Patients at a single institution were eligible if they were enrolled with the Proton Collaborative Group (PCG; REG001-09, NCT01255748, informed consent obtained), which is a prospective multi-institutional observational study for patients treated at proton therapy centers within the United States. Institutional Review Board (IRB 19-000891) was also obtained.

Patients were included in this study if they were treated with curative intent for breast cancer using either photon or proton therapy between March 2016 and June 2019. Patients with metastatic disease were eligible if the treating physician determined the radiation therapy to be curative. All curative dosing fractionations were allowed. Patients undergoing whole breast, chest wall, or accelerated partial breast irradiation (APBI) were included for analysis. Technique to address lymph nodes at the time of surgery was not required for this study. Patients could have received systemic therapy as part of their treatment regimen at any time point.

Patients must have completed at least one baseline and one post-radiation survey which addressed PRO-CTCAE; otherwise, they were excluded from this analysis. In our department, patient surveys are routinely administered before, during, and after completion of radiotherapy (baseline, end of treatment, 3 months, 6 months, 12 months, and annually thereafter). We utilize the Patient-Reported Outcomes Measurement Information System 10 (PROMIS-10), which has been validated across multiple medical specialties (11, 12). Additional surveys administered include the Expanded Prostate Cancer Index breast-specific questions and an institution-specific PRO survey. Patients are not required to complete every survey provided to them.

Radiation technique (proton or photon), dose fractionation, lumpectomy boost, lymph node coverage, and target volumes were all determined by the treating physician and treatment planning team. Bolus was not used for any patients. Organ-at-risk (OAR) dose constraints generally followed the institutional guidelines. Hypofractionation (HF) was defined as a dose per fraction >2 Gy. Proton therapy was delivered by a Hitachi pencil beam scanning treatment deliver system.

Patient, tumor, and treatment characteristics were collected from the PCG database. When necessary, additional data were obtained from electronic medical record system. As described above, three surveys related to patient-reported outcomes and quality-of-life measures are routinely administered to patients in our department. From these surveys, 14 symptom-related questions were selected to be included for evaluation as they reflected PRO-CTCAE items. Symptoms included anxiety, appetite, concentration, constipation, cough, insomnia, interference with breathing, nausea, sadness, severity of breathing interference, skin toxicity, stools (diarrhea), swallowing, and tingling. Each item is scored on a five-point Likert scale with regard to frequency, severity, and interference. Physician reported toxicities were also gathered using CTCAE version 4.0. Toxicities were considered acute or late whether they occurred before or after 3 months following adjuvant radiotherapy, respectively.

Descriptive statistics were used to report patient, tumor, treatment, and PRO-CTCAE characteristics. For univariate and multivariate analysis, categorical variables were analyzed by Fisher’s exact test and continuous variables by Wilcoxon rank sum test. PRO-CTCAE items graded ≥2 and ≥3 were specifically analyzed between patients who received HF versus standard fraction therapy by Chi-square test.

Three hundred thirty-one patients met inclusion criteria, 246 of which underwent HF and 85 underwent (CF). Baseline patient and tumor characteristics between patients treated with HF vs. CF are listed in Table 1 . The median age at radiation therapy (RT) start was 60.0 years (62 HF vs. 54 CF). Most patients were Caucasian (312, 95%) with an Eastern Cooperative Oncology Group performance status (ECOG) 0 (219, 66%). Left-sided breast cancer (168, 51%) was slightly more common than right-sided breast cancer. Histology was most commonly invasive ductal carcinoma (283, 85%) and grades 1–2 (219, 66%). Estrogen receptor positivity was noted in 271 (82%) tumors, whereas human epidermal growth factor receptor 2 (HER-2) positivity was found in 74 (23%) patients. Pathologic tumor stage was T1–T2 in 309 (94%) patients. Eighty-seven (29%) patients were node-positive and only four (1%) patients had a metastasis at the time of initial diagnosis. Metastatic sites of disease in the four patients include liver (two), posterior mediastinal lymph node (one), and bone (one). All patients with metastatic disease received neoadjuvant chemotherapy and/or HER-2–directed therapy followed by modified radical mastectomy, followed by chest wall and regional nodal irradiation. These patients were then treated with adjuvant therapy such as endocrine therapy and or HER-2–directed therapy such as pertuzumab and trastuzumab. Of the two patients with liver metastases, one underwent partial hepatic resection. The other patient developed a clinical and radiographic response following neoadjuvant chemotherapy and did not undergo hepatic resection. The patient with bone metastases received palliative radiation 30 Gy in 10 fractions to multiple sites of bony metastasis. The patient initially presenting with a posterior mediastinal lymph node developed an excellent response to neoadjuvant chemotherapy and continued with HER-2–directed therapy after adjuvant radiotherapy.

Table 1 also highlights the treatment characteristics in this patient cohort. Breast-conserving surgery, which consisted of lumpectomy, occurred in 202 (82%) and 47 (55%) who underwent adjuvant hypofractionated radiotherapy and conventionally fractionated radiotherapy, respectively. Mastectomy was performed in 53 (16%) patients, and lumpectomy was performed in 278 (84%). CF was more likely to be performed following mastectomy, with 17 (7%) HF vs. 31 (37%) CF (p < 0.001). A re-excision of a positive margin was performed in 34 (10%). Reconstruction occurred in 6 (2%) HF and 20 (24%) CF (p < 0.001). Sentinel lymph node biopsy (SLNB) occurred in 223 patients (67%) total, which consisted of 181 (55%) HF and 42 (12%) CF. CF was also more likely to be performed after axillary lymph node dissection with 17 (7%) undergoing HF and 54 (64%) undergoing CF (p < 0.001). A total of 37 (11%) patients did not undergo SLNB or ALND. Chemotherapy was administered—either in the neoadjuvant or adjuvant setting—to 129 (39%) patients: 71 (21.4%) HF and 58 (18%) CF.

Photons were utilized in 281 (85%) and protons were utilized in 50 (15%) patients. Photons were delivered in 281 (85%), and protons were delivered in 50 patients (15%). Protons were more likely to be delivered in patients receiving CF (26/85; 31%) vs. HF (24/246; 10%). Target coverage was to the whole breast in 254 (77%) patients, chest wall in 71 (22%) patients, and partial breast in five (1%) patients. Whole breast radiation was more likely to be HF (225/246; 92%) than CF (29/85; 35). Patients undergoing radiation following mastectomy were more likely to receive CF (54/85; 64.3%) compared with HF (17/246; 91.5%).

Forty-eight (36%) patients receiving lymph node coverage, with patients receiving CF more likely to have axillary lymph nodes included in the target volume than HF (34.1% vs. 7.7%, p < 0.001). The median dose/fractionation was 40.05 Gy/15 fractions for the HF group and 50 Gy/25 fractions for the CF group. The median boost dose/fractionation was 10 Gy/5 fractions for both groups. Forty-eight (36%) patients had lymph node coverage included in radiation treatment: 19(8%) HF and 29 CF (34%).

Two hundred forty-seven patients (75%) experienced any PRO-CTCAE grade ≥2 ( Table 2 ), and 92 (28%) patients experienced any PRO-CTCAE grade ≥3 toxicity ( Table 3 ). Table 4 highlights the individual grade ≥2 PRO-CTCAE toxicities. Grade 2+ PROs were experienced by 176/246 (72%) HF and 71/85 (84%), with CF correlating with higher rates of grade 2+ PRO-CTCAE (p < 0.03). When evaluating modality, there was a statistically significant difference (p = 0.021) in grade ≥2 PRO-CTCAE in patients treated with protons (44/50; 88%) vs. photons (203/281; 72%). However, there was no difference in PRO-CTCAE grade ≥2 skin toxicity between protons and photons (14.2% vs. 11.5%). In addition, there was no difference in grade >3 PRO-CTCAE between protons (21/50; 42%) and photons (85/281; 30%) (p = 0.10). Skin toxicity (204, 76%) and insomnia (133, 40%) were the two most common PRO-CTCAEs graded ≥2. CF also correlated with higher rates of grade 3+ PRO, with 65/246 (26%) HF and 41/85 (48%) experiencing grade 3+ PRO (p < 0.001). The most common PRO-CTCAE ≥3 was skin toxicity (87, 33%) and concentration (92, 28%). When analyzing individual items by CF versus HF, CF was associated with increased risk of grade ≥2 skin toxicity (p = 0.04) and swallowing (p < 0.01), whereas HF was associated with an increased risk of grade ≥2 insomnia (p = 0.04) ( Table 2 ). CF was also found to be associated with increased risk of grade ≥3 skin toxicity, swallowing, and nausea (all p < 0.01). When evaluating items most likely related to radiation (anxiety, cough, severity of breathing, interference with breathing, nausea, skin burns, and swallowing), CF was significantly associated with increased risk of PRO-CTCAE grade ≥3 (OR 2.96, p < 0.01).

Univariate and multivariate analyses of clinical and treatment variables for any PRO-CTCAE ≥2 and ≥3 were performed ( Table 5 ). The following variables were significantly associated with an increase in any PRO-CTCAE ≥2 after univariate analysis: ECOG ≥1, proton therapy (vs. photon), and chest wall radiation (vs. whole breast) ( Table 3 ). HF was associated with decrease in any PRO-CTCAE ≥2. However, only ECOG ≥1 remained significant after multivariate analysis (OR 2.21, p = 0.01). When analyzing any PRO-CTCAE ≥3, ECOG ≥1, node positivity, and total dose were all univariately significant for increased risk of analyzing any PRO-CTCAE ≥3. HF was protective against PRO-CTCAE ≥3 ( Table 3 ). In the multivariate model, ECOG ≥1 (OR 1.78, p = 0.02) and HF (OR 0.48, p < 0.01) remained significant.

Per physician reporting, a grade ≥2 adverse event was found in 114 patients: 52 (21%) HF and 62/85 (73%) CF. Radiation dermatitis was the most common adverse event reported with grade 2+ RT dermatitis occurring in 37 (72.5%) HF and 57 (70.4%) CF. Grade 3 radiation dermatitis occurred in four patients receiving CF, whereas there was none in the HF group.