Cost–utility analysis of MR imaging-guided transurethral ultrasound ablation for the treatment of low- to intermediate-risk localised prostate cancer

Abstract

Background

Magnetic resonance-guided transurethral ultrasound ablation (MR-TULSA) is a new focal therapy for treating localised prostate cancer that is associated with fewer adverse effects (AEs) compared with established treatments. To support large-scale clinical implementation, information about cost-effectiveness is required.

Objective

To evaluate the cost–utility of MR-TULSA compared with robot-assisted radical prostatectomy (RARP), external beam radiation therapy (EBRT) and active surveillance (AS) for patients with low- to favourable intermediate-risk localised prostate cancer.

Design, setting and participants

A Markov model was developed targeting 60-year-old men diagnosed with low- to intermediate-risk localised prostate cancer over a time horizon of 40 years from the German Statutory Health Insurance (SHI) perspective. To assess the robustness of the results, deterministic and probabilistic sensitivity analyses were performed.

Intervention

Four different treatment strategies were compared: minimally invasive MR-TULSA, two definitive approaches (RARP and EBRT) and one observational strategy (AS).

Outcome measurements and statistical analysis

Outcomes were measured in overall costs, quality-adjusted life years (QALYs) and the incremental cost-effectiveness ratio (ICER).

Results

AS generated the highest number of QALYs (12.67), followed by MR-TULSA (12.35), EBRT (12.35) and RARP (12.20). RARP generated the lowest costs (€ 46 997) over one patient’s lifetime, while MR-TULSA was a slightly more expensive alternative (€48 826). The incremental cost-effectiveness ratio (ICER) of AS compared with RARP was €11 600 per QALY and of MR-TULSA compared with RARP was €12 193 per QALY, while EBRT was dominated. At a willingness-to-pay of €20 000 per QALY, the probability of being cost-effective is 44% for AS, 25% for RARP, 25% for MR-TULSA and 6% for EBRT.

Conclusions

All treatment options for 60-year-old men diagnosed with low- to intermediate-risk localised prostate cancer are affected by considerable uncertainty. Accepting high follow-up costs by applying a higher willingness-to-pay, AS is the most favourable treatment option.

Strengths and limitations of this study.

• This is the first modelling study targeting focal therapy, which is based on a lifelong time horizon.

• For comparators data on transition probabilities and adverse events were obtained from a large prospective multi-centre randomised controlled trial with 15 years of follow-up.

• Data on utility were based on a large cohort study with long-term survivors of prostate cancer.

• Due to a lack of high-quality clinical data on magnetic resonance-guided transurethral ultrasound ablation, transition probabilities were based on a prospective single-arm study and a prospective cohort study assessing magnetic resonance high-intensity focused ultrasound.

Background

According to the Federal Statistical Office in Germany, prostate cancer was responsible for 15.4% of all deaths in men in 2020.¹ In addition, the economic burden of prostate cancer is significant due to direct medical costs, which were estimated to be (inflation-adjusted) €326 million in Germany in 2023.²

Among all prostate cancer patients, those with low- and favourable intermediate-risk (of) prostate cancerlow or favourable intermediate risk of prostate cancer who have a low risk of tumour spreading are the largest group.³ For these patients, there are three main treatment strategies: active surveillance (AS), radical prostatectomy (RARP) or external beam radiation therapy (EBRT). Both definitive treatment options (RARP and EBRT) are associated with a significant risk of long-term adverse effects (AEs), mainly erectile dysfunction (ED), urinary incontinence (UI) and bowel problems (BP). However, compared with AS, both RARP and EBRT result in more favourable oncologic outcomes (ie, overall survival, disease progression and metastases).⁴

