Introduction

Lower urinary tract symptoms (LUTS) suggestive of benign prostatic hyperplasia (BPH) are common in aging men. For decades, the symptoms of this condition have been treated with phytotherapy. Various preparations are available, mainly consisting of extracts from saw palmetto fruit, pumpkin seed, pygeum africanum bark, nettle root, or willow herb [1, 2]. Meanwhile, selective alpha1-receptor blockers (ARBs) and 5-alpha-reductase inhibitors (5-ARIs) have increasingly served as first-line medical treatments since they came on the market in the 1990s. However, synthetic drugs have potential side effects that might negatively influence quality of life, such that many patients prefer herbal products due to their excellent safety profile. Thus, plant extracts have been continuously used in clinical practice [3,4,5].

In recent years, symptom bother and quality of life have become key criteria for therapy decisions [6, 7]. Considering patient preferences has led to debates about the role of phytotherapy particularly in patients with a low risk of disease progression [4, 8, 9].

Since 2014, the German S2e guideline suggests the option of treating LUTS/BPH with herbal preparations of proven superiority to placebo in patients who have mild-to-moderate complaints and refuse chemical compounds [6].

Notably, relevant placebo responses have regularly been observed in randomized controlled trials in patients with LUTS/BPH [10]. The placebo effect is rapid and may account for 40-60% of the overall symptom relief but tends to diminish over time [10,11,12]. Consequently, the International Consultation on BPH emphasized the importance of placebo control and follow-up periods of at least 12 months for clinical research in the early 1990s [13].

Nonetheless, most studies with herbal preparations have shortcomings, such as brief follow-up periods up to only 6 months, lack of placebo control, and small samples with less than 100 participants per treatment group [6, 14, 15]. To date, no more than five randomized placebo-controlled long-term studies in patients with LUTS/BPH have been reported [6]. Notably, two of these 12-month studies investigated pumpkin seed soft extract (PSE)Footnote 1 [16, 17].

The first study, published by Bach [16], showed significant IPSS improvement with PSE versus placebo, and its 12-month follow-up period was recognized as an exception in clinical research on phytotherapy for LUTS/BPH [6].

The second study (GRANU study) was a three-arm trial testing the efficacy and safety of pumpkin seed (open study arm) and PSE capsules against matching placebo. This study showed no difference between PSE and placebo [17].

Nonetheless, these two placebo-controlled studies have established the long-term safety of the extract; less than 1% of more than 700 patients treated with PSE reported adverse events with a possible causal relationship to treatment [16, 17]. The reactions were nonserious, and most of them were gastrointestinal complaints. The types and frequency of all recorded adverse events in the PSE group were similar to those in the placebo group. Serum PSA levels and other laboratory safety parameters showed no relevant changes from baseline after 12 months of treatment [16, 17]. An observational study confirmed the excellent safety profile of PSE in a large population of 2245 men with LUTS/BPH. After 3 months of treatment, the patients reported symptomatic relief and improved quality of life. Simultaneously, only mild gastrointestinal complaints were reported in no more than 4% of the patients [18].

In terms of efficacy, we performed a meta-analysis to estimate the benefits of PSE compared to placebo using the data from the original reports of the Bach and GRANU study. Since we used the individual patient listings, the meta-analysis also involved the reanalysis of the initial studies enabling us to provide supplementary information about the individual studies. In addition, we aimed to explore the possible influences of covariates on treatment response using statistical models.

Materials and methods

Description of the studies included in the meta-analysis

Design, conduct, and target populations

The Bach and the GRANU study followed the clinical research criteria established by the WHO-sponsored International Consultation on BPH [13]; both studies were randomized and placebo-controlled, had a treatment period of 12 months, and used the IPSS as the primary efficacy variable; eligible patients had to have an IPSS of at least 13 points (Table 1, Supplementary Table A).

Table 1 Design, treatment and follow-up schedule of pooled studies
Full size table

Intervention

The active treatment in both studies consisted of pumpkin seed soft extract (PSE) given in capsules (500 mg BID). PSE is a proprietary extract manufactured from the seeds of Uromedic® pumpkin, a company-owned registered cultivar of Cucurbita pepo L. convar. Citrullinina GREB. var. styriaca GREB (extraction solvent ethanol 92% [w/w]; drug-extract ratio: 15-25:1). The extract has a high content of Δ7-phytosterols (Δ7,25-stigmasterol, spinasterol, Δ7-avenasterol) that are specific putative active constituents of pumpkin seeds [19]. One capsule with 500 mg PSE contains approximately 15 mg of Δ7-phytosterols [20]. Placebo capsules contained the active capsules’ excipients and macrogol 400 as an inert substitute for the extract.

