Five-fraction Radiotherapy for Breast Cancer: FAST-Forward to Implementation Clinical Oncology

Introduction: The phase 3 FAST-Forward trial reported outcomes for 26 and 27 Gy schedules delivered in 5 fractions over 1 week versus 40 Gy in 15 fractions over 3 weeks in 4000 patients. We discuss concerns raised by the radiotherapy community in relation to implementing this schedule. Ipsilateral Breast Tumour Relapse (IBTR): Published estimated 5-year IBTR with 95% CI after 40 Gy in 15 fractions was 2.1% (95% CI 1.4 e 3.1), 1.7% (1.2 e 1.6) after 27 Gy and 1.4% (0.2 e 2.2) after 26 Gy, emphatically showing non-inferiority of the 5-fraction regimens. Subgroup analyses comparing IBTR in 26 Gy versus 40 Gy show no evidence of differential effect regarding age, grade, pathological tumour size, nodal status, tumour bed boost, adjuvant chemotherapy, HER2 status and triple negative status. The number of events in these analyses is small and results should be interpreted with caution. There was only 1 IBTR event post-mastectomy. Normal tissue effects: The 26 Gy schedule, on the basis of similar NTE to 40 Gy in 15 fractions, is the recommended regimen for clinical implementation. There is a low absolute rate of moderate/marked NTE, these are predominantly moderate not severe change. Subgroup analyses comparing clinician-assessed moderate or marked adverse effect for 26 Gy versus 40 Gy show no evidence of differential effects according to age, breast size, surgical de ﬁ cit, tumour bed boost, or adjuvant chemotherapy. Radiobiological considerations: The design of the FAST-Forward trial does not control for time-related effects, and the ability to interpret clinical outcomes in terms of underlying biology is limited. There could conceivably be a time-effect for tumour control. A slight reduction in a / b estimate for the late normal tissue effects of test regimens might be a chance effect, but if real could re ﬂ ect fewer consequential late effects due to lower rates of moist desquamation. Conclusion: The 26 Gy 5-fraction daily regimen for breast radiotherapy can be implemented now.


Introduction
Over the last 30 years, breast radiotherapy fractionation has been systematically investigated and debated. Moderate hypofractionation, using 15 or 16 fractions over 3 weeks delivering total doses in the range 40e42.5 Gy, has become the widespread international standard [1e3]. Recent published studies of five-fraction breast radiotherapy describe safe, effective and simpler regimens that will probably become a new standard of care. The phase III randomised FAST-Forward trial [4] reported outcomes, in relation to both tumour control and normal tissue effects (NTE), for 26 and 27 Gy in five fractions over 1 week versus 40 Gy in 15 fractions over 3 weeks in more than 4000 patients. Selection of total doses for the five-fraction schedules was informed by earlier trials, including the FAST trial of 915 patients testing 28.5 and 30 Gy in five fractions delivered once weekly versus 50 Gy in 25 fractions over 5 weeks, which has now published 10-year results [5].
At a consensus meeting held in October 2020 under the auspices of The Royal College of Radiologists, the UK breast cancer radiotherapy community voted for five-fraction breast radiotherapy as its new national standard for breast radiotherapy (https://www.rcr.ac.uk/publication/postoperativeradiotherapy-breast-cancer-hypofractionation-rcr-consensusstatements). However, some commentators suggest caution in adopting the schedule now. In the Editorial accompanying the FAST-Forward results, Levy and Rivera [6] agree that results are practice-changing for low-risk patients, but want longer-term disease outcomes and clinically defined subgroup analyses. Offersen et al. [7] argued that 26 Gy in five fractions is expected to be less effective than 40 Gy in 15 fractions based on conventional a/b estimates. We explore these issues, together with recurring themes that have come up when presenting the data since publication.

Ipsilateral Breast Tumour Relapse
FAST-Forward is a non-inferiority trial with ipsilateral breast tumour relapse (IBTR) as the primary end point. Based on START trial data [8,9] and incorporating subsequent improvements in surgical technique and systemic therapy, it was anticipated that the incidence of IBTR by 5 years in the FAST-Forward control group giving 40 Gy in 15 fractions would be 2%. Following discussions with clinicians and patient advocates, a non-inferiority margin of 1.6% was prespecified in the protocol, which required a sample size of 4000 patients. The estimated 5-year incidence of IBTR after 40 Gy in 15 fractions was 2.1% (95% confidence interval 1.4e3.1), 1.7% (1.2e1.6) after 27 Gy and 1.4% (0.2e2.2) after 26 Gy. These upper confidence limits excluded an increase in IBTR of >1.6% after both five-fraction schedules, with P ¼ 0.0022 and P ¼ 0.00019 for non-inferiority of 27 and 26 Gy schedules, respectively, compared with 40 Gy in 15 fractions.
Commentators refer to subgroups for whom hypofractionation may not be so effective, but the empirical data are reassuring. The 2011 American Society for Radiation Oncology evidence-based guidelines [12] were unable to confirm agreement on hypofractionation in patients <50 years. The small number of patients included in trials and an increased risk of IBTR at a young age were cited, but no evidence of an adverse outcome by age after hypofractionation has been reported. A post-hoc analysis of tumour grade in the Ontario Clinical Oncology Group trial [11] suggested an interaction of grade and randomisation group, but subsequent central analysis of tumour blocks reported no trend for patients with high-grade tumours to be disadvantaged after 42.5 Gy in 16 fractions [13]. They also found that tumour grade and molecular subtype did not predict response to hypofractionation. A subgroup metaanalysis of locoregional relapse was carried out of the START-P [14], -A [8] and -B [9] trials in 5861 patients reporting 10-year results [10]. Treatment effects of hypofractionation in terms of tumour control were not significantly different from 50 Gy in 25 fractions when examined by age, type of primary surgery, axillary node status, tumour grade, use of adjuvant chemotherapy or boost radiotherapy. The 2018 American Society for Radiation Oncology evidence-based guidelines approved hypofractionated breast radiotherapy with 40/42.5 Gy in 15/16 fractions over 3 weeks, irrespective of age, tumour grade or receptor status [1].
Wang et al. [15] reported on 810 patients in a single institution randomised non-inferiority trial of hypofractionated radiotherapy post-mastectomy. All patients underwent axillary dissection and were at least four-node positive or T3e4, unless they received neoadjuvant chemotherapy, in which case either clinical stage III or pathological axillary node-positive patients were eligible. The hypofractionated schedule was 43.5 Gy in 15 fractions over 3 weeks versus 50 Gy in 25 fractions over 5 weeks as standard. The radiotherapy target volume included the chest wall, level 3 axilla and supraclavicular fossa. The 5-year locoregional recurrence rate was 8.3% (90% confidence interval 5.8e10.7) with the 15-fraction schedule and 8.1% (90% confidence interval 5.4e10.6) with the 25-fraction schedule. With a P < 0.0001 for non-inferiority they concluded that the hypofractionated regimen was non-inferior to standard.
FAST-Forward reported non-inferior IBTR for both fivefraction schedules and given the preceding arguments we would not expect inferiority to be observed in any subgroups, but what did we find? In total, 1545 (37.8%) patients were in the high-risk category as defined by age <50 years, grade 3 tumour or both, and this was a stratification factor at randomisation. Retrospective subgroup analyses comparing IBTR in 26 Gy versus 40 Gy provided no evidence of a differential effect according to age, grade, pathological tumour size, nodal status, tumour bed boost, adjuvant chemotherapy, HER2 status and in triple-negative patients ( Figure 1). Confidence intervals for the hazard ratios overlapped for the subgroups, although the number of events in these analyses was small (52); hence, results should be interpreted with caution, as the statistical power was low. Subgroup analysis according to type of primary surgery was not possible as there was only one IBTR event postmastectomy in a control group patient (out of 91) and none in the 173 patients treated with five fractions. Table 1 shows the frequencies and number of patients in each category of age <50 years, grade 3, post-mastectomy, triplenegative and HER2-positive tumours. No evidence to signal concern was seen for the five-fraction schedules. The use of boost and dose/fractionation, both declared prior to randomisation, was balanced between the three treatment groups, minimising the risk of bias in dose intensity between trial groups.

Normal Tissue Effects
In FAST-Forward, late NTE assessed by clinicians, patients and photographs were key secondary end points. The 26 Gy in five daily fractions schedule, on the basis of similar NTE to 40 Gy in 15 fractions, is the recommended regimen for clinical implementation. By 'similar' we mean that NTE were neither statistically nor clinically significantly different from 40 Gy in 15 fractions with respect to clinician-or patient-assessed outcomes, including photographic assessments conducted blind to treatment allocation. The 27 Gy five-fraction schedule was statistically significantly different to the 40 Gy standard for many late NTE and also to the 26 Gy schedule, confirming the sensitivity of trial outcome measures to detect a difference in dose intensity corresponding to 3 Gy in 2 Gy equivalents assuming a/b ¼ 2 Gy (see below). The 27 Gy five-fraction regimen exhibited late NTE rates of comparable magnitude to 50 Gy in 25 daily fractions. To provide some perspective for the late NTE after five-fraction regimens, 40 Gy in 15 fractions is equivalent to about 46 Gy in 2 Gy fractions in terms of late NTE compared with 50 Gy in 25 fractions according to START trial outcomes [11]. In FAST-Forward, the 26 Gy regimen is comparable to 47 Gy in 2 Gy equivalents in terms of late NTE.
Early NTE are much less responsive to fraction size than late NTE, the contribution of total dose being relatively more important [16]. FAST-Forward offers a good example in that breast erythema was less intense and also settled a fortnight earlier after five-fraction than 15-fraction schedules [17]. In this context, the milder erythema was a response to 26 and 27 Gy total dose levels much more than to fraction sizes of 5.2 and 5.4 Gy. Acute reactions were also milder in both five-fraction arms (total doses 28.5 and 30 Gy) of the FAST trial than the 50 Gy schedule [18].
Induration is a key late NTE that is expected to increase with the passage of time, irrespective of radiation schedule. Other factors contributing to breast appearance include fat necrosis and oedema, particularly in the early years [19]. Table 2 shows FAST-Forward breast assessments recorded separately by patients and clinicians. It is important to consider the absolute frequencies of events as well as the relative comparisons between schedules. For example, for breast shrinkage, the most frequent of the clinicianassessed effects, the prevalence of moderate or marked effects at 5 years was 5.5% in 40 Gy and 6.8% in 26 Gy, and the 5-year prevalence of moderate or marked induration outside the tumour bed was only 0.1% and 2.1% in 40 and 26 Gy, respectively. For all clinician-assessed events documented in the moderate/marked change categories, most were moderate rather than marked in severity.
With regards to increasing frequency of late NTE with time, stability of the hazard ratio at longer timepoints is  clinically relevant, as shown for START-B [10] and FAST [5] in Table 3. The principle of the relative difference between test and control group changing little with time can therefore be applied to FAST-Forward, again noting the low absolute levels of marked and moderate events. The Danish-led HYPO trial of 1864 patients tested the non-inferiority of 40 Gy in 15 fractions in terms of breast induration at 3 years compared with 50 Gy in 25 fractions [7]. This important study reproduced the START-B results with 3-year rates of induration of 11.8% (95% confidence interval 9.7e14.1%) in the 50 Gy group and 9.0% (95% confidence interval 7.2e11.1%) in the 40 Gy group (risk difference 22.7%; 95% confidence interval 25.6e0.2%; P ¼ 0.07). Low uptake of hypofractioned whole-breast radiotherapy in the USA, due in part to concerns of safety for patients receiving a tumour bed boost, chemotherapy or having large breast size, led to a randomised non-inferiority trial [20]. The standard treatment of 50 Gy in 25 daily fractions with a 10e14 Gy boost in five to seven fractions was tested against 42.56 Gy in 16 daily fractions with a 10e12.5 Gy boost in four to five fractions. In total, 106 (36.9%) of the 287 patients were defined as large breast size, with a bra cup size of at least D. Adverse patient-reported cosmetic outcome, the primary end point, was 5.4% lower (8.2% versus 13.6%, P ¼ 0.002 for non-inferiority) in the hypofractionated arm overall and 18.6% lower (90% upper confidence limit 8% lower) for large breasted patients. They concluded that this offers strong reassurance for hypofractionation not compromising cosmetic outcome based on large breast size.
Tsang et al. [21] looked at dose heterogeneity with regards to the FAST trial and the risk of 'triple trouble'. In total, 390 full computed tomography planning data sets were reviewed for patients where there was a baseline and a 2-year photographic assessment, the primary end point of FAST. The two five-fraction groups were combined for analysis and there was no significant difference between these and the control for breast volume or for patient tumour and treatment characteristics from the whole FAST population. Multiple logistic regression analyses showed that after adjusting for breast size (and surgical deficit), there was no evidence of late NTE associated with dose inhomogeneity using various definitions of hotspots. The effect of inhomogeneity was not significantly different for any of the dosimetric parameters between control and fivefraction schedules. In FAST-Forward, the a/b estimate for any clinician-assessed moderate or marked NTE was barely different unadjusted or when adjusted for breast size, using whole-breast planning treatment volume as the proxy for breast size. The same lack of change was found with photographic assessment and breast size. We can conclude that breast size is an established factor for increased NTE following breast radiotherapy but that hypofractionation, including five-fraction schedules, is not an additional concern for larger breasted patients.
In FAST-Forward, retrospective subgroup analyses comparing the time to first clinician-assessed moderate or marked adverse effect in the breast or chest wall for 26 Gy versus 40 Gy provided no evidence of a differential effect of the five-fraction schedule according to age, breast size, surgical deficit, tumour bed boost or adjuvant chemotherapy, as confidence intervals for subgroups overlapped, although the power for these retrospective subgroup analyses was low (Figure 2).
What about other organs at risk? The heart is often mentioned, particularly as a very long follow-up is required to assess the full risk, although there is no specific reason to expect an increased cardiac sensitivity to hypofractionation. Darby et al. [22] have shown that there is no safe dose to the heart and therefore the effort is to reduce or eliminate cardiac dose. At this early stage, after imaging and further investigation, excluding cases confirmed not to be radiotherapy-related, for left-sided radiotherapy there were six cases of ischaemic heart disease in the 40 Gy group and three cases in the 26 Gy group. The most frequent specialist referral we have seen is to lymphoedema clinics for breast lymphoedema: 90 patients (6.6%) following 40 Gy, 122 (8.9%) after 27 Gy and 106 (7.7%) after 26 Gy. Breast oedema is predominantly an early side-effect, which we have seen settling, such that at 5 years the moderate/marked incidence on clinician assessment was seven (0.7%) patients after 40 Gy, 18 (1.8%) after 27 Gy and 17 (1.7%) patients after 26 Gy, with no oedema in 94, 92 and 93%, respectively. These rates are low and not clinically or statistically significantly different.

Some Radiobiological Considerations: Tumours
In a review of the linear-quadratic model and implications for practice, Brand and Yarnold [23] present FAST-Forward as an example of a trial evaluating five-fraction hypofractionated accelerated radiotherapy. To make sense of FAST-Forward in terms of fraction size effects, the START trials offer a good entry point. The START-P/-A trials [8,10,14] (1986e2003) each compared 50 Gy in 25 fractions over 5 weeks (control) with two test dose levels of a 13-   (Table 4), yet based on 95 IBTR events in 2215 patients (4.3%), the test schedule was non-inferior to 50 Gy (hazard ratio 0.77; 95% confidence interval 0.51e1.16, P ¼ 0.21). In fact, the point estimate 10-year IBTR rate of the 3-week regimen was 1% lower than the 5-week control regimen (ns). A post-hoc analysis asked the question 'If this difference is real, what would it tell us about the impact of treatment time?' [24]. We know that in laryngeal carcinomas, at least 0.5 Gy/day can be 'wasted' compensating for accelerated repopulation from the fourth week of treatment onwards, first described by Withers et al. [25] in patients treated with primary radiotherapy and confirmed by Lyhne et al. [26] in a randomised clinical trial comparing 60 Gy in 30 fractions delivered five versus six times per week. Breast cancers have relatively low mitotic rates at presentation, but they might be in an accelerated phase of repopulation by the time radiotherapy starts several weeks or months after primary surgery and/or chemotherapy. In the context of the START-B result, if the post-hoc analysis (hypothesis generating) estimated 0.6 Gy/ day (95% confidence interval 0.1e1.8, P ¼ 0.02) 'wasted' dose in control group patients during weeks 4 and 5 is true, it implies about 14 Â 0.6 ¼ 8 Gy of the control regimen (50 Gy) is 'wasted'. This implies a time-corrected EQD 2/3.5 of 42 Gy for the 5-week regimen compared with 45 Gy for the 3week schedule. The Danish-led HYPO trial [7] offers an independent test of START-B in a comparable group of patients, in whom the 9-year risk of locoregional recurrence was 3.3% (95% confidence interval 2.0e5.0) in the 50 Gy in 25 fractions group compared with 3.0% (95% confidence interval 1.  Table 4). The difference in estimated anti-tumour effect between this EQD 2/3.7 ¼ 41 for the five-fraction schedule and EQD 2/3.7 ¼ 45 Gy of 40 Gy in 15 fractions would be too small to detect at such high levels of local control. Nevertheless, a robust clinical conclusion can be drawn, namely that the five-fraction regimen has shown non-inferiority in relation to the predefined 1.6% excess IBTR boundary set in the protocol. Questions have been raised whether 26 Gy in five fractions has any anti-tumour effect at all [27]. With a 5-year incidence of IBTR of 2.1% (95% confidence interval 1.4e3.1) after 40 Gy in 15 fractions, the incidence without any radiotherapy would be expected to be about 6% at 5 years and 10% at 10 years, according to systematic overviews of radiotherapy effects [28]. The observed 5-year incidence of IBTR after 26 Gy in five fractions is hardly consistent with an absence of effect.

Some Radiobiological Considerations: Late Reacting Normal Tissues
The discussion of tumour responses above has a lot to say about the potential impact of time on IBTR. Turning to late NTE, meticulous data generated in human skin are consistent with a minimal measurable effect of time. Turesson and Thames [29] reported a tiny time effect for telangiectasia associated with a complete absence of mitotic figures in capillary endothelium on serial skin biopsies over many weeks of radiotherapy, the lack of mitoses excluding repopulation as a mechanism. The effect was thought to probably represent a very slow component of repair decaying with a T 1/2 of around 40 days. The same post-hoc investigation of a time effect in breast cancer in START-B described above included analysing the effect of time on late NTE as a negative control, yielding an estimate of 0.14 Gy/day (95% confidence interval e0.09 to 0.34 Gy/day, P ¼ 0.29) for a change in photographic breast appearance [26].
The reason for providing this level of detail is that the selection of FAST-Forward test dose levels 27 and 26 Gy assumed, firstly, no clinically significant time effect for late NTE between 1 and 3 weeks; secondly, complete sublethal damage repair between fractions; and thirdly, an a/b of 2.8 Gy for late NTE, the last assumption based on the combined estimates of a/b in START-A and FAST. On this basis, the EQD 2/2.8 of all FAST-Forward schedules relative to 50 Gy in 25 fractions are shown in Table 5, where negative values indicate estimated NTE rates lower than 50 Gy in 25 fractions.
Although the 27 Gy test dose level was predicted to be iso-effective for NTE with 40 Gy in 15 fractions, the observed iso-effect for NTE at 5 years was closer to 26 Gy, suggesting a slightly lower a/b value (see Table 4). The a/b point estimates are all around 2 Gy, corresponding to EQD 2/ 2 of about 47 Gy for 26 Gy in five fractions. This compares to EQD 2/2.8 of about 46 Gy for 40 Gy in 15 fractions, the latter using the combined estimate of a/b ¼ 2.8 for this regimen based on START-A and FAST.
The 95% confidence intervals of a/b point estimates for all NTE scored in the FAST-Forward trial fall within the confidence intervals of a/b estimates for all late NTE in the FAST and START-P/-A trials. One interpretation, and statistically speaking the likeliest, is that they are all internally consistent with each other. Alternatively, the differences in a/b estimates is real, and late NTE are truly slightly more likely after 26 Gy in five fractions. If so, we exclude repopulation for the reasons described above, leaving slow (>24 h) repair between daily fractions reported by Turesson and Thames [29] and modelled in the FAST-Forward trial protocol (appendix 2) as the likely explanation [4]. Alternatively, lower rates of moist desquamation (high a/b) after five-fraction regimens may cause less consequential late NTE (same high a/b), enough to reduce the a/b estimate of late NTE compared with conventional fractionation. These somewhat esoteric considerations should not obscure the all-important clinical conclusion that 26 Gy in five daily fractions offers patients comparable NTE rates and non-inferior IBTR to 40 Gy in 15 fractions.

UK Consensus and Recommendations for Implementation
The UK consensus meeting [2] included an in-depth review of the FAST-Forward results, including many of the clinical aspects examined here. The results of FAST-Forward were planned to be taken together with those of IMPORT LOW [30], which had the same control regimen of 40 Gy in 15 fractions to the whole breast and, therefore, are applicable to partial-breast radiotherapy. There is no clinical rationale for excluding groups that were under-represented unless there is a logical argument for doing so. The decisions taken at the consensus meeting were to adopt 26 Gy in five daily fractions of 5.2 Gy for whole-breast, partialbreast and chest wall radiotherapy as the standard regimen.
The coronavirus pandemic has unexpectedly given clinicians and centres all over the world experience of 26 Gy in five daily fractions. Audit of that experience by centre, region or country should aid confidence in incorporating it into national or international guidelines. The original publication [4] and appendices include links to the trial protocol as a resource; the UK consensus weblink in this document is also a resource and the FAST-Forward team have provided advice both to individual centres and via webinars to international groups over the last year. The START trials' [8,9] 5-year outcomes were published in 2008 and the 40 Gy in 15 fractions schedule was adopted as UK standard of care in 2009 (https://www.nice.org.uk/guidance/ng101) as a result, albeit that the 10-year outcomes [10] gave clinicians and patients confidence that the regimen was safe and effective in the longer term. Similarly, the 26 Gy five-fraction schedule is ready for adoption globally, and indeed is already going through that process in some countries. Although, based on the START [10] and FAST [5] data, it is anticipated that outcomes will remain non-inferior at 10 years, it is important to continue collecting data to the 10year timepoint to provide reassurance around the longerterm safety and efficacy of the five daily fraction schedule.