In recent years, focal approaches (eg, high-intensity focused ultrasound (HIFU), cryotherapy, vascular-targeted photodynamic therapy and magnetic resonance imaging-guided transurethral ultrasound ablation (MR-TULSA)) have been developed to achieve oncologic outcomes similar to those of the definitive treatments and, simultaneously, to reduce the risk of AEs. MR-HIFU is still the most performed among the focal approaches, but the share of German patients treated with MR-HIFU decreased from 90% in 2017 to 75% in 2019. In contrast, since 2018, there has been a steep increase in the application of MR-TULSA procedures in Germany.⁵ MR-TULSA is a minimally invasive technique, performed with a transurethral robot-guided ultrasound device that thermally ablates the prostate cancer tissue, under real-time MR-imaging monitoring. According to recent clinical studies, MR-TULSA leads to fewer AEs than RARP and EBRT but has similar oncologic outcomes.^{5 6}

In a recently published cost-effectiveness modelling study with a 10 year time horizon from the perspective of the British National Health Service, focal therapy (based on cryosurgery and HIFU) dominated EBRT and RARP for patients with non-metastatic prostate cancer.⁷ To evaluate the clinical and economic long-term consequences of applying MR-TULSA in low- and favourable intermediate-risk (of) prostate cancer in Germany, we compared MR-TULSA with three conventional treatment strategies (ie, AS, EBRT and RARP) from the perspective of the German Statutory Health Insurance (SHI).

Methods

To reflect the lifetime effects and costs of the treatment strategies, we developed a Markov model using TreeAge Pro Healthcare 2024 (figure 1). Markov models are a frequently employed methodology for the assessment of interventions with long-term costs and effects, which are challenging to ascertain through the analysis of follow-up data derived from clinical trials. Men with (i) newly diagnosed low- to favourable intermediate-risk prostate cancer according to the D’Amico risk classification and (ii) life expectancy of more than 10 years entered the model at the age of 60. Low risk was defined as cT1–cT2a, Grade Group 1, prostate-specific antigen (PSA) <10 ng/mL and favourable-intermediate risk as one intermediate risk factor (cT2b-c or PSA 10–20 ng/mL) assigned to Grade Group 1 or 2 (<50% biopsy cores positive). When entering the model, patients were assumed to have no functional limitations in terms of ED, UI or BP.

Figure 1. Markov model structure.

The model had a time horizon of 40 years, and 3 month cycles were used to reflect the schedule of follow-up visits in Germany (online supplemental tables S1 and S2). The analysis was conducted from the perspective of the German SHI and is reported according to the CHEERS checklist.⁸

Strategies for the comparison

In line with international recommendations,⁹ in Germany the costs of screening have to be borne by the patient. To draw a realistic depiction of clinical practice in Germany, for patients with clinically detected prostate cancer, we chose four different treatment strategies as comparators for the model: (i) MR-TULSA, (ii) RARP, (iii) EBRT and (iv) AS. According to the 2024 German guideline on prostate cancer, patients newly diagnosed with favourable localised prostate cancer should be offered clear information on the benign course of the disease, with AS as the recommended strategy. However, patients can still opt against AS and decide on an interventional curative treatment option.¹⁰ In this model, we included RARP and EBRT as comparators because they represent more than 90% of procedures performed in Germany.⁵ In contrast to options (ii)–(iv), MR-TULSA was included to meet the latest technical developments and to add an innovative focal approach with increasing therapy applications to our model.

• MR-TULSA, including follow-up with magnetic resonance (MR)-fusion biopsies. In case of clinical progression, patients could either be treated with a second MR-TULSA or with a salvage RARP.

• RARP was assumed to be performed with the robot-assisted approach because the use of the conventional open technique is declining.¹¹ For salvage treatment after RARP, patients were assumed to receive EBRT and 6 months of androgen deprivation therapy (ADT) with the gonadotropin-releasing hormone analogue leuprorelin.

• EBRT was assumed to be performed with intensity-modulated and image-guided techniques with 74–80 Gy.¹⁰ Salvage treatment options are a salvage RARP or ADT with leuprorelin for 2 years.