PSE and placebo capsules were indistinguishable by taste, smell, size, shape, and color. Independent statisticians who were not involved in the study analyses generated permuted, balanced block randomization lists stratified by study site. According to the randomization schedule, the study medications were packed in identical, neutral boxes, each with a patient number. After the 1-month run-in period, eligible patients were randomly assigned to 12-month treatment with either PSE (500 mg BID) or placebo (Table 1).

Only complete random blocks were allocated to each study center. The block size was not stated in the study protocols. The investigator assigned patients to a treatment using the study medication with the next consecutive random number. In the partially blinded GRANU study, opaque, sealed, serially numbered envelopes were used as previously described [17].

Until the studies were unblinded, staff involved in the study conduct and analyses were unaware of the treatment allocation. Randomization lists were only opened after completing data cleaning and patient assignments to the analysis sets.

Efficacy variables

The International Prostate Symptom Score (IPSS) was the primary efficacy variable. Patients were classified as responders if they had a decrease in IPSS of at least 5 points from baseline after 12 months of treatment. In addition, the mean changes from baseline in IPSS and related quality of life (IPSS QoL) were analyzed for all visits.

The IPSS is a self-administered 7-item patient questionnaire covering the symptoms incomplete emptying, frequency, intermittency, urgency, weak urinary stream, straining, and nocturia. An additional question assesses the IPSS-related quality of life (IPSS QoL). On a 6-point scale, patients rate the frequency of each symptom from 0 (‘never’) to 5 (‘almost always’). Total IPSS ranges between 0 and 35 points. Although alone not suitable to diagnose the presence of BPH, the IPSS is a clinically reliable and worldwide accepted tool to assess and follow-up the severity of LUTS/BPH [6, 13]. According to the IPSS total score, symptom severity is categorized as mild (IPSS< 8), moderate (IPSS = 8-19), or severe (IPSS = 20-35) [7, 21]. An IPSS reduction by a minimum of 3 points is considered a perceivable change for the patient [21,22,23,24]. Thus, a 5-point IPSS decrease represents a clinically relevant therapeutic response [24].

The answers to the single IPSS QoL question range between 0 (delighted) and 6 (terrible) [13, 21].

Secondary parameters included maximum urinary flow rate (Qmax), postvoid residual volume (PVR), and prostate volume (PV) measured at baseline and after 12 months of treatment.

Statistical methods

The methods of data handling and analysis were prespecified in a statistical analysis plan. All statistical evaluations were performed using the software package SAS release 9.4 (SAS Institute Inc.).

Data extraction and handling

The individual patient data owned by the present manufacturer of PSE were used to perform the meta-analysis. For the Bach study, an electronic database was not available. Therefore, data from paper-based lists were captured in electronic form. When entering the data, the IPSS QoL score values ranging from 1 to 7 were reduced by one unit to adjust them to the more common range from 0 to 6 used in the GRANU study.

For the GRANU study, PSE and placebo patient data were extracted from the existing electronic database.

Analysis sets

The intention-to-treat (ITT) analysis set consisted of all randomized patients who took at least one study medication and had baseline and at least one postbaseline IPSS assessment. For the per-protocol (PP) analysis set, the original assignments of each study were retained (Fig. 1).

Fig. 1
figure 1

Patient disposition. Number of patients: N, pooled; n1, Bach study; n2, GRANU study. PSE pumpkin seed soft extract

Full size image

The meta-analysis was performed for the ITT set using last-observation-carried-forward (LOCF). Each analysis was repeated for the PP set with observed data.

Efficacy analysis

The primary efficacy time point was the last follow-up visit after 12 months. According to the original studies’ definition, the response criterion was a 5-point IPSS reduction after 12 months of treatment. Response rate differences between the PSE and placebo groups were statistically tested by Fisher’s exact test. The 95% confidence intervals (CIs) were calculated using Wilson’s score method.

Statistical significance refers to a two-sided type I error of 5%. No adjustments for multiplicity were made.

To compare the changes from baseline in IPSS and secondary parameters between treatment groups, P values were calculated using the t test, assuming equal variances. Confidence intervals (CIs) were calculated using t distribution.

