained more than 1 year before surgery were excluded to ensure accurate prostate volume measurement due to the expected annual prostate gland growth rate of 0.6 mL, with a range of −9.9 mL to 62.1mL [13]. In patients with multiple MRI studies the MRI most proximate to surgery was used. Operations were performed by five different surgeons and the decision for surgery was made using a shared decision‐making model in which urodynamic data was one factor. Comprehensive clinical data were collected from the electronic medical record including demographics, preoperative clinical characteristics, imaging data, and operative details. A fellowship trained abdominal radiologist experienced in prostate MRI (SAW) estimated WGV and TZV from the axial‐oblique Fast Spin Echo (FSE) T2‐weighted sequence (TR 4119 ms; TE = 92.4 ms; matrix =320 × 256; FOV = 260 × 234 mm²; slice thickness = 3 mm; true spatial resolution = 0.81 × 1.02 mm; flip angle = 90°; receiver bandwidth = 122.07 ± 31.25 kHz) obtained on a 3.0 T MRI (Discovery 750, GE Healthcare, Waukesha, WI) with a 32‐channel surface coil (NeoCoil, Pewaukee, WI) using a clinically available, semi‐automated 3D software (DynaCAD, Philips, Amsterdam, NL). Semi‐automated calculation of prostate TZV and WGV was performed to reduce intra‐observer volume estimation inconsistency. TZI was calculated as: TZI = TZV/WGV.

All urodynamic evaluations were performed with Medical Microsystems and conducted in concordance with International Continence Society guidelines. If a patient was able to void, the investigation began with noninvasive uroflowmetry. Following this, a seven French bladder and nine French rectal catheter were placed, zeroed at the level of the pubic symphysis, and cystometry and pressure flow study performed with room temperature saline instilled at 50 mL/min in a standing position. Sphincter electromyographic activity was recorded from cutaneous electrodes placed on the perineum. Typically, two or more pressure‐flow studies were performed. Data from the first study was used for analysis unless precluded by artifact, in which case data from the first subsequent study free of artifact was used. One author interpreted or verified all the urodynamic tests and reported the parameters including Qmax and PdetQmax. BOOI was calculated from the formula BOOI = PdetQmax – 2Qmax. BOOI was used to classify patients as non‐obstructed (BOOI ≤ 40) or obstructed (BOOI > 40).

Relevant clinical and demographic data between groups were compared utilizing GraphPad Prism version 10.2.0, which involved two‐tailed Mann‐Whitney U tests for continuous variables or Fisher's exact tests for categorical variables. Differences were considered significant if p ≤ 0.05. Univariate linear regression, univariate logistic regression, multivariate logistic regression, and empirical cutpoint estimation were performed with STATA 18.0 software (StataCorp, College Station, TX, USA).

3. Results

3.1. Clinical and UDS Characteristics

We identified 944 patients who underwent HoLEP during the study period. Of these, 45 (4.8%) underwent both preoperative UDS and MRI within 1 year of HoLEP and were included in the study. Overall mean patient age was 69.2 ± 8.7 years, preoperative prostate volume was 84.0 ± 50.1 cc, preoperative PSA was 5.1 ± 6.3 mg/mL, preoperative International Prostate Symptom Score (IPSS) total was 20.4 ± 6.3, and postoperative IPSS total was 6.9 ± 5.0. Of 45 patients, 18 were non‐obstructed and 27 were obstructed. Table 1 compares group clinical characteristics. Non‐obstructed and obstructed groups were similar with respect to patient age, BMI, preoperative catheter status, postoperative catheter status, race/ethnicity, and IPSS scores. Obstructed patients had a higher mean TZI (0.69 ± 0.14 vs. 0.53 ± 0.17), WGV (94.6 ± 48.7 vs. 68.2 ± 49.4 mL), TZV (68.9 ± 40.2 vs. 42.6 ± 46.9 mL), and PSA (6.6 ± 7.5 vs. 2.7 ± 2.5 ng/mL), and a shorter time interval between MRI and surgery (54.7 ± 54.4 vs. 89.3 ± 82.8 days) compared to non‐obstructed patients (p < 0.05 for all). Table 2 compares group UDS characteristics of non‐obstructed and obstructed patients. First sensation, first desire, strong desire, capacity, and postvoid residual (PVR) were similar between groups. Obstructed patients had lower threshold to normal desire (248.8 ± 88.0 mL vs. 378.2 ± 141.3), lower Qmax (4.44 ± 2.22 mL vs. 7.94 ± 5.65), higher PDetQmax (86.2 ± 22.4 vs. 39.9 ± 14.3), and higher BOOI (77.3 ± 23.2 vs. 24.0 ± 13.3) compared to non‐obstructed patients (p < 0.05 for all).

TABLE 1.

Demographic and clinical characteristics of the study population.

	Overall N = 45	Non‐Obstructed N = 18	Obstructed N = 27	p‐value
Race, n (%)
White	39 (86.7)	14 (77.8)	25 (92.6)	0.076
Black or African American	2 (4.4)	1 (5.6)	1 (3.7)
Asian	2 (4.4)	2 (11.1)	0 (0)
Not Specified	2 (4.4)	1 (5.6)	1 (3.7)
Ethnicity, n (%)
Hispanic	0 (0)	0 (0)	0 (0)	0.999
Non‐Hispanic	43 (95.6)	17 (94.4)	26 (96.3)
Not Specified	2 (4.4)	1 (5.6)	1 (3.7)
Clinical Characteristics, mean ± SD
Age at Imaging (years)	69.2 ± 8.7	67.4 ± 9.5	70.4 ± 8.0	0.364
BMI	28.0 ± 4.6	28.9 ± 3.7	27.5 ± 5.1	0.076
Whole Gland Volume (mL)	84.0 ± 50.1	68.2 ± 49.4	94.6 ± 48.7	0.039
Transition Zone Volume (mL)	58.4 ± 44.4	42.6 ± 46.9	68.9 ± 40.2	0.012
Transition Zone Index (TZI)	0.62 ± 0.17	0.53 ± 0.17	0.69 ± 0.14	0.001
Time between Imaging and Surgery (days)	68.5 ± 68.5	89.3 ± 82.8	54.7 ± 54.4	0.036
Preoperative IPSS Total	20.4 ± 6.3	20.2 ± 6.0	20.6 ± 6.9	0.820
Postoperative IPSS Total	6.9 ± 5.0	8.2 ± 5.1	6.7 ± 5.1	0.601
Preoperative PSA (ng/mL)	5.1 ± 6.3	2.7 ± 2.5	6.6 ± 7.5	0.023
Voiding Status Preop, n (%)
Voiding	30 (66.7)	12 (66.7)	18 (66.7)	0.999
Catheter Dependent	15 (33.3)	6 (33.3)	9 (33.3)
Voiding Status Postop, n (%)
Voiding	41 (91.1)	17 (94.4)	24 (88.9)	0.640
Catheter Dependent	4 (8.9)	1 (5.6)	3 (11.1)

Abbreviations: BMI, body mass index; IPSS, international prostatic symptom score; PSA, prostate‐specific antigen; SD, standard deviation.

TABLE 2.

Urodynamics parameters of the study population.

	Overall (n = 45)	Non‐Obstructed (n = 18)	Obstructed (N = 27)	p‐value
Urodynamic Parameters, mean ± SD
First Sensation (mL)	136.0 ± 90.5	156.5 ± 107.0	121.5 ± 75.8	0.376
First Desire (mL)	205.8 ± 135.6	236.8 ± 177.8	185.2 ± 97.3	0.707
Normal Desire (mL)	295.4 ± 124.6	378.2 ± 141.3	248.8 ± 88.0	0.027
Strong Desire (mL)	342.9 ± 230.1	401.0 ± 314.4	304.1 ± 146.2	0.657
Capacity (mL)	372.6 ± 230.6	446.4 ± 296.4	324.3 ± 164.2	0.241
PVR (mL)	212.7 ± 188.1	278.1 ± 257.4	180.0 ± 137.0	0.557
Qmax (mL/sec)	5.84 ± 4.27	7.94 ± 5.65	4.44 ± 2.22	0.014
PDetQmax (cmH2O)	67.7 ± 30.0	39.9 ± 14.3	86.2 ± 22.4	0.0001
BOOI	56.0 ± 32.9	24.0 ± 13.3	77.3 ± 23.2	0.0001

Abbreviations: PDetQmax, detrusor pressure at maximum flow; PVR, post‐void residual; Qmax, maximum flow; SD, standard deviation.

3.2. Association of Prostate TZI and WGV With Urodynamic Outlet Obstruction

Univariate linear regression showed a positive association between TZI and BOOI (R² = 0.373, p < 0.001) and WGV and BOOI (R² = 0.214, p < 0.001). Logistic regression showed that increasing TZI was associated with urodynamic outlet obstruction (BOOI > 40) [OR 1.07, 95% CI 1.02–1.11, p = 0.005] but WGV was not [OR 1.01, 95% CI 1.00–1.03, p = 0.092] (Table 3). Receiver operating characteristic (ROC) curves showed an area under the curve (AUC) of 0.750 and 0.663 for TZI and WGV, respectively (Figure 1). The Youden index was used to create the optimal cutpoints to maximize the sensitivity and specificity of the ROC curve for predicting urodynamic outlet obstruction. These optimal cutpoints were 0.528 and 61.5 mL for TZI and WGV, respectively.

TABLE 3.

Univariable and multivariable logistic regression assessing the association of prostate whole gland volume and transition zone index with urodynamic bladder outlet obstruction: treated as continuous variables.

	Univariable		Multivariable
Variable	OR (95% CI)	p‐value	OR (95% CI)	p‐value
TZI	1.07 (1.02–1.11)	0.005	1.08 (1.02–1.14)	0.013
WGV	1.01 (1.00–1.03)	0.092	1.00 (0.98–1.01)	0.614

Abbreviations: CI, confidence interval; TZI, prostate transition zone index; WGV, prostate whole gland volume.

FIGURE 1.

Receiver operating characteristic (ROC) curves for diagnosis of bladder outlet obstruction based on transition zone index and whole gland volume. MRI measured prostate transition zone index (TZI) is more highly predictive of bladder outlet obstruction compared to prostate whole gland volume (WGV). Area under TZI curve is 0.750, area under WGV curve is 0.663.

Using these cutpoints, logistic regression showed a positive relationship between TZI > 0.528 and urodynamic obstruction [OR 16.0, 95% CI 3.40–75.34, p = 0.001] as well as for WGV > 61.5 mL and urodynamic obstruction [OR 7.43, 95% CI 1.94–28.47, p = 0.003] (Table 4). Multivariable logistic regression showed that increasing TZI was independently associated with urodynamic obstruction when considered as either a continuous [OR 1.08, 95% CI 1.02–1.14, p = 0.013] or categorical variable [OR 25.0, 95% CI 3.40–183.58, p = 0.002] (Tables 3 and 4). Prostate WGV was not associated with BOOI on multivariable analysis (Tables 3 and 4).

TABLE 4.

Univariable and multivariable logistic regression assessing the association of prostate whole gland volume and transition zone index with urodynamic bladder outlet obstruction: treated as binary variables.

	Univariable		Multivariable
Variable	OR (95% CI)	p‐value	OR (95% CI)	p‐value
TZI > 0.528	16.0 (3.40–75.34)	0.001	25.0 (3.40–183.58)	0.002
WGV > 61.5 mL	7.43 (1.94–28.47)	0.003	0.993 (0.98–1.01)	0.452

Abbreviations: CI, confidence interval; TZI, prostate transition zone index; WGV, prostate whole gland volume.

4. Discussion & Conclusion

Prostate WGV is the most often assessed anatomic parameter during BPH evaluation and management. However, prostate gland zones do not contribute to BPH equally. The transition zone experiences proliferative growth during BPH and accounts for most of the age‐related increase in prostate volume [14, 15]. We used noninvasive MRI and a common, clinically available software to estimate the prostate TZI and evaluate the association with BOO on urodynamic study before HoLEP. We found that increasing TZI was highly and independently associated with BOO on multivariate analysis. In addition, we identified a TZI cutpoint of 0.528 above which patients are very likely to have BOOI ≥ 40. As prostate MRI utilization increases generally and specifically in evaluation of elevated PSA, TZI is a useful, noninvasive, and available adjunct to predict likelihood of BOO and guide need for further evaluation. For example, a patient with severe BPH/LUTS with TZI < 0.528 could be considered for invasive UDS to evaluate for BOO and assess alternative causes of symptoms such as detrusor underactivity before pursuing BPH surgery.

Prior studies have assessed associations between prostate WGV, TZV, and TZI with urodynamic findings and clinical outcomes with mixed results [9, 10, 16, 17, 18, 19]. One such study correlated transrectal ultrasound (TRUS) measured prostate WGV and TZI with AUA symptom score, Qmax, and PDetQmax in 61 men with symptomatic BPH [9]. The authors found that TZI was positively associated with increasing AUA symptom score (r = 0.75, p < 0.001) and PDetQmax (r = 0.43, p < 0.001) and negatively associated with Qmax (r = 0.71, p < 0.001). Further, the authors identified a TZI cutpoint of 0.5 above which there was a significant difference in symptom score, Qmax, and PdetQmax [9]. Another study of 350 Japanese BPH patients correlated TRUS measured TZI with urodynamic obstruction classified as BOOI > 40, as in our study [20]. They found that mean TZI was higher in obstructed versus unobstructed men (0.59 vs. 0.49, p < 0.0001) with an AUC of 0.719; however, the TZI was not associated with BOOI on multiple logistic regression [20]. Other studies have not shown a consistent relationship between TZI, symptom scores, and UDS metrics. One study of 93 men with TRUS derived WGV and TZV, found no relationship between TZI and symptom score (r² = 0.022, p = 0.165) and reported that a TZI cutoff of 0.5 did not identify men with more severe symptoms [16]. This study also demonstrated weak correlations between TZI and Qmax (r² = 0.147, p < 0.001), WGV (r² = 0.160, p < 0.001) and TZV (r² = 0.156, p < 0.001). Another study conducted with 35 men in India with TRUS‐based WGV and TZV showed that TZI was not predictive of bladder outlet obstruction (OR = 1.99, p = 0.24) [19].

The controversy in literature is likely explained by differing methods of prostate volume measurement, comparison of different urodynamic and clinical outcome metrics, and/or differences in patient cohorts. For example, most prior studies estimated prostate volume using TRUS linear measurements and ellipsoid volume formula. While TRUS is accessible and inexpensive, it has been shown to underestimate prostate volumes compared to MRI [11]. Additionally, the ellipsoid method has demonstrated greater interobserver variability compared to the 3D volumetric method in calculation of prostate volumes, such as was used in our study [12]. Variability in prostate volume calculations may impact the strength of relationship with BOO and other clinical parameters. Therefore, we propose that volume estimation using a readily available 3D software that is familiar to most radiologist who perform/interpret prostate MRI may be a more accurate method to measure prostate WGV and TZV in circumstances where these metrics may change clinical management. Furthermore, differences in patient populations and average prostate volumes may have contributed to differences in study results. Average WGV in earlier highlighted studies fell between 47 and 52 mL compared to our average of 84.0 mL [9, 16, 19, 20]. Whether or not our results are generalizable to all patients, and in particular those with smaller prostates, is worthy of further study.

Perhaps due to the conflicting reports on the clinical utility of TZI, recent research focuses on the development of other imaging biomarkers that strongly correlate to BOO. This includes but is not limited to intravesical prostatic protrusion, high bladder neck elevation, peripheral zone index, bladder wall thickness, detrusor wall thickness, and resistive index [21, 22, 23]. With an OR of 25.0 (3.40–183.58 95% CI) most bothersome BPH/LUTS patients will have UDS demonstrated BOO with a TZI > 0.528, yet patients may present with an elevated TZI that are unobstructed on UDS. For these patients, a combination of imaging‐based metrics may help to better characterize urologic symptoms and counsel patients in the future than utilizing TZI alone, reducing the discrepancy between noninvasive imaging determined BOO and gold‐standard UDS determined BOO. Further, as clinical BPH likely represents a confluence of factors in any individual patient, combining imaging‐based metrics may be more powerful than any single metric in evaluation of BOO. As an example, one study found that the combination of TRUS measured intravesical prostatic protrusion and resistive index was more predictive of BOO than the individual clinical metric [20]. Additional studies are needed to evaluate the potential value of this approach and we propose that MRI derived metrics will be increasingly relevant as prostate MRI becomes increasingly integrated into the care pathway of men with BPH/LUTS.

Our study has limitations that should be considered. First, our study is subject to the selection bias inherent in any retrospective surgical series. Second, our study population is majority white and our findings may not be generalizable to other racial and ethnic groups. Third, interobserver variability in reporting of urodynamic data may have impacted the observed relationships between TZI, WGV, and BOOI; however, we attempted to address this by having a single physician verify the accuracy of all UDS interpretations. Fourth, our study used MRI for estimations of prostate WGV or TZV. While MRI may be more accurate than CT and US [11], TRUS is still a common method of prostate imaging before surgery in the United States. MRI and 3D volume estimations are highly accurate, but this technology and workflow may not be available at all clinical sites. Yet, as the use of MRI becomes more widespread in the adjudication of elevated PSA, 3D prostate volume estimation will be increasingly available in clinical practice as a less invasive and perhaps more reproducible means of accurately estimating prostate volume [24, 25]. Fifth, our study utilized a semi‐automated 3D software to estimate prostate TZV and WGV. This software may not be available at all clinical sites limiting the generalizability of our findings, however, TZV and WGV may be estimated from MRI without the use of this software but possibly subject to greater intra‐observer variability. Finally, our study examined patients with larger average prostate WGV than other studies and may not be generalizable to men with smaller prostates.

Despite these limitations, our data holds promising implications for expanded usage of MRI in preoperative evaluation of patients before BPH surgery. We demonstrated that noninvasive MRI based estimation of TZI was highly and independently predictive of BOO before HoLEP and was superior to WGV alone. In the future, MRI‐based measurement of TZI singularly or in tandem with other imaging metrics may act to inform patient selection for invasive urodynamic study and diagnosis of BOO. This may assist in patient selection, counseling, and treatment satisfaction. Further investigation is needed into the relationship between MRI‐measured TZI and other imaging metrics and clinical outcomes following BPH surgery in a larger, more diverse, cohort.

Ethics Statement

This study was reviewed and approved by the University of Wisconsin‐Madison's Institutional Review Board, under IRB ID 2021‐0284 as a retrospective study without patient interaction. No clinical trial registration was necessary for these reasons. No ethical objections were raised.

Data Availability Statement

Research data are not shared, PHI information was utilized. Data may be available after appropriate approval from member institutions, adherence to HIPAA guidance, and IRB approval.