DIAGNOSTICS ASSESSMENT PROGRAMME

Benefits of MRI fusion biopsies from the trials

3.2 The committee agreed that the evidence for MRI fusion technology looked promising, but that there was a lot of uncertainty around the performance of the technology when compared with cognitive fusion. The EAG's network meta-analyses used pooled data from different software fusion technologies for its base-case analysis. The results suggested that software fusion, compared with cognitive fusion, may detect more International Society of Urological Pathology (ISUP) grade 2 or higher prostate cancer. But whether software fusion truly detected more higher-grade cancer, and if so by how much, was very uncertain. The EAG explained that few studies in the meta-analysis included people with higher-grade cancer, and few cases were detected in these studies.

3.3 The committee discussed the limitations of the clinical evidence included in the EAG's meta-analyses. None of the studies were done in the UK. The EAG judged all studies used in the meta-analysis to be at high risk of bias, and it stated that no high-quality randomised controlled trials have been published. The meta-analyses also showed moderate heterogeneity that could not be explained by differences in individual software fusion devices.

Performance differences between the different MRI fusion biopsy systems

3.4 The committee was uncertain how appropriate it was to use data generated using 1 technology to show the performance of others. There was limited data directly comparing the different technologies, and evidence levels varied across the different software fusion technologies. The EAG combined data from different technologies in its base-case analysis, based on advice from clinical experts. The committee noted that there are fundamental differences between the MRI fusion software systems which may influence outcomes. Only 2 software fusion technologies had more than 1 study included in the meta-analysis. For 1 of these (KOELIS Trinity), all studies used a previous version of the software included in the current device. The company commented that the KOELIS Trinity uses an updated version of the same software included in the KOELIS Urostation (which is now discontinued). The EAG did not identify any evidence for Fusion Bx 2.0 or FusionVu, and no identified studies for the bkFusion or ISR'obot Mona Lisa met the inclusion criteria for the meta-analysis. The EAG stated that evidence was insufficient to conclude whether any software fusion technologies were superior to others.

Potential impact on waiting times of adopting MRI fusion biopsy

3.5 The committee considered that adopting MRI fusion software could prolong waiting time for prostate biopsies, and that this was not captured in the cost-effectiveness estimates. The clinical experts explained that radiologist services are at full capacity in the NHS. Clinical experts with experience using fusion software highlighted that more preparatory time is needed per biopsy with fusion software than with cognitive fusion, although this does decrease with greater experience in using the technology. This lowers the number of biopsies that can be done in a day. In its economic model, the EAG assumed that additional procedure time for software fusion use would be 10 minutes in a high throughput centre (5 minutes radiologist time to import and obtain appropriate MRI sequences and 5 minutes during the biopsy), based on expert advice. It was suggested that this would be longer when the technologies were first used, because of lack of experience. The additional time was added to the cost of doing software fusion biopsy in the model (for example, for staff time), but the model did not include any impact of this on slowing biopsy throughput. The clinical experts commented that software fusion could improve patient throughput if it reduced the repeat biopsies needed (for example, if there was more confidence in a negative biopsy). The EAG commented that how much software fusion increases wating times would depend on whether the capacity constraint was access to the biopsy procedure or to the initial MRI (which would be the same for software and fusion biopsy).

Improving detection of higher-grade prostate cancer makes the technology more cost effective in the model

3.6 The cost-effectiveness estimates for software fusion biopsy (with or without systematic biopsy), compared with cognitive fusion, were generally favourable, but uncertain. Potential adverse impacts on the capacity of services to do biopsies (see section 3.5) were not considered in the model. The EAG's analysis showed that cost effectiveness was most affected by the improved detection of localised or locally advanced higher-grade cancers (ISUP2 or higher). When targeted fusion was combined with systematic biopsy, there were more cost savings and health benefits in the long-term model for software fusion than for cognitive fusion.

3.7 The EAG cautioned that the uncertainties in the clinical evidence should be considered alongside the overall cost effectiveness, because the evidence used in its network meta-analysis underpinned its economic model. The committee recalled that although the data suggested that software fusion may detect more higher-grade cancers, this was very uncertain (see section 3.2). It noted that a scenario analysis done by the EAG in which any benefit to detecting cancer for software fusion was removed from the model, the incremental cost-effectiveness ratios (ICERs) increased to over £500,000 per quality-adjusted life year (QALY) gained. The committee concluded that the uncertainty about how much better software fusion is at detecting higher-grade cancers means that the cost effectiveness of these technologies, while potentially favourable, is also uncertain.

Benefits to less experienced health professionals

3.8 The committee acknowledged that the technology could benefit less experienced healthcare professionals, and help to standardise care across different centres and improve accessibility. The clinical experts noted that biopsies are done by various healthcare professionals including urologists, radiologists and nurse specialists. The level of experience in doing biopsies varies, and specialist centres are more likely to have professionals with more experience. They suggested that the accuracy of cognitive fusion is highly dependent on the operator. The EAG identified 1 study that reported cancer biopsy positivity rates by operator experience (Stabile et al. 2018). It reported an increase in the biopsy positivity rates in the first 60 procedures, where it plateaued, regardless of the biopsy approach. Operator experience predicted the biopsy positivity rate of targeted biopsies, particularly for transrectal software fusion biopsy compared with transrectal cognitive fusion or transperineal software fusion.

3.9 The clinical experts suggested that using MRI fusion biopsy could help standardise biopsy quality. They said that this could minimise geographical variation in procedure standard caused by the variation in operator expertise. Improving the ability of less experienced operators to do the procedure may allow it to be available more widely, improving accessibility.

MRI fusion biopsy shows promise, but better quality evidence is needed

3.10 The committee recognised that MRI fusion software shows promise, could help improve standards across the NHS and patient access (see section 3.9) and may be cost effective. But it felt there was too much uncertainty about how much it would improve detection of prostate cancer to recommend its use for routine adoption (see sections 3.6 and 3.7). Further research is needed to reduce this uncertainty, and so also reduce the uncertainty about the cost effectiveness. The committee was also concerned about the potential impact of adopting software fusion technology on how many biopsy procedures can be done per day, and consequently on waiting times for this procedure (see section 3.5). Further data on this would help reduce concerns about the impact of adopting software fusion technologies.

An ongoing study may fill some evidence gaps

3.11 The ongoing IP7-PACIFIC trial (NCT05574647, accessed 14 November 2022) is likely to provide further, potentially high quality, data on software fusion biopsy for detecting clinically significant prostate cancers, compared with cognitive fusion biopsy. The primary outcome of the trial is the proportion of clinically significant cancers (defined as ISUP 2 or above) detected in people who had a biopsy with a suspicious MRI (MRI score 3, 4 or 5 on either the Likert or Prostate Imaging-Reporting and Data System [PI-RADS] schema). The EAG commented that the study aims to recruit 3,600 people with suspected prostate cancer in the UK, but that not all of these participants will provide data on software fusion performance (the study is 2 linked randomised controlled trials, and is also comparing biparametric and multiparametric MRI to detect clinically significant prostate cancers). The EAG emphasised that to help inform cost-effectiveness estimates from its model, the numbers of cancer detected by ISUP grade should be reported (rather than just the number of ISUP grade 2 or higher). Given the uncertainty about the generalisability of data generated using 1 software fusion technology to others (see section 3.4), it would be beneficial if data was reported separately for each fusion software. The committee concluded that this will be a useful trial for reducing uncertainty about software fusion biopsy performance, but noted that the estimated study completion data is not until 2026. It stated that people should be encouraged to participate in the trial.

Existing data and resources to collect data

3.12 The clinical experts highlighted that professional bodies such as the British Association of Urological Surgeons and the Royal College of Radiologists may have data that could contribute to a future assessment. They also suggested that establishing further registries may help data collection. The committee suggested that centres should consider contributing to existing audit tools such as the National Prostate Cancer Audit where possible.