Furthermore, response rates and mean changes in IPSS from baseline to 12 months were investigated using logistic regression and analysis of covariance (ANCOVA) models with treatment, study, baseline IPSS, and center size as independent variables. Study sites were classified according to their number of recruited patients into low (1-3 patients), medium (4-10 patients), and high (> 10 patients) center sizes. P values were calculated using the Wald test.

Ethical aspects

Approvals from responsible ethics committees and authorities were obtained for the original studies. All participants provided written informed consent before any screening examination. Thus, further institutional approval was not required for the present analysis.

Results

Patient disposition

The pooled analysis set comprised 1432 patients (PSE, 714; placebo, 718). Forty-three patients (PSE, 27; placebo, 16) were excluded from analysis, mostly due to missing postbaseline IPSS assessment. The ITT and PP sets consisted of 1389 and 973 patients, respectively (Fig. 1).

Baseline characteristics

At baseline, treatment groups were homogeneous for demographic and clinical characteristics (Table 2 [ITT], Supplementary Table B [PP]). On average, the ITT patients were approximately 64.4 years old and had IPSS and IPSS QoL scores of 16.6 and 3.4 points, respectively The means of Qmax, PVR, and PV were 10.1 mL/s, 38.6 mL, and 30.2 mL, respectively (Table 2).

Table 2 Baseline demographic and disease characteristics (ITT set)
Full size table

Approximately 11%, 29%, and 60% of the ITT patients (N = 1389) were treated in low-, medium-, and high-recruiting centers, respectively (Table 3).

Table 3 Patients in low-, medium- and high-recruiting centers (ITT set)
Full size table

Primary efficacy

The IPSS continuously declined in both treatment groups (Fig. 2, Supplementary Table C). After the 12-month treatment period, the response rates in the PSE group exceeded those in the placebo group by 3% (ITT) and 5% (PP) (Table 4). Figure 3 shows the forest plot for the primary analysis set (ITT set using LOCF). The mean IPSS changes from baseline after 12 months were − 5.3 points in the PSE group and − 4.8 points in the placebo group (ITT using LOCF, Supplementary Table C). The mean change difference between the treatment groups was 0.6 points in favor of PSE (Fig. 4).

Fig. 2
figure 2

Mean IPSS changes from baseline during 12-month treatment (ITT set). Vertical bars indicate standard error. PSE pumpkin seed soft extract; BL baseline

Full size image
Table 4 Responders after 12 months of treatment and differences between treatments
Full size table
Fig. 3
figure 3

Forest plot showing the difference (with 95% CI) between treatments (PSE minus placebo) in IPSS response rates after 12 months of treatment; ITT set using LOCF. PSE pumpkin seed soft extract. The response rate is the number of responders divided by the total number of patients treated in the respective group

Full size image
Fig. 4
figure 4

Forest plot showing the mean difference (with 95% CI) between the treatments (placebo minus PSE) in IPSS change from baseline after 12 months of treatment; ITT set using LOCF. PSE pumpkin seed soft extract

Full size image

Logistic regression results

The logistic regression analysis of response rates in the ITT set resulted in an odds ratio (OR) of 1.17 for comparing PSE and placebo treatment and of 1.28 for comparing the Bach and GRANU studies (Table 5). The baseline IPSS had a relevant influence on treatment response (OR, 1.17 [per one-point increase in the IPSS baseline score]). Additionally, the variable center size affected a patient’s chance of achieving the response criterion. The effect was conspicuous (OR, 0.6) when comparing low- and high-recruiting centers (Table 5).

Table 5 Odds ratios (OR) for treatment response in IPSS estimated by logistic regression (ITT set using LOCF)
Full size table

ANCOVA results

After 12 months, the mean IPSS change from baseline estimated by ANCOVA showed a difference between treatments of 0.7 points (95% CI 0.1, 1.2) in favor of PSE. Again, the influence of the variables study (Bach vs. GRANU), baseline IPSS, and center size (low- vs. high-recruiting centers) on the magnitude of IPSS change was present (Table 6).

Table 6 Mean IPSS changes from baseline to 12 months estimated by ANCOVA (ITT set using LOCF)
Full size table

Secondary variables

In the ITT set, the mean IPSS QoL scores were reduced by 0.9 and 0.8 points during the first 3 months of treatment in the PSE and placebo groups, respectively. After 12 months, the mean IPSS QoL scores were 1.2 points (PSE) and 1.1 points (placebo) lower than the scores at baseline (Fig. 5, Supplementary Table D).

