Effect of hypofractionation on the incidental axilla dose during tangential field radiotherapy in breast cancer

The Z0011 trial [1] demonstrated that use of sentinel lymph node biopsy (SLNB) alone compared with axillary lymph node dissection (ALND) does not result in inferior survival in patients with one or two positive sentinel lymph nodes. Interestingly, the regional recurrence rate after SLNB alone was very low (<1%) even though approximately 27% of patients had lymph node metastases in the undissected axillary nodes. Since tangential field irradiation delivers a relevant dose to the axillary levels, adjuvant radiotherapy contributes to eradication of microscopic disease in the undissected axilla and accounts for the good oncologic outcomes after SLNB only. In accordance with this, current guidelines recommend omitting ALND only if patients receive systemic therapy and adjuvant radiotherapy (RT) [1, 2]. The MA-20 [3] and EORTC 22922-10925 [4] studies have shown that elective lymph node irradiation (supra/infraclavicular and internal mammary) is effective in lowering regional and distant metastases and prolongs disease-free survival, which emphasizes the potential of lymph node irradiation in breast cancer.

Several previous studies focused on the dose distribution in the axillary levels during tangential field irradiation [5‐8].The dose distribution varies largely depending on patient anatomy and treatment technique (e.g., high tangent vs. conventional tangent). Usually, a smaller dose per fraction compared to the prescribed dose is delivered to the axillary lymph node areas [9]. This has a direct impact on the biological effective dose (BED) and tumor control probability (TCP) of microscopic disease in the axillary lymph nodes [9‐11].

Nowadays, moderate hypofractionated radiotherapy (40–42.5 Gy in 15–16 fractions [Fx]) is the recommended therapy for adjuvant radiotherapy after breast-conserving surgery for most patients [1, 12‐14]. Several trials have been initiated to test the outcome of even higher doses per day (extreme hypofractionation). The UK FAST trial [14] tested 30 or 28.5 Gy in 5 Fx once a week and showed promising results with only two local relapses in 915 patients at 3 year median follow up. The UK FAST-Forward trial (N = 4000) [15] recruited patients between 2012 and 2014 and tested 27 or 26 Gy in 5 Fx/week against 40 Gy in 15 Fx. Since the study remains in follow-up, results are pending.

Nonetheless, two different approaches have changed simultaneously in breast cancer treatment: hypofractionation is considered a standard therapy and at the same time, de-escalation of axillar surgery is conducted for an increasing number of patients. The effect of fractionation on the incidental radiotherapy dose to the axilla during whole-breast radiotherapy has not been addressed in the literature so far. Thus, it remains unclear whether hypofractionated radiotherapy might impair the local outcome in the axilla.

The aim of the current study was to evaluate if different fractionation schedules have a potential effect on TCP during incidental lymph node irradiation in the areas at risk.

In a previous study, 580 F18-FDG-PET/CT-positive lymph node metastases of breast cancer patients were detected and mapped in a CT template [16]. 431 of these 580 lymph nodes were located in axillary levels I–III. Using the previously published methodology, these 431 lymph node metastases were registered rigidly and non-rigidly to the three different patients listed in Table 1. The patients were chosen to represent different breast sizes (ranging from small to large) and breast shape. The dose in every lymph node was assessed in each patient for the equilateral tangent treatment plan and the different prescription doses. The dosimetric data was then transferred to SPSS (IBM statistics version 25, IBM Corp., Armonk, NY, USA).

Current guidelines recommend the omission of ALND in case of 1) negative SLNB, 2) micrometastases seen in SLN, or 3) 1–2 positive lymph nodes and T1–2 tumors after BCS without preoperative chemotherapy and planned whole-breast RT [5]. According to these recommendations, residual tumor cells in the axillary lymph nodes must be expected in a relevant number of patients prior to whole-breast irradiation. Large randomized trials (START Pilot, START A/B, Ontario trial) suggest that hypofractionated radiotherapy can be safely used in most breast cancer patients even if the total dose is reduced by approximately 10%. In 2018 an updated American Society for Radiation Oncology (ASTRO) evidence-based guideline was published, recommending hypofractionation as the standard scheme during whole breast irradiation [1]. However, the effect of incidental irradiation in the axilla remains unclear in these studies. The START trials [20] reported only relapses in the ipsilateral axilla if it had been within the irradiated target volumes. Regional relapses outside the radiotherapy target volume were excluded from the analysis of locoregional relapse. In the Ontario trial [21], all patients were treated with axillary dissection and were staged as pN0. Furthermore, axillary relapses were not reported separately and can therefore not be compared between the two schedules. Furthermore, in the Z0011 trial, which is one of the foundations of today’s guidelines, high tangents were used in approximately 50% of the patients and 17% received supraclavicular fields [17]. Hence, the outcome of patients treated with SLNB only receiving hypofractionated radiotherapy to the breast remains unclear.

The TCPs in our study were calculated according to a model by Plataniotis and Dale [10]. The authors defined TCP as (failure rate without RT − failure rate with RT)/failure rate without RT. Thus, the calculated values relate only to the beneficial effect of radiotherapy on local control. When interpreting our data, it needs to be taken into consideration that TCPs most likely do not correctly reflect the actual expected tumor control in axillary lymph nodes. Firstly, because the model was calculated for local breast cancer recurrences not for axillary recurrences. A pre-RT average clonogen number per tumor of 59.2 was assumed, which does not necessarily reflect the clonogen in the axillary lymph nodes in case of residual tumor. Secondly, the datapoints used for the model are all clustered around relatively high BEDs, causing considerable uncertainties for low BEDs. In addition, Plataniotis and Dale et al. used an alpha/beta ratio of 4 for calculation of BEDs. However, the alpha/beta ratio of breast cancer cells is still controversial and recent analyzes proposed values clearly lower than 4 [22]. The alpha/beta ratio has a direct impact on the calculated BEDs as well as the TCP formula. Despite these limitations, the publication currently provides the best model linking BED and TCP for microscopic breast cancer cells taking both hypofractionated and normofractioned schemes into account. Even though the model might not predict the exact control probability in the lymph nodes, it clearly helps to understand the effect of hypofractionation during incidental irradiation.

In accordance with other studies, the model by Plataniotis and Dale suggest an s‑shape of the TCP curve, with increasing slope from low to moderate BEDs and increasing slope from moderate to high BEDs [10, 20]. Thus, for high BEDs (e.g., in the boost target volume), changes of BED result in minor changes of TCP, whereas changes of lower BEDs during incidental irradiation result in large differences of TCP due to the steepness of the slope (Fig. 1). Our study focuses on the effect of hypofractionation on local control, even though the rate of axillary recurrences is very low in breast cancer patients. Nevertheless, taking the results from the MA-20 [3] and EORTC studies [4] into account, treatment of microscopic disease in the lymph node system has a potential effect on distant metastases and disease-free survival. Thus, hypothetically, the TCP in the axillary lymph node regions might not only affect locoregional recurrences. However, patients without an indication for regional lymph node irradiation usually have a lower baseline risk of lymphatic tumor spread. Thus, it is unclear whether the (theoretical) reduction of TCP has a (measurable) clinical impact in early breast cancer patients and hypofractionation remains the standard regimen for most patients.

As tangential field irradiation is the standard technique for breast radiotherapy in many departments [21, 23], we analyzed the dose distribution to the axillary lymph nodes during tangential field irradiation. Several studies report average doses in the axillary lymph node levels that range widely in dependence of patient anatomy [6, 9, 11]. To account for differences related to anatomy and thus field design, we choose three different patients with very different breast sizes and shapes. Hence the degree of incidental lymph node irradiation varied between the patients from 12.1 ± 19.2% of the prescribed dose (patient III) to 64.8 ± 17.8% of the prescribed dose (patient II). Differently from our study, most previous studies estimated the dose distribution in the lymph node areas as defined by RTOG or ESTRO guidelines [5, 6, 9, 11, 24, 25]. However, these guidelines define a treatment volume and do not necessary reflect the whole axillary lymph node drainage system and both primary lymph node metastases and lymph node recurrences occur outside the ESTRO and RTOG margins [16, 26]. The 2018 published 3D lymph node atlas [16] allows dose evaluation and calculation of the TCP for actual “areas at risk” in which lymph node metastases frequently occur. Still, it should be considered that detected areas at risk in the lymph node atlas (including also recurrent and metastatic breast cancer patients) might not be representative for primary early breast cancer patients receiving SLNB only.

The total treatment time between conventional fractionation and the analyzed hypofractionated schemes differed by up to 28 days. Based on the START trials, previous studies generated the hypothesis that overall treatment time is a significant determinant of local cancer control. The assumption was prompted by a suggestion of lower locoregional relapse rates after 40 Gy in 15 fractions in in START‑B trial. Based on the START trials, Haviland et al. estimated an overall treatment time effect for locoregional relapse of 0.6 Gy/day [27]. However, the presumed time effect conflicts with the results of the Ontario trial testing 42.5 Gy in 16 fractions without any differences in long-term outcome. Furthermore, the literature regarding the impact of a treatment delay are inconsistent and the repopulation kinetics of microscopic breast cancer are known to be slow [28]. In the model by Plataniotis and Dale [10], repopulation during treatment was assumed to be small and was not taken into account. Even though the subject remains controversial, neglecting the potential impact of overall treatment time in the TCP formula (and therefore in our calculations) can be seen as a relevant limitation of our study. Given the limitations of the TCP calculation, the level of significance was set at <0.001 to keep statistical uncertainties as low as possible.

For a better accuracy of the model, we analyzed only schemes with 5 Fx per week. In addition to the recommended (standard) fractionation of 42.67 or 40.05 Gy, we analyzed the FAST-Forward schemes. These schedules are yet not recommended, but currently being tested in prospective trials [15]. Since the UK-FAST trial showed promising results after 3 years of follow-up, extreme hypofractionation with higher single doses could become more important. Thus, the potential effect of hypofractionation in the axillary lymph nodes should be considered for further prospective studies as well as during interpretation of results.

According to the available radiobiological models, hypofractionated RT leads to lower TCPs in areas of risk containing microscopic disease in the axilla compared to normofractioned RT. Due to low rates of axillary recurrences in early breast cancer, it is unclear whether this finding has a clinical significance. Nevertheless, special attention should be paid to this issue in further studies on this subject.