• AS, defined as monitoring the PSA value and performing digital rectal examinations (DRE), multi-parametric MRIs (mpMRIs) and biopsies following a pre-set protocol.¹⁰ In case the cancer progresses, or the patient opts out of the AS strategy, he received RARP or EBRT following the same treatment protocols described in (ii) and (iii).^{10 12}

If a patient develops a second, non-metastatic elevation of the PSA-value during any of the salvage treatment options described above, he receives a permanent ADT with leuprorelin.

Model overview

In accordance with previously published cost-effectiveness analyses,^{13 14} our Markov model comprises five health states (figure 1). All patients started in the state of ‘stable disease’. Patients who received a definitive treatment moved to the state ‘remission’. If a disease progression was biochemically identified, patients could move from ‘remission’ to ‘clinical progression’. ‘Clinical progression’ after RARP was defined as two consecutive PSA-value measurements exceeding 0.2 ng/mL.¹⁵ ‘Clinical progression’ after EBRT or MR-TULSA was defined as exceeding 2 ng/mL following the Phoenix definition¹⁶ or any other progression of the diseases (eg, local recurrence) not categorised as a metastasis. In the case of ‘clinical progression’, patients received the salvage therapy depending on their first treatment (without the option of returning to the ‘stable disease’ state). From each of the described health states, patients could move to the states ‘metastasis’ or ‘death’. The health state ‘metastasis’ was defined as metastatic hormone-sensitive prostate cancer (mHSPC). In the case of metastasis, patients were considered to be hormone sensitive because the treatment strategies (i–iv) do not comprise definitive hormonal treatment, and thus, patients cannot achieve hormone resistance. AEs of the different treatments were depicted within the inherent health state.

Model input parameters

Sources of data

Data used for the calculation of transition probabilities and utility values were obtained from literature reviews. To identify appropriate input parameters, systematic literature searches in MEDLINE were conducted. Studies were selected based on methodological quality and applicability to the study context. All relevant literature was either obtained through open-access publications or institutional access. The search strategies are given in online supplemental table S3.

Probabilities

Probabilities of AEs after MR-TULSA were obtained from a single-arm, prospective cohort (n=115 patients) with 1 year follow-up (online supplemental table S4). Because in that study there were no incident AEs between 1 and 3 years, in our model AEs were assumed to be stable in the long term.^{6 17}Because follow-up in most studies on MR-TULSA was up to 1 year and/or was based on small sample sizes (n<30),^{6 17} we assumed that the probability of clinical progression or metastasis after MR-TULSA would be similar to that evaluated in a prospective cohort of intermediate-risk prostate cancer patients for MR-HIFU.¹⁸

Transition probabilities and probabilities of AEs (online supplemental table S4) for the three comparators RARP, EBRT and AS were obtained from the ProtecT trial, a prospective multi-centre randomised controlled trial with 1643 participants over 15 years.^{4 19} Data from the ProtecT trial were preferred because data on the PREFERE trial were limited due to poor participation of patients.¹²

All-cause mortality was age-adjusted, taken from the German federal statistical office.²⁰ Patients with metastases were assigned to have higher cancer-specific mortality.²¹

Utility values

Utilities were taken from a standard gamble study with 1884 prostate cancer survivors from the USA, the CaPSURE study,²² which addressed various disease states of prostate cancer (table 1). A temporary decrease in utilities was assigned due to definitive treatments,²³ while a permanent reduction was assumed due to the occurrence of AEs (ie, UI, ED and BP), according to the proportion of specific AEs per treatment strategy (online supplemental table S4).^{19 24} To combine utility values, the multiplicative method was used.

Table 1. Input parameters.

