Nasopharyngeal carcinomas (NPCs), most of which are undifferentiated (UCNT), present a unique oncological and therapeutic challenge, particularly an intermediate-incidence area like Tunisia, where the incidence remains ranging from 3 to 9 new cases per 100,000 habitants annually. ^{1, 2} Radiotherapy, in combination with concomitant chemotherapy, remains the cornerstone of treatment. However, the anatomical complexity of the nasopharyngeal region poses a dual challenge: ensuring optimal tumor coverage while meticulously sparing adjacent critical structures. ³ Over the past two decades, remarkable progress in imaging modalities—especially with MRI, and more recently PET-CT—has transformed the landscape of radiotherapy planning. These technologies, coupled with advanced tools for image fusion and multiplanar reconstruction, have greatly refined the precision of target volume delineation. ^{4, 5} In parallel, the emergence of advanced irradiation techniques such as intensity-modulated conformal radiotherapy (IMRT), has allowed clinicians to deliver highly conformal dose distributions that optimize tumor coverage while respecting dose constraints for organs at risk (OARs). ^{6, 7} These dose constraints have been established over time by expert radiation oncology societies, based mainly on retrospective dose–effect correlation studies and meta-analyses. ^{8, 9} For the brachial plexus, dose constraints remain controversial. Indeed, the complex anatomy of the brachial plexus and its variable relationship with adjacent structures lead to significant inter- and intra-observer variability in contouring, which remains challenging. Moreover, variability in fractionation schedules, dose per fraction, overall treatment time, and total dose across institutions adds further complexity to retrospective analyses, partly explaining the variation in the recommended maximum dose to the brachial plexus, which typically ranges between 60 and 66 Gy according to international guidelines. ^{10– 12} Understanding and addressing these variations is crucial, not only to improve therapeutic outcomes but also to minimize the risk of debilitating complications such as radiation-induced brachial plexopathy.

This study aimed to assess the brachial plexus dosimetry in 50 patients treated with IMRT for nasopharyngeal carcinoma at the Salah Azaiz Institute. We investigated correlations between dosimetric parameters and both acute and late neurological toxicity affecting the brachial plexus. Secondly, we aimed to evaluate the conformity of brachial plexus delineation with international contouring atlas guidelines.

The average age at diagnosis for the included patients was 44 years (10-68). Three patients were under 18 years of age. The sex ratio in our study was 2.1 (34 males, 16 females). None of the patients had personal or family histories of neurological diseases. Five patients had cardiovascular conditions (two coronary patients and one on anticoagulants) (10%), sixteen patients were diabetic (32%), sixteen were smokers (32%), and one patient consumed alcohol occasionally (2%). The TNM classification of the patients showed a varied distribution of tumor and nodal stages. Specifically, 50%, 24%, 8%, and 18% of the patients were at stages T2, T4, T1, and T3, respectively. For lymph node involvement, 40%, 34%, 14%, and 12% of the patients were at stages N2, N3, N0, and N1, respectively. Bilateral involvement was seen in 20 patients (40%), unilateral on the left side in 12 patients (24%), and unilateral on the right side in 11 patients (22%). Level II was involved in 84% of the cases, level III in 74%, level IV in 62%, level V in 38%, and high cervical involvement (levels I and VII) in 44%, respectively.

After a median follow-up of 36 months [22-68], 5 patients, representing 10% of the cohort, reported the onset of grade 1 neurological symptoms in the upper limbs according to the LENT/SOMA scale. The average delay was 30.75 months, with extremes ranging from 12 to 36 months after the end of radiotherapy. The table below summarizes the results by detailing the characteristics of the reported symptoms ( Table 2).

The median volume of the brachial plexus among the 50 patients studied was 10.35 cm ³ (range: 7.30–17.21 cm ³). The average maximum dose received by the brachial plexus was 73.57 Gy, with values ranging from 61.47 Gy to 82.64 Gy. The mean dose received by 2% of the volume was 70.97 Gy (range: 60.91–81 Gy), while 30% and 70% of the volume received mean doses of 61 Gy and 50.56 Gy, respectively ( Table 3).

The maximum dose to the brachial plexus increases with the extent of nodal involvement ( p < 0.05). The analysis showed a significant difference in the dose received by 0.03 cm ³ of the volume across the different nodal stages ( p < 0.05). As the nodal stage increases (N2, N3), the dose to this volume tends to exceed 70 Gy ( p < 0.05). The dose received by 2% of the brachial plexus volume tends to increase with nodal involvement, and thus with the N stage ( p < 0.05) ( Table 4).

Moreover, no significant dose differences were found between nodal stages for the other evaluated volume percentages of the brachial plexus (D30%, D70%) ( p > 0.05). Dosimetric data related to the levels of nodal involvement revealed significant differences between groups ( p < 0.05). The maximum dose (Dmax) as well as the D0.03 cm ³ to the brachial plexus were higher in patients with lower cervical nodal involvement (levels III and IV) ( p < 0.05) ( Table 5).

The RIBP is a relatively common complication after radiotherapy for breast or lung cancer but remains less frequent in head and neck cancers. Studies have indicated that the average annual incidence of RIBP among patients irradiated at the cervical or apical pulmonary levels is between 1.8% and 2.9%. Interestingly, a meta-analysis by Yan et al, which reviewed studies reporting RIBP in radiotherapy-treated patients, found a 0% incidence in those treated for head and neck cancers. ^{20, 21} In our series, after a median follow-up of 37.5 months (range: 22–68 months), three patients developed clinical signs of late RIBP, two at 24 months and one at 36 months post-treatment. Moreover only 5 patients treated exhibited grade 1 neurological symptoms in the upper limbs, according to the LENT/SOMA scale. It is worth noting that all radiological and electrophysiological investigations remained negative, suggesting either subclinical damage or delayed manifestation. This low incidence in nasopharyngeal cancer could be explained by several factors. First, the radiation fields used for nasopharyngeal tumors are generally located farther from the brachial plexus compared to fields used in breast and lung cancers, making plexus involvement less likely. Second, the often-long interval between radiation exposure and the onset of RIBP symptoms complicates clinical surveillance. Clinically, transient or early RIBP typically appears within 4 to 6 months after radiotherapy, with spontaneous resolution within 3 to 6 months and without permanent sequelae. Conversely, late RIBP generally occurs around 40 months post-treatment, with a widely variable reported incidence (1–73%), depending on diagnostic criteria, treatment techniques, and patient populations. At the histopathological level, radiation-induced nerve damage involves functional changes, such as abnormal action potentials and conduction delays, as well as structural injuries, including demyelination, enzymatic alterations, nerve sheath damage, and vascular remodeling. Over time, these changes decrease the brachial plexus’s ability to resist microtrauma from daily activities, potentially leading to progressive fibrosis and secondary nerve compression which is dose dependent. ^{22– 25}

An important and original finding of our study was the identification of significant correlations between nodal stage and brachial plexus dosimetry. Specifically, the Dmax delivered to the brachial plexus increased significantly with higher nodal stages ( p < 0.05). ²⁶ Similarly, the involvement of lower cervical nodal levels, particularly levels III and IV, was associated with higher Dmax values ( p < 0.05). Patients with advanced nodal disease (N2–N3) received significantly higher Dmax compared to those with early-stage disease (N0–N1). Furthermore, the dose delivered to 0.03 cm ³ of the brachial plexus (D0.03 cm ³) also showed an increasing trend with higher nodal stages ( p < 0.05). These results are in line with previous studies. ^{27, 28} For instance, Prakash et al. reported that patients with advanced nodal stages (N2–N3) received an average of 4.2 Gy higher doses to the brachial plexus compared to those with limited nodal involvement (N0–N1) ( p = 0.0001). ⁹ This observation is likely due to the anatomical proximity between the cervical lymph nodes and the brachial plexus, increasing the risk of overlap between the brachial plexus and the planning target volumes (PTVs). Indeed, it has been well established that the brachial plexus runs adjacent and medial to the elective cervical nodal levels II to IV. Therefore, in cases of large nodal volumes, it becomes challenging to avoid this overlap, especially when applying the recommended expansions for clinical target volumes according to Grégoire et al.’s consensus guidelines. ²⁷ Further supporting these findings, Truong et al. reported an 8.1 Gy increase in Dmax to the brachial plexus among patients with N2–N3 disease compared to N0–N1 (52.8 Gy vs. 60.1 Gy). ²⁸ Similarly, Xu et al. found that in patients with N3 disease, the brachial plexus Dmax reached 78.5 Gy, and the incidence of overdose at the plexus was as high as 68.5%. ²⁹ In our own cohort, the maximum dose to the brachial plexus also increased significantly with nodal involvement ( p < 0.05), and 83% of patients who received doses ≥66 Gy had advanced nodal stages (N2–N3), confirming these observations. The clinical relevance of brachial plexus dosimetry is further highlighted in the study by Chen et al., which showed that the volume of the ipsilateral brachial plexus receiving more than 70 Gy was predictive of neurological complications. Specifically, the risk of complications increased significantly when more than 10% of the plexus volume was exposed to doses ≥70 Gy ( p < 0.001). Similarly, for patients not undergoing neck dissection, a volume greater than 4% receiving 74 Gy (V74) was predictive of neurological symptoms ( p = 0.038). ²⁸ In our study, the mean percentage volume of the brachial plexus receiving 70 Gy was 15.96%, and that receiving 74 Gy was 6.61%. Despite these relatively high values, no confirmed cases of plexopathy were recorded, although the possibility of asymptomatic focal damage or late manifestations cannot be excluded. However, considerable discordance persists in the literature regarding brachial plexus contouring practices. The RTOG 0618 guidelines proposed by Hall et al. (2008) were among the first to standardize plexus contouring for head and neck cancers based on anatomical landmarks, primarily vertebral bodies and neck muscles. ¹⁹ Nevertheless, this protocol had limitations, particularly in its failure to account for anatomical variations such as root emergence from C5 to T1 and fusion of C2–C3 vertebrae, which could lead to contouring errors. ^{18, 30} Recognizing these limitations, Sun Ki Yi et al. conducted a study emphasizing the impact of contouring variability on dosimetry. They identified that the most significant source of inter-observer variability concerned the axillary portion of the plexus. ¹⁹ Consequently, they proposed modifications to the RTOG guidelines by adding two new items, incorporating additional anatomical landmarks such as the neurovascular bundle, sternoclavicular and glenohumeral joints, and the first two ribs. In our study, brachial plexus delineation was based on the RTOG guidelines complemented by Sun Ki Yi’s modifications. We adopted this approach because it represents the only contouring protocol developed within clinical trials and supported by dosimetric analyses, thereby allowing for more reliable conclusions when comparing our results with the existing literature. ^{31– 34}

Finally, regarding brachial plexus volumes, the study by Platteaux et al. which relied on MRI-assisted contouring — reported a mean volume of 6.7 cm ³ (range: 3–16.4 cm ³). ³³ Similarly, Truong et al. (using MRI guidance) reported a mean volume of 8.2 cm ³ (range: 3.7–12.7 cm ³). ³³ In contrast, Thomas (2015), who delineated the plexus without MRI based on RTOG recommendations, reported a mean volume of 11.6 cm ³. ³² Although MRI was not used for delineation in our study, the mean volume we obtained (10.39 cm ³, range: 7.30–17.21 cm ³) remains consistent with the values reported in the literature.

Despite the strengths of our study, several limitations must be acknowledged. The main limitations include its retrospective design and the relatively small sample size. The use of induction chemotherapy was not taken into account in the analyses. Although brachial plexus delineation was performed according to RTOG guidelines and reviewed by both a radiation oncologist and a head and neck radiologist, accurate visualization of the plexus remained extremely challenging — even with high-resolution CT — which may have led to over- or underestimation of its true anatomy. Additionally, a longer follow-up period is needed to fully assess the incidence and clinical impact of late radiation-induced brachial plexopathy in patients treated for nasopharyngeal carcinoma.

Our study highlights the low incidence of clinically apparent RIