Estimating the percentage of patients who might benefit from proton beam therapy instead of X-ray radiotherapy

Abstract

Objectives: High-energy Proton Beam Therapy (PBT) commenced in England in 2018 and NHS England commissions PBT for 1.5% of patients receiving radical radiotherapy. We sought expert opinion on the level of provision.

Methods: Invitations were sent to 41 colleagues working in PBT, most at one UK centre, to contribute by completing a spreadsheet. 39 responded: 23 (59%) completed the spreadsheet; 16 (41%) declined, arguing that clinical outcome data are lacking, but joined six additional site-specialist oncologists for two consensus meetings. The spreadsheet was pre-populated with incidence data from Cancer Research UK and radiotherapy use data from the National Cancer Registration and Analysis Service. 'Mechanisms of Benefit' of reduced growth impairment, reduced toxicity, dose escalation and reduced second cancer risk were examined.

Results: The most reliable figure for percentage of radical radiotherapy patients likely to benefit from PBT was that agreed by 95% of the 23 respondents at 4.3%, slightly larger than current provision. The median was 15% (range 4-92%) and consensus median 13%. The biggest estimated potential benefit was from reducing toxicity, median benefit to 15% (range 4-92%), followed by dose escalation median 3% (range 0 to 47%); consensus values were 12 and 3%. Reduced growth impairment and reduced second cancer risk were calculated to benefit 0.5% and 0.1%.

Conclusions: The most secure estimate of percentage benefit was 4.3% but insufficient clinical outcome data exist for confident estimates. The study supports the NHS approach of using the evidence base and developing it through randomised trials, non-randomised studies and outcomes tracking.

Advances in knowledge: Less is known about the percentage of patients who may benefit from PBT than is generally acknowledged. Expert opinion varies widely. Insufficient clinical outcome data exist to provide robust estimates. Considerable further work is needed to address this, including international collaboration; much is already underway but will take time to provide mature data.

Figure 1.

The core data ‘tab’ showing main tumour sites, estimated 10-year survival, total patient numbers, numbers estimated to receive curative RT, the percentage of all curative RT for each site and how these patients are distributed across the age group categories of 5 years. The oldest category is ‘Age 90 and over’. Children were defined as 16 and under, with numbers interpolated within the 15–19 years age category. Note that the generic assumptions of percentages of (1) cancer patients receiving any RT (50%) and (2) the percentage treated with curative intent (60%) are shown at the top left (Cells C2 & C3, blue). In this scenario, the radical RT numbers are dominated by non-melanoma skin cancer (32%, 37895 patients), breast (12%), lung (10%) and prostate (10%). To use the NCRAS site-by-site data, the specific tumour RT data were entered into Columns E & F site-by-site, the dependency to Cells C2 and C3 was removed, and the numbers in each age category were scaled. Note that the total cases per annum receiving radical RT shown here with the generic assumptions (50%, 60% in cells C2 & C3) is 120,026 while the actual NCRAS data show that only 86064 patients received radical RT (see Table 2). H&N = Head and Neck, medullo = medulloblastoma, Mel = melanoma, GI = gastro-intestinal, Leuk = leukaemia.

Figure 2.

Dose escalation spreadsheet ‘tab’ that participants were asked to complete. Note that the columns for ages ‘Under 1’ and ‘1-4’ were pre-populated with 0%. Participants did not need to know patient numbers in each cell since this was pre-set in the spreadsheet, from the Core Data tab, shown in Figure 1.

Figure 3.

a. Numbers of participants who felt able (59%) to complete the spreadsheet or unable (41%) because they felt there was insufficient clinical information to make reasoned recommendations. b. Distribution of professional roles in the 23 participants who completed the full spreadsheet. The professional roles of the remaining 16 were: radiation (clinical) oncologists 2, clinical scientists in radiotherapy physics 9 and proton research scientists 5.

Figure 4.

a. Distribution of individual responses for overall percentage of patients who might benefit from PBT, counting patients only once and calculated from NCRAS site-by-site data (the graph for generic RT usage (50% receive RT, 60% radical) is identical). One outlier estimated that 92% might benefit. Although almost two-thirds (65%) of participants estimated the overall percentage of patients who might benefit to be between 0 and 20%, the minimum figure estimated by 95% of participants was 4.3%. b. Distribution of individual responses for potential benefit from MoB 2, ‘Reduction in dysfunction and toxicity in normal tissues’, and from MoB 3, ‘Dose escalation’. One participant estimated that a very high percentage (92%) of patients might benefit as a result of a very high estimate of gain from reduction in normal tissue toxicity (MoB 2). A different participant estimated a high benefit (35%) from dose escalation (MoB 3). This was the only participant who prioritised dose escalation ahead of reduction in normal tissue toxicity.

Figure 5.

Distribution of individual responses for the study presented here (n=23) compared to opinion results from the International Symposium on Proton Therapy (ISOP) held in Heidelberg in 2018 (results for adults only; expert international audience) (n=50) and for the Oncology Forum conference, held on-line in 2020 (non-expert oncology audience) (n=16). Note that children (in the NCRAS site-by-site dataset) account for only 0.5% of cases. Median for the individual participants in our study was 15%, while the consensus figure was 13%. Approximate medians for ISOP and the Oncology Forum were both 15%, with the majority estimating likely benefit from PBT ≤20%. The three data sets are quite similar although there may be differences between countries in the perceived potential benefit, manifest in the generally higher estimates shown in the ISOP results.