Parameter	Mean	SD	Source
Utility values
Stable disease	0.907	0.121	22
Remission	0.869	0.151	22
Biochemical recurrence	0.865	0.156	22
ADT without metastasis	0.833	0.187	22
Metastasis	0.826	0.190	22
Prostatectomy	0.670	0.290	23
Radiation	0.730	0.300	23
MR-TULSA	0.775	0.233	23/self-calculated
Potent	0.901	0.132	22
Impotent	0.870	0.159	22
Continence (no pad/day)	0.922	0.116	22
Incontinence ≥11 pad/day)	0.886	0.141	22
Bowel function (no problem)	0.923	0.114	22
Bowel function (problem)	0.822	0.195	22
Health state transition probabilities
AS
Stable disease—primary treatment	2.12%	0.62%	4
Stable disease—metastasis	0.71%	0.36%	4
RARP (primary)
Remission—biochemical Recurrence	0.8%	0.38%	4
Remission—metastasis	0.35%	0.25%	4
EBRT
Remission—biochemical Recurrence	0.84%	0.39%	4
Remission—metastasis	0.37%	0.26%	4
MR-TULSA (primary and salvage)
Remission—biochemical Recurrence	2.52%	0.882%	18
Remission—metastasis	0.20%	0.251%	18
Salvage EBRT with 6 months of ADT
BCR—metastasis	2.64%	0.057%	36
Salvage ADT
BCR—metastasis	8.97%	3.278%	37
Salvage RARP
BCR—metastasis	2.58%	0.789%	38
Any treatment
Metastasis—death	6.45%	1.592%	21
Any state—death (60–65 years)	1.16%	0.34%	20
Any state—death (65–70 years)	1.83%	0.42%	20
Any state—death (70–75 years)	2.64%	0.51%	20
Any state—death (75–80 years)	4.2%	0.63%	20
Any state—death (80–85 years)	6.93%	0.8%	20
Any state—death (85–90 years)	12.66%	1.05%	20
Any state—death (90–95 years)	21.57%	1.3%	20
Any state—death (95–100 years)	30.84%	1.46%	20

ADTandrogen deprivation therapyASactive surveillanceBCRbiochemical recurrenceEBRTexternal beam radiation therapyMR-TULSAmagnetic resonance imaging-guided transurethral ablationRARPradical prostatectomy

Input parameters are shown in table 1.

Costs

Cost data applied in the model were calculated based on clinical guidelines and consultation with experts to reflect the resource consumption in Germany.¹⁰ The direct medical costs included the costs of performing different treatment strategies including specific follow-up protocols. A detailed cost breakdown included in the analysis is provided in online supplemental tables S5–S10. Costs related to the treatment of AEs were included depending on their coverage by the SHI (eg, the costs of treatment for UI are covered, while those for ED are not) (table 2).

Table 2. Cost calculation.

Treatment strategy	Costs (in €) per year
	1	2	3	4	5	6 +
Primary treatment
MR-TULSA	11 378	1470	279	279	139	139
AS	1676	1 676	707	707	279	421
RARP	11 066	557	279	279	139	139
EBRT	8028	557	279	279	139	139
Clinical progression
Salvage ADT with leuprorelinn for 2 years	1859	1859	139	139	139	139
Permanent salvage ADT with leuprorelin	1859	1859	1859	1859	1859	1859
Salvage radiation + 6 months of leuprorelin	6055	557	279	279	139	139
Salvage RARP		557	279	279	139	139
Salvage MR-TULSA	10 774	1470	279	279	139	139
Metastasis
Enzalutamid + leuprorelin	39 609	39 609	39 609	39 609	39 609	39 609

ADTandrogen deprivation therapyASactive surveillanceBCRbiochemical recurrenceEBRTexternal beam radiation therapyMR-TULSAmagnetic resonance imaging-guided transurethral ablationRARProbot-assisted radical prostatectomy

Costs for treatment, surveillance and follow-up were valued according to publicly available sources for the reimbursement of patients insured in the German SHI (eg, the German diagnosis-related groups (DRG)-catalogue). For the calculations, all costs were adjusted to €2024. The generalisability of our results for the German context was assured by averting regional differences in prizing (eg, the lump sums for reimbursements of inpatient treatment were calculated with the base case value proposed by the DRG research group).²⁵

Base case analysis

To capture the differences between strategies, the incremental cost-effectiveness ratio (ICER) between the four examined treatment strategies was calculated as costs per quality-adjusted life year (QALYs). Because low- to favourable intermediate-risk prostate cancer usually has a benign course, with a 15 year cancer-specific survival rate, the main differences between the different strategies derive from adverse events affecting the quality of life of patients. Therefore, calculating the ICER as cost per QALY was deemed more meaningful than cost per life year gained. An annual discount rate of 3% for costs and utility values was applied in line with German methodological guidelines.²⁶

Sensitivity analyses and model validation

To assess parameter uncertainty, we carried out deterministic sensitivity analyses (DSAs) of all variables (ie, varying input parameters within the 95% CI and assessing the impact on the ICER). For inpatient costs, German-DRG lump sums were varied according to the minimum and maximum length of stay. The costs of MR-TULSA and RARP varied within the same adapted range.¹³

In probabilistic sensitivity analysis, all input parameters were varied simultaneously according to predefined distributions: utility values and transition probabilities were assumed to be beta-distributed and costs to be gamma-distributed. Results from the probabilistic sensitivity analysis were plotted in a cost-effectiveness acceptability curve, showing the probability of each strategy being cost-effective at different thresholds of willingness-to-pay(WTP).

In a structural sensitivity analysis, we assessed the impact of applying shorter time horizons for the model (ie, 5, 10 and 20 years).

Validation efforts are reported according to the AdViSHE tool and reported in online supplemental table S11.²⁷

Patient and public involvement

Patients or the public were not involved in the design, conduct or reporting of our research.

Results

Base case

AS generated the highest number of QALYs (12.67), followed by EBRT (12.35), MR-TULSA (12.35) and RARP (12.20). In contrast, RARP was the strategy with the lowest lifetime costs (€46 997). MR-TULSA (€48 826), AS (€52 449) and EBRT (€54 263) were more expensive strategies. EBRT was an absolutely dominant treatment option (more expensive and less QALYs than AS). Compared with RARP, the additional costs of AS were €5452, resulting in an ICER of €11 600 per QALY (MR-TULSA vs RARP: €12 193 per QALY). Results from the base case are shown in figure 2 and online supplemental table S12.

Figure 2. Base case results.

Sensitivity analyses

Results from DSA are shown in online supplemental figure S1. The parameters most affecting the model were the probability of metastasis after any treatment alternative and the direct costs of RARP, EBRT and MR-TULSA. For all comparisons, a lower/higher probability of metastasis after stable disease/remission resulted in the largest range of the cost-effectiveness ratio to (MR-TULSA vs RARP: cost-saving to €280 000; RARP vs AS: cost saving to €24 000 and EBRT vs RARP: €9000 to €53 000).

The probabilistic sensitivity analysis showed that at a WTP of zero RARP would be the preferred option (because of the lowest costs). Assuming a WTP of €10 000 or more a decision maker would favour AS (figure 3 and online supplemental figure S2). Assumed WTPs above 80 000 €/QALY, for MR-TULSA the probability of cost-effectiveness was between 20% and 30%.

Figure 3. Cost-effectiveness acceptability curve.

Structural sensitivity analyses showed that for a time horizon of 5 or 10 years, AS dominates all other strategies (ie, less costly and more QALYs). For a time horizon of 20 years, RARP becomes the cheapest strategy, followed by MR-TULSA and AS. Thus, for a time horizon of 20 years, the ICER for AS is € 3131 per QALY and the ICER for MR-TULSA is € 1733 per QALY (online supplemental table S12).

Validation

Cross validity was assessed by comparing two models from the perspectives of the French National Health Insurance and the British National Health Service (NHS).^{7 13 14 28} The comparison revealed that these differed with regard to the chosen time horizons, the model structure, input data used for the model (eg, utilities and transition probabilities) and the strategies compared (focal therapy vs AS, focal therapy vs RARP or EBRT). Detailed results of the validation can be found in online supplemental table S11. All assumptions made in the model are detailed in online supplemental table S13.

Discussion

This is the first cost–utility analysis of MR-TULSA for the treatment of low- and favourable intermediate-risk localised prostate cancer. Our results show that over a lifetime horizon, RARP is the cheapest treatment alternative, whereas AS and MR-TULSA are cost-effective alternatives with an ICER of €116,600 per QALY and €12 193 per QALY, respectively. Compared with the definitive treatment options RARP and EBRT, MR-TULSA would meet the economic criteria for positive reimbursement decisions in German hospitals.^{29 30} However, for patients accepting or even preferring a non-definitive treatment option, AS would yield the highest benefit at acceptable costs.

The most influential parameters for the cost-effectiveness of the MR-TULSA strategy were the costs of the procedure and the post-treatment probability of metastasis or clinical progression. In the probabilistic sensitivity analysis, the probability of MR-TULSA being cost-effective only ranges between 16% and 37% depending on the willingness-to-pay threshold. Moreover, the structural analysis showed that MR-TULSA is not cost-effective for shorter time horizons (5–10 years) due to high initial treatment costs. In the model, the costs of MR-TULSA are offset by the long-term benefits at longer time horizons. However, these benefits are yet to be demonstrated with prospective long-term follow-up.

If future studies can confirm the short-term benefits for long-term oncologic outcomes, MR-TULSA is likely to be a cost-effective focal treatment option for low- to intermediate-risk localised prostate cancer. To date, evidence on MR-TULSA is promising for functional outcomes but still immature for long-term safety and efficacy.³¹ A multicentre, prospective two-arm RCT, the ‘CAPTAIN’ trial (Clinical Trials registration number: NCT05027477), is ongoing to assess the effectiveness of MR-TULSA (including the proportion of patients free from treatment failure and overall survival) compared with RARP over a period of 10 years.³² Once this trial is finished by 2031, an update of our model will be opportune. Additionally, improved clinical outcomes from MR-TULSA are expected to result from learning curve effects and a more targeted patient selection (eg, prostate calcifications, elderly persons and anticoagulation).^{5 6}

Our results are opposite to those of a cost-effectiveness modelling published in 2023.⁷ According to Reddy et al, focal therapy dominated EBRT and RARP for patients with non-metastatic prostate cancer, while AS was not considered as a treatment option. The study differed from ours in the chosen time horizon (10 years vs lifelong in our analysis) and the focal treatment modalities (cryotherapy and HIFU vs MR-TULSA). In addition, in contrast to our study, transition probabilities were derived from a series of prostate cancer registries that reported clinical outcomes for patients undergoing RARP, EBRT and focal therapy, whereas adverse events were not considered.⁷ The superiority of focal therapy was mainly driven by the low costs of cryotherapy and HIFU in the UK, which were among the most influential parameters on the cost-effectiveness ratio.⁷ In addition, while Reddy et al estimated the primary cost for focal therapy to be half of that of RARP, the current reimbursement for MR-TULSA (including follow-up costs) by the German SHI is significantly higher. To date, the lump sum reimbursed for MR-TULSA is based on local arrangements between healthcare providers and the SHI; that is, once MR-TULSA is included in the general German-DRG catalogue, the lump sum will be renegotiated. Furthermore, follow-up plans for MR-TULSA include cost-intensive mpMRIs and MR-fusion biopsies and, from year 2 onwards, patients require more frequent follow-up visits to urologists than what is required for RARP and EBRT (online supplemental table S1). If the long-term oncologic safety of MR-TULSA is confirmed, the follow-up scheme for patients treated with MR-TULSA could become less resource-demanding, similar to the follow-up schemes after EBRT and RARP.

This is the first cost-effectiveness analysis for MR-TULSA for low- to intermediate-risk prostate cancer patients. Reddy et al compared focal therapy (including cryotherapy and HIFU) to intermediate- and high-risk prostate cancer patients, for whom active surveillance is an unsuitable option.⁷ In a literature-based modelling study from the French National Health Insurance perspective, AS dominated focal therapy for a 30 year time horizon. Indeed, the probability that focal therapy is cost-effective was 45.5% at a WTP of €30 000/QALY, indicating a high level of uncertainty (as in our study). Similarly to our study, the vast majority of the uncertainty resulted from—among others—transition probabilities related to focal therapy cancer.²⁸ However, for the comparison between a definitive treatment option and AS, the patient’s preference may be more directive for the treatment decision than cost-effectiveness. The patients’ choice between a definitive treatment and AS depends on the individual risk preference between maintaining the short-term quality of life (ie, due to avoidance of treatment-related AEs) and improving long-term quality of life (ie, due to decreased risk of cancer progression).³³ MR-TULSA could fill the gap, compromising good cancer control and high quality of life and therefore should be offered as a third treatment alternative to patients besides invasive and observational approaches.

Strengths and limitations

The main strength of our model is that—in contrast to previous analyses—we could rely on 15 year follow-up data on oncologic outcomes from the ProtecT trial.⁴ To respond to the degree of uncertainty of long-term outcomes (eg, mortality) from different treatment strategies for low-risk prostate cancer, a series of structural sensitivity analyses were conducted, exploring shorter time horizons for the model (10, 15 and 20 years). These analyses were informed directly by the data from the ProtecT trial.⁴

In addition, we could apply utility data from the CaPSURE study, a large cohort of long-term survivors from 2019.²² Because cancer therapies have developed over the last two decades, the availability of these updated clinical evidence and utility values reflects the clinical course of prostate cancer patients appropriately.

Some limitations have to be acknowledged. First, due to a lack of long-term data on MR-TULSA, the long-term data of a 5 year follow-up of patients treated with MR-HIFU were used as a proxy. This choice was justified by the similarity in mechanisms of action between these methods, which are expected to lead to similar oncological outcomes.^{18 31} In addition, this assumption was validated by clinical experts (ie, face validity) and cross-validation (ie, comparison to previous models). Previous models have already assumed the interchangeability of different focal therapies (cryotherapy and MR-HIFU) with regard to the expected related oncologic outcomes.⁷ Therefore, we consider our model a proper and sufficient analysis that can serve as a solid basis for deciding to adopt the MR-TULSA in the German SHI system or postponing its adoption until long-term evidence is available.

Second, the high overall costs of the EBRT strategy were mainly driven by the costs of ADT, while re-irritation (ie, brachytherapy, stereotaxic radiotherapy or EBRT) was not considered an alternative salvage treatment option for patients with loco-regional failure. In addition, innovative technical features in EBRT such as intensity-modulated radiotherapy were not considered in the ProtecT trial (and thus also excluded from the model).⁴

A further concern could be that our model included only 60-year-old men based on PSA testing, which makes the eligibility of these findings for Germany questionable. According to the German guideline on prostate cancer, PSA testing should only be performed (i) after clarification that the risk of overdiagnosis is not offset by the oncologic outcomes and (ii) if the patient strongly desires to undergo screening (in that case the cost of the test should be borne by the patient).¹⁰ However, PSA screening is still often performed in Germany,³⁴ and the present cost–utility analysis did not address a PSA-based screen-and-treat strategy for prostate cancer; rather it compares treatment strategies for patients who were already diagnosed as low- and favourable intermediate-risk prostate cancer. In addition, in contrast to the German guideline, the European Association of Urology recommends PSA as the primary screening test.³⁵ This recommendation is followed in several European countries with public health insurance (eg, Sweden). Until this divergence in clinical guidance can be solved, it remains relevant to provide preliminary evidence of the cost-effectiveness of different treatment options for patients with low-risk prostate cancer.

Conclusion

AS is the most cost-effective treatment modality for patients with low- to favourable intermediate-risk prostate cancer. Considering the current evidence base, MR-TULSA can be cost-effective from the perspective of the German SHI.

Data availability statement

Data are available upon reasonable request.