Natural history of prostate cancer on active surveillance: stratification by MRI using the PRECISE recommendations in a UK cohort

Most policy groups around the world now recommend active surveillance (AS) as an appropriate option for patients with low-risk prostate cancer (PCa) [1, 2]. However, the current position is less than satisfactory as most of the evidence base to support its use is from single centres, no standard AS protocol exists and we lack a method of reliably ascertaining when true progression (by either grade or volume) occurs.

Multi-parametric MRI (mpMRI) has provided us with an opportunity to move away from random prostate biopsies and address the challenge of discriminating true progression from mere re-classification [3]. Recent evidence suggests that pre-biopsy MRI should be performed before confirmatory systematic transrectal ultrasound (TRUS)–guided biopsies during AS, together with MRI-targeted biopsies when indicated [4]. The ASIST trial initially showed no difference in the upgrade rate between patients having standard re-biopsy and those having MRI with 2 cores targeted to a lesion (upgrade to Gleason Grade Group 2 (GGG 2) was 21% vs 23% respectively, p = 0.9), although there were marked differences between the study centres [5]. However, after 2 years of follow-up, the authors found that baseline mpMRI before confirmatory biopsy resulted in 50% fewer failures of surveillance and less progression to higher-grade cancer, confirming the value of mpMRI during AS [6].

Addressing the serial use of MRI in AS, there are a small number of studies at present [7‐11] and there is a lack of standardised reporting in serial MRI data across cohorts [12].

Following the standardisation work initiated with the Prostate Imaging Reporting and Data System (PI-RADS) guidelines, the Prostate Cancer Radiological Estimation of Change in Sequential Evaluation (PRECISE) panel was convened in 2016 [13]. The panel included ten experts in urology, eight in radiology and one in radiation oncology from the UK, Europe and North America and their objective was to define the conduct and reporting standards for patients on AS having serial MRI scans. A set of 394 statements relevant to prostate MRI reporting during AS was scored for agreement on a 9-point scale by each panelist individually prior to the meeting. All scores were collated and consensus (or lack of consensus) for each statement was assessed using the RAND/UCLA criteria. Each statement was then discussed and rescored anonymously by each panelist during the face-to-face meeting, and the PRECISE checklist and case report template for reporting prostate MRI during AS were developed according to the consensus reached. The key features of the PRECISE recommendations are the measurement of each lesion at every time point, and a determination of the likelihood of radiological progression using a 1-to-5 Likert scale (PRECISE score) as shown in Table 1.

At present, there is limited literature on the application of the PRECISE recommendations in a clinical setting [14‐17].

At University College London Hospital, we have a cohort of > 620 patients who have undertaken MRI-guided AS. The cohort was established in 2005 in a prospective manner and is defined by patients who have had a prostate MRI and a biopsy-confirmed low- to intermediate-risk PCa (i.e. Gleason ≤ 3 + 4 and prostate-specific antigen (PSA) ≤ 20 ng/ml) as per UK National Institute for Health and Care Excellence (NICE) guidelines [18], and who have chosen AS as their initial management option. No maximum cancer core length or number of positive cores was stipulated for eligibility, due in part to the extensive use of targeted biopsies that can result in ‘risk inflation’ and exclude patients from AS unnecessarily. We also note that the number of involved cores is not part of disqualification criteria from AS according to recent European and UK NICE guidelines [19, 20]. At our institution, all clinical records and MR images are routinely reviewed as part of an audit performed for the internal evaluation of the AS service (Fig. 1) and no institutional review board approval was required [21].

In detail, the timing of MRI on AS at our centre is based on both baseline risk (i.e. presence of a visible lesion at mpMRI) and PSA/PSA density changes during follow-up. Biopsy recommendations are based either on the suspicion of progression on MRI, or on adverse PSA kinetics without MRI changes. All biopsies are performed using a transperineal approach.

Data are updated to February 15, 2020.

Our final cohort comprised a total of 553 patients on AS with two or more prostate MR scans, all of diagnostic quality (Fig. 2).

Overall, 232/553 (42%) patients were exclusively scanned on a 1.5-T and 8/553 (1.4%) exclusively on a 3-T scanner. Supplementary Table 2 shows the number of patients and MR scans included in the study.

Table 2 shows baseline and follow-up characteristics of our population. Overall, 306/553 (55%) patients had at least an additional biopsy, 178 (58%) of which were targeted by visual registration. All follow-up biopsies were performed using a transperineal approach.

Table 3 lists the histopathological data of the whole population stratified by PRECISE score. If we have a closer look at those patients classified as PRECISE 4 (n = 211), 141/211 (67%) developed a new lesion, 36 of which (26%) were upgraded on histology, whilst 70/211 (33%) showed an increase in lesion size or conspicuity from baseline MRI, 58 of which (83%) were upgraded on histology. More in detail, if we focus on those patients who showed biopsy progression in the PRECISE 4 group (94/211; 45%), 36/94 (38%) developed a new lesion whilst 58/94 (62%) showed an increase in lesion size or conspicuity from baseline MRI. In this subcohort (n = 94), the median baseline PSA was 5.9 ng/ml (4.45–8.55) and the median baseline PSA density was 0.15 ng/ml/ml (0.11–0.21).

This is the first report that applies the PRECISE criteria for radiological progression in a UK cohort of patients with PCa on an MRI-led AS programme.

We have shown that radiological stability is associated with 97% freedom from clinical progression at 5 years. We have also demonstrated that a PI-RADS score of 4 or 5 at entry to surveillance is associated with PRECISE 4 or 5 in up to 68% at a median of 5 years—that is, more than two-thirds of visible lesions on MRI will increase in size, conspicuity or stage over a 5-year period.

Additionally, an increase in PSA density over time is significantly associated with radiological progression (PRECISE 4–5).

The most appropriate definition of progression in patients on AS is a challenging one. Whilst inclusion of patients with low-volume GGG 2 is increasing, it is also sometimes stated as a threshold for initiation of active treatment. We chose GGG 3 as the histological threshold for clinical progression, as this fits with our clinical practice of recommending treatment in all of these patients. Some patients, particularly those who have GGG 2 at enrolment on AS, and who show radiological progression, wish to avoid a biopsy and proceed with active treatment. As they had already met the criteria for choosing active treatment at baseline, this approach was permitted.

We therefore chose a combined endpoint of either histological progression to GGG 3, or initiation of active treatment as the definition of clinical progression.

In our MRI-led AS programme, biopsies are not offered on a protocol basis. Instead they are offered to patients ‘for cause’. The triggers for the recommendation comprise adverse changes in MRI or PSA kinetics. This approach was developed in response to our work in this field, where we demonstrated a high negative predictive value of MRI for GGG 3 disease. Whilst this protocol differs from all other published protocols, we know that in practice, compliance rates with protocol biopsy can be as low as 20 to 30%, in both routine practice and formal studies such as PRIAS [25], and actual biopsy rates may be more similar in the two approaches than the protocols would suggest [26, 27].

Whilst some would see the lack of additional biopsy as a drawback, we believe that it has significant advantages in terms of patient acceptability and compliance, and that the risk of ‘loss of opportunity for cure’ is no higher using this approach than standard biopsy alone. For example, in our cohort, 1/553 patients (< 1%) developed nodal disease and 2/553 (< 1%) patients had bone metastases whilst on surveillance, with median follow-up of 6.3 years. This compares well to the Sunnybrook cohort [28], where 13/980 (1.3%) patients developed lymph node disease, and 18/980 (1.8%) patients developed bone metastases, with a median follow-up of 6.3 years.

Our study suggests that the widespread use of the PRECISE recommendations might have two main clinical consequences: (i) identification of patients on AS who progress (PRECISE 4–5) in a timely manner, promoting re-biopsy/treatment; (ii) avoidance of repeat biopsy (and a lower intensity surveillance schedule) for patients with PRECISE 1–3, reducing the burden of surveillance for the individual and the healthcare system.

We do not promote the adoption of our AS protocol in every clinical setting, but we provide here the first application of the PRECISE recommendations in a cohort of patients monitored according to the UK NICE guidelines.

Future randomised studies (e.g. a study design of ‘biopsy’ versus ‘no biopsy’ according to PRECISE, with presence or absence of clinical progression as the ultimate reference) will be more informative.

We acknowledge that the PRECISE score will require further validation in prospective multicentre cohorts, using a range of reporters. Further analysis in larger cohorts, including the Movember GAP3 cohort of over 15,000 patients on AS across 25 countries, is planned [2].