Fig. 5
figure 5

IPSS QoL score during 12-month treatment (ITT set). Bars indicate 95% confidence intervals. BL baseline; PSE pumpkin seed soft extract

Full size image

Table 7 shows the changes in Qmax, PVR, and PV from baseline to the 12-month follow-up.

Table 7 Statistical characteristics of changes from baseline to 12 months for urologic parameters (ITT set)
Full size table

Specific characteristics of the individual studies

Baseline

In both studies, all baseline characteristics were well balanced between PSE and placebo groups in the ITT (Table 2) and PP sets (Supplementary Table B). However, when comparing the data between the two studies, conspicuous differences were found for some demographic and clinical parameters. On average, the patients in the GRANU study were approximately 3 years older than the Bach study patients. The mean baseline IPSS was 16.0 points in the GRANU study compared to 17.6 points in the Bach study (ITT sets, Table 2).

The upper PV limit of 40 mL for inclusion in the GRANU study resulted in a lower mean PV of 29 mL compared to 35 mL in the Bach study. The mean Qmax was 9.6 mL in the GRANU study and 11.0 mL in the Bach study (Table 2).

IPSS outcome

In the Bach study, the response rate of the PSE group (65%) was significantly higher than that of the placebo group, with a difference of 11% (95% CI 1%, 20%) (Table 4). For the mean IPSS change from baseline, the difference between PSE and placebo after 12 months was − 1.1 points (95% CI -2.0, − 0.1) in favor of PSE (Supplementary Table C). Conversely, in the GRANU study, response rates and changes from baseline after 12 months of treatment did not differ between PSE and placebo patients (ITT set, Table 4). Details of these results were comprehensively published in 2015 [17],

IPSS QoL

In each study, the IPSS QoL index decreased in both treatment groups, with major improvement occurring during the first 3months. After 12 months, the mean reduction in the IPSS QoL score from baseline in the Bach study was 1.3 and 1.1 points in the PSE and placebo groups, respectively, and the mean decreases in the GRANU study were 1.2 points (PSE group) and 1.0 points (placebo group) (Supplementary Table D).

Urological measurements

Qmax, PVR, and PV changes from baseline to 12 months of treatment are shown in Table 7. In either study, the Qmax improved in both treatment arms. In the Bach study, the increase was 3.9 mL/s and 3.3 mL/s in the PSE and placebo groups, respectively, and in the GRANU study, the mean difference from baseline was the same for both treatments (+ 3.6 mL/s).

The median reduction in PVR was 10 mL in both treatment groups of the Bach study, while in the GRANU study, which excluded patients having a PVR over 100 mL, the median change from baseline after 12 months was zero (0.0 mL) in either group (Table 7).

The individual changes in PV from baseline to the 12-month follow-up varied between a 123 mL increase and a 70 mL decrease in the Bach study. In addition to potential measurement errors, transient prostatic congestions most likely concomitantly present at baseline or after 12 months accounted for large volume changes in individual patients. The mean (SD) change from baseline was − 1.7 (15.1) mL and - 4.0 (17.7) mL in the PSE and placebo groups, respectively. Conversely, in the GRANU study, PV values were increased at study end compared to baseline. The mean/median increases after 12 months were 2.4 / 1.0 mL and 2.6 / 2.0 mL in the PSE and placebo groups, respectively (Table 7).

Discussion

In everyday clinical practice, phytotherapy is part of standard therapy for men with uncomplicated but bothersome LUTS suggestive of BPH. The favorable safety profile of herbal extracts is well recognized. However, the efficacy evidence is mainly based on highly heterogeneous studies, and meta-analyses should be interpreted with caution [4, 6, 7].

The present meta-analysis on the efficacy of soft extract from Uromedic® pumpkin seed (PSE) pooled two studies with basically homogenous designs. Both the Bach and GRANU studies were placebo-controlled and had the same follow-up period of 12 months.

Consistent with the primary studies, an IPSS reduction of ≥5 points from baseline was the threshold of response in the pooled analysis. Thus, responders had an improvement that exceeded the minimum 3-point improvement a patient needs to perceive a clinical benefit [21,22,23].

The analyzed population consisted of 1389 men aged 64 years on average. At baseline, the patients had mean IPSS and IPSS QoL scores of 16.6 and 3.4 points, respectively. This level of moderate LUTS and related impact on quality of life is typical for patients who seek medical advice and treatment for symptomatic relief [2].

In these patients, the meta-analysis results suggest beneficial effects of PSE in terms of symptomatic relief. After 12 months, the response rate in the PSE group was 54% (ITT-LOCF) compared to 51% in the placebo group. An advantage of PSE treatment was also observed in the PP set, with response rates of 58% and 53% in the PSE and placebo groups, respectively. Additionally, the mean IPSS change from baseline was greater with PSE (− 5.3 points) than with placebo (− 4.8 points).

The magnitude of symptomatic relief achieved with PSE after 12 months compares favorably with the findings by Debruyne et al. [25] for hexanic saw palmetto extract and tamsulosin. With these treatments, only 49% of the patients reached an IPSS decrease of ≥5 points, and the mean IPSS change from baseline was − 4.4 points. Notably, based on that study, the European herbal monograph has granted well-established use status for hexanic saw palmetto extract [26].

In a recent meta-analysis, Russo et al. [27] found minimal differences in IPSS response between saw palmetto extracts and placebo. However, this analysis was based on studies with immense variability regarding essential features. The small number of patients resulted in significant imbalances in baseline IPSS with up to a 2.9-points difference between the treatment groups in 4 out of the 5 studies.

Another meta-analysis of all available published data for hexanic saw palmetto extract showed a mean IPSS decrease of 5.73 [2, 15]. However, only 6 of the primary studies were randomized controlled trials, of which only one had a 12-month follow-up period, and none was placebo-controlled [15, 25].

In contrast, the two PSE studies pooled for the present meta-analysis were equal regarding essential requirements of clinical research in patients with LUTS/BPH, such as placebo control and 12-month follow-up [13]. As we used patient-level data, our study also repeated the analysis of the individual studies [28]. All re-evaluation results were consistent with the initially reported study data [16, 17].

In both studies, the baseline characteristics were well-balanced between treatment groups. However, we noticed possibly important differences between the studies due to the strict selection criteria in the GRANU study conducted approximately a decade after the Bach study [16]. During this period, ARBs and 5-ARIs had become standard prescription drugs for symptomatic LUTS in Germany [6, 29]. Thus, the ethics committee tightened the inclusion criteria because treating patients with severe LUTS was no longer accepted in placebo-controlled studies. Consequently, the upper IPSS limit for inclusion was 19 points in the GRANU study [17].

To some extent, the lower baseline scores in the GRANU study might have reduced the magnitude of IPSS improvement. Notably, the logistic regression and ANCOVA models showed that symptom severity at baseline substantially influenced the IPSS reduction. These findings correspond to other observations, e.g., Roehrborn et al. reported that patients with baseline IPSS above 19 points reached a 3-point reduction more often than patients having lower scores at study start [23]. Similarly, Debryne et al. [25] reported a better treatment response in patients with severe symptoms compared to patients with moderate symptoms.

Of the patients studied in the present meta-analysis, approximately two-thirds (924 out of 1389) included in our meta-analysis had been treated in the GRANU study. Thus, a study population with less severe symptoms and no difference between PSE and placebo dominated the present results. Nonetheless, in the pooled analysis of both studies, the estimated response rates in the PSE group exceeded those in the placebo group by 3% and 5% in the ITT and PP sets, respectively. Furthermore, according to the logistic regression model, a patient’s likelihood of achieving a 5-point decrease in the IPSS was 17% higher in the PSE group than in the placebo group. Moreover, regarding the mean IPSS changes from baseline to 12 months the difference between the PSE and placebo groups also favored PSE treatment. In both treatment groups, the decrease in IPSS was accompanied by an improvement in the IPSS-QoL index.

Conclusions

Herbal preparations such as pumpkin seed soft extract may be an option for the symptomatic treatment of men with mild-to-moderate LUTS/BPH and low risk of progression. According to our meta-analysis of two 12-month studies, more patients treated with pumpkin seed soft extract achieved the clinically relevant response of an at least 5-point decrease in total IPSS than patients in the placebo group. The extract was well tolerated. No treatment-related changes were observed in laboratory safety parameters, including PSA values after 12 months of treatment [16, 17].

Accordingly, for men seeking symptomatic relief of LUTS, pumpkin seed soft extract may offer a favorable balance between desirable and undesirable outcomes, particularly for those who refuse or do not tolerate synthetic drugs. The possible association between treatment response and baseline IPSS warrants confirmation in future studies.