Hypofractionated Adjuvant Whole Breast Radiotherapy: Progress and Prospects

Abstract

Published results of randomised trials involving >7000 women confirm the safety and efficacy of hypofractionated schedules of adjuvant radiotherapy for women with early breast cancer using fraction sizes between 2 and 3 Gy assuming appropriate downward adjustments to total dose. Unnecessary concerns relating to heart tolerance, suboptimal dose distribution and duration of follow up need not discourage the routine adoption of 15- or 16-fraction schedules in women treated by breast conservation surgery for early breast cancer. Regardless of fractionation regimen, dose escalation to the index quadrant in high risk subgroups will result in a greater relative increase in late adverse effects than tumour control, a therapeutic disadvantage that can only be overcome by exploiting a marked dose-volume effect. A 15-fraction schedule of whole breast radiotherapy is unlikely to represent the lower limits of hypofractionation, and the preliminary results of a 5-fraction regimen are encouraging.

Key Words::

• Radiotherapy
• Hypofractionation
• Breast cancer
• Local tumour control
• Normal tissues
• Clinical trials

It has long been assumed that human cancers are relatively insensitive to fraction size compared to late-reacting normal tissues [1,2]. However, breast cancer appears to be more sensitive to fraction size than thought in the past, perhaps approaching the sensitivity of late reacting normal tissues [2–4]. If true, small fractions (2.0 Gy) spare the cancer as much as the late-reacting normal tissues, and larger fractions have no disadvantages. A critical point is that appropriate downward adjustments to total dose must be made to compensate for non-linear increases in normal tissue effects (NTE) as fraction size increases [5,6]. The cell and molecular processes underlying the adjustments are not clear, but a mechanistic understanding is not needed to apply the linear quadratic model in a given clinical situation. For fraction sizes in the range 2.0 – 6.0 Gy, the linear quadratic model appears to offer a reliable estimate of dose reductions needed to match a conventionally fractionated regimen in terms of specific late NTE. Reliability hinges on appropriate choices of α/β values, assuming complete repair between fractions and no time dependency for late NTE. The clinical benefits depend on whether and how local tumour control is affected by changes in fractionation.

Summary of randomised trials

The international standard regimen for whole breast radiotherapy delivers a total dose of 50 Gy in 25 fractions (daily doses) over 5 weeks following surgical resection of primary tumour in women with early breast cancer. Attempts in the UK to reduce the number of fractions in the 1970s made inadequate downward adjustments to total dose, resulting in unacceptable rates of late complications [7]. The Danish experience of hypofractionated post-mastectomy radiotherapy is also informative in this respect [8]. The maximum physical absorbed dose delivered by the anterior photon field to the axilla in the 12-fraction group ranged from 43 to 52 Gy (median 47.3 Gy) and in the electron field to the chest wall from 44 to 50 Gy (median 47.9 Gy). Assuming an α/β value of 3.0 Gy for late adverse effects, the median total doses of 48.0 Gy in 12 fractions are equivalent to 67 Gy in 2.0 Gy equivalents. A high rate of marked late adverse effects was, in retrospect, only to be expected. A 12-fraction regimen equivalent to 50 Gy in 25 fractions in terms of late adverse effects, assuming an α/β value of 3.0 Gy, would have delivered 39.6 Gy in 12 fractions of 3.3 Gy.

Miscalculations based on the Nominal Standard Dose formula of iso-effective dose schedules for late adverse effects inhibited further research in breast radiotherapy fractionation for decades. Interest in fewer larger fractions delivered over a shorter overall treatment time has been rekindled by randomised clinical trials based on a better understanding of normal tissue and tumour responses. Four randomised trials involving a total of >7000 women have compared a lower total dose in fewer larger fractions against 50 Gy in 25 fractions, and all have reported favourable results in terms of local tumour control and late adverse effects, see Tables I and II [9–13].

Table I. Randomised clinical trials testing fraction size in whole breast external beam radiotherapy after local excision of early breast cancer. All trials used a standard arm delivering 50 Gy in 25 fractions over 5 weeks.

Trial	Test schedules*	Number patients	Breast conserving surgery (%)	Patients prescribed boost (%)	Median follow up (months)
START pilot 1986–98 [10,11]	39.0/13/5.0 (3.0)	1410	100	74.5	116
	42.9/13/5.0 (3.3)
Ontario 1993–96 [9,15]	42.5/16/3.2 (2.66)	1234	100	0	>132
START A 1999–02 [12]	39.0/13/5.0 (3.0)	2236	85	60.6^1	61
	41.6/13/5.0 (3.2)
START B 1999–01 [13]	40.0/15/3.0 (2.67)	2215	92	42.6^1	72

*Total dose (Gy), fraction number, treatment time (weeks) and fraction size (in brackets, Gy).

¹Breast conservation patients only.

Table II. Results of randomised clinical trials testing fraction size in whole breast external beam radiotherapy after local excision of early breast cancer. All trials used a standard arm delivering 50 Gy in 25 fractions over 5 weeks.

		5-year rates of:
Trial	Dose schedules*	Any change in breast appearance (%)	Good/excellent breast cosmesis (%)	Local tumour relapse (%)
START pilot 1986–98 [10,11]	50.0/25/5.0 (2.0)	35.4	–	12.1
	39.0/13/5.0 (3.0)	27.4	–	14.8
	42.9/13/5.0 (3.3)	42.3	–	9.6
Ontario 1993–96 [9,15]	50.0/25/5.0 (2.0)	–	79.2^1	3.2^2
	42.5/16/3.2 (2.66)	–	77.9^1	2.8^2
START A 1999–02 [12]	50.0/25/5.0 (2.0)	42.9	–	3.2
	39.0/13/5.0 (3.0)	32.1	–	4.6
	41.6/13/5.0 (3.2)	43.6	–	3.2
START B 1999–01 [13]	50.0/25/5.0 (2.0)	42.2	–	3.3
	40.0/15/3.0 (2.67)	36.5	–	2.0

*Total dose (Gy), fraction number, treatment time (weeks) and fraction size (in brackets, Gy).

¹71.3% and 69.8% at 10 years.

²6.7% and 6.2% at 10 years.

The Royal Marsden Hospital/Gloucestershire Oncology Centre and Ontario trials entering a total of 2644 women with mainly axillary node negative tumours <5 cm diameter were the subject of a 2008 Cochrane review of altered radiotherapy fractionation in early breast cancer [14]. Radiotherapy fractions larger than 2.0 Gy did not appear to affect: a) local-recurrence free survival (absolute difference 0.4%, 95% CI -1.5% to 2.4%), b) breast appearance (risk ratio (RR) 1.01, 95% CI 0.88 to 1.17; p = 0.86), c) survival at five years (RR 0.97, 95% CI 0.78 to 1.19; p = 0.75), d) late skin toxicity at five years (RR 0.99, 95% CI 0.44 to 2.22; p = 0.98, or e) late radiation toxicity in subcutaneous tissue (RR 1.0, 95% CI 0.78 to 1.28; p = 0.99). The review concluded that the use of unconventional fractionation regimens did not affect breast appearance or toxicity, nor appear to affect local cancer relapse.

The UK Royal Marsden Hospital/Gloucestershire Oncology Centre (START pilot) and START A trials both tested two dose levels of a 13-fraction regimen in terms of late NTE and tumour control, allowing direct estimates of α/β for both endpoints [10–12]. Based on a combined total of 278 local-regional tumour relapses, the adjusted α/β value for tumour control was estimated to be 4.6 Gy (95% CI, 1.1 – 8.1), similar to the estimate of 3.4 Gy (95% CI, 2.3 – 4.5) for late change in photographic breast appearance. The two trials identified 13-fractions of 3.2 Gy over 5 weeks to be as safe and effective as 50.0 Gy in 25 fractions. The results of the Ontario and START B trials were consistent with the findings of the START pilot and START A trials. The Ontario study compared 42.5 Gy in 16 fractions of 2.66 Gy (3.2 weeks) with 50.0 Gy in 25 fractions over 5 weeks [9, 15]. Rates of breast cosmesis at a median follow up of >11 years were virtually identical in both treatment arms, consistent with an α/β value of 3.0 Gy, assuming no effect of treatment time on late normal tissue responses. If the 2-week difference in treatment duration had no impact on tumour control either, the identical 10-year local control rates are consistent with the same α/β estimate of 3.0 Gy. The UK START B trial compared 40.0 Gy in 15 fractions of 2.67 Gy (3 weeks) against 50.0 Gy in 25 fractions (5 weeks), and recorded a slightly lower rate of change in breast appearance after the 15-fraction regimen (HR = 0.83, 95% CI, 0.66 – 1.04, p = 0.06) [12]. The favourable hazard ratio for late changes in breast appearance is likely to be real, since 40.0 Gy in 15 fractions is equivalent to 45–45.5 Gy in 2.0 Gy fractions, if α/β is in the range 3.0–3.4 Gy. In other words, 40.0 Gy in 15 fractions is gentler on the late reacting normal tissues than 50.0 Gy in 25 fractions. The important question is whether 40.0 Gy in 15 fractions is also gentler on breast cancer. If the α/β for tumour control is ≥10 Gy, tumour control would be inferior after such a large reduction in total dose (from 50.0 Gy to 40.0 Gy) unless compensated by a big effect of shortened overall time from 5 to 3 weeks. The hazard ratio for tumour control for the 15-fraction regimen was 0.79 (95% CI, 0.48 – 1.29), suggesting a comparable rate of local relapse after 40.0 Gy in 15 fractions and 50.0 Gy in 25 fractions. The residual imprecision indicated by the upper and lower 95% confidence intervals for the absolute difference between regimens suggest that local tumour relapse is unlikely to be more than 1% higher, and conceivably 1 or 2% lower, than after 50.0 Gy in 25 fractions.

On the basis of all 4 trials, there appears to be no reason to avoid modest hypofractionation in the adjuvant treatment of women requiring whole breast or post-mastectomy chest wall radiotherapy, but there are some residual concerns expressed in the literature that need to be addressed [16]. Concerns that trial follow up is too short are unjustified, in that the START pilot trial published at a median follow up of 9.7 years (inter-quartile range 7.8 – 11.8) [10,11]. The Ontario trial has published 10-year results consistent with an earlier 5-year analysis [15]. The START A and B trials published at a median follow up of only 5 and 6 years, respectively, but although absolute rates of late NTE and local tumour relapse are expected to increase for many years, the relationships between schedules are not expected to alter significantly, as shown in the START pilot trial [10,11]. This expectation is consistent with 15-year follow up of the EORTC tumour bed boost dose trial, in which late NTE and local tumour relapses continued to occur after 5 years without altering the hazard ratios for either adverse effects or tumour control [17].

Concerns about the sensitivity of the heart to hypofractionation have been expressed but the heart is vulnerable to radiotherapy whatever fractionation schedule is used, and there appears to be no lower dose threshold for injury [18]. The priority is therefore to protect the heart as much as possible whatever dose regimen is used. The heart apex can often be shielded using a multileaf collimator after breast conservation surgery, although after mastectomy, some form of respiratory gating may be needed. Where the brachial plexus is concerned, 40.0 Gy in 15 fractions is expected to be a safe schedule if delivered using a technique and reference point of proven safety with 50.0 Gy in 25 fractions. Assuming an α/β value of 2.0 Gy for brachial plexus injury, 40.0 Gy in 15 fractions is equivalent to 47.0 Gy in 2.0 Gy equivalents. Only if α/β for nerve damage is as low as 1.0 Gy, is 40.0 Gy in 15 fractions equivalent to 50.0 Gy in 2.0 Gy equivalents. In other words, the 15-fraction regimen is likely to be less damaging to the brachial plexus than 50.0 Gy in 25 fractions. In the START Trials 11% of patients received regional radiotherapy, with only one confirmed case of brachial plexopathy in a patient treated with 41.6 Gy in 13 fractions.

Exploiting a volume effect for dose escalation

The START pilot and START A trials confirmed a steep dose response for radiotherapy adverse effects, generating a γ50 value of 1.4 for late change in photographic breast appearance [12]. The γ50 describes the slope of the sigmoid dose response curve at the 50% level of effect, and a value of 1.4 means that the absolute rate of late NTE rises by 1.4% for each 1% increase in total dose e.g. by 2.8% per 2.0 Gy fraction delivered above 50 Gy. This contrasts with the shallow dose response for tumour control near the top of the sigmoid dose response curve in the adjuvant setting. The γ95 in the START pilot and START A trials was 0.2, corresponding to a 0.4% increase in local control per 2.0 Gy fraction delivered at this level (95%) of effect. The implication is that in women at low (5%) 10-year risk of local relapse after the equivalent of 50.0 Gy in 25 fractions, a dose reduction to the equivalent of 46.0 Gy in 23 fractions would increase local relapse from 5% to 6.5% in exchange for a 16% reduction in the rate of late NTE. Depending on how patients balance the beneficial and adverse effects of radiotherapy, 50.0 Gy in 25 fractions may be a higher dose intensity than necessary for low risk subgroups in the current era of falling local relapse rates [19]. For other populations, especially young women at higher than average risk of local relapse, the same considerations mean that dose escalation causes a much steeper increase in late NTE than in tumour control, an unfavourable relationship.

The therapeutic ratio of dose escalation at relatively high levels (>90%) of local tumour control is improved by taking advantage of the volume effect, which describes the effect of treatment volume on the probability of an adverse effect following a given dose of radiotherapy. The magnitude of the volume response can be generated in different ways, most directly by comparing outcome of the same dose prescription delivered to different partial volumes of breast tissue. In a retrospective study of RT boost dose delivered using radioactive implants in 404 patients, a 4-fold increase in risk of breast fibrosis (hardness) was reported for each 100 cm³ increment in boost volume, suggesting a very steep volume response [20]. This finding was consistent with univariate analysis of 364 patients randomised to a RT tumour bed boost dose after whole breast radiotherapy reporting a hazard ratio for poor cosmesis of 0.45 (95% CI, 0.29 – 0.76) for boost volumes ≤200 cm³ compared to >200 cm³ [21]. The volume response can also be quantified by comparing the increased risk of late adverse effects after a boost dose to the tumour bed with the same dose increment applied to the whole breast. This analysis was performed in the START pilot trial for 723 patients randomised after 5 weeks of whole breast radiotherapy to no tumour bed RT versus tumour bed RT (15.5 Gy to 100% isodose in 7 fractions) [10]. Patients randomised to tumour bed boost RT had a 17% higher risk of moderate or marked breast hardness at 10 years. The same trial compared two randomised dose levels of whole breast radiotherapy in a factorial 2 × 2 design. The dose of whole breast radiotherapy causing a 17% increased risk of breast hardening could be estimated by interpolation to be 4.5 Gy. This is much lower than the 15.5 Gy dose causing the same level of hardening when confined to the tumour bed (10–30% of whole breast volume).

What are the limits of whole breast hypofractionation?

There is no reason to assume that a 15- or 16-fraction regimen represents the lower limit of hypofractionation for whole breast radiotherapy. There is a history of once-weekly fractions delivered to women too frail or otherwise unable to attend for conventional schedules. In a French series of 115 patients undergoing primary radiotherapy for non-metastatic breast cancer, the whole breast was treated with 2 tangential fields and 5 once-weekly fractions of 6.5 Gy [22]. 101 were given additional tumour bed boost doses, 7 with 1 fraction, 69 with 2 fractions and 25 with once-weekly fractions of 6.5 Gy using electrons. Late effects were classified as grade 1 in 19 cases, grade 2 in 21 cases and grade 3 in 6 patients. The 5-year local progression-free rate was 78% (95% CI, 66.6–88.4). In a separate French series, 5 once-weekly fractions of 6.5 Gy to the whole breast, with no boost, were given to 50 women after local tumour excision [23]. Grade 1–2 induration was reported in 33% of the patients at a median follow up of 93 months (range 9–140). The 7-year local relapse free survival was 91%. Five fractions of 6.5 Gy are equivalent to 62.0 Gy in 31 fractions, assuming an α/β value of 3.0 Gy, so the modest rate of late NTE are reassuring. In the UK, 5 once-weekly fractions of 6.0 Gy and 5.7 Gy were tested in a 1:1:1 ratio against 50.0 Gy in 25 fractions (N = 915) [24]. At a median follow up of 28.3 months (inter-quartile range 24.1–36.3), the lower dose level of the 5-fraction regimen (5 fractions of 5.7 Gy) most closely matched the control regimen in terms of change in breast appearance and clinical assessments of breast hardness. The upper dose level (5 fractions of 6.0 Gy) is estimated to be equivalent to 54 Gy in 27 fractions in terms of late NTE. The primary endpoint in the UK trial related to late adverse effects, and a phase III trial is needed to test tumour control. One trial (FAST Forward) is planned in the UK that will test a 5-fraction schedule delivered in 5 days against the 15-fraction regimen that has become the UK standard of care.

Conclusions

Modest hypofractionation, using fraction sizes between 2.0 and 3.0 Gy, is a safe and effective way to deliver adjuvant whole breast radiotherapy, assuming appropriate adjustments are made to total dose. A schedule delivering 40.0 Gy in 15 fractions of 2.67 Gy in 3 weeks is now standard of care in the UK after breast conservation surgery or mastectomy. The schedule is gentler on the healthy tissues, including brachial plexus, than 50.0 Gy in 25 fractions, and there is no evidence of inferiority in terms of tumour control. Testing the risks and benefits of fraction sizes >5.0 Gy requires further evaluation in randomised trials with sufficient follow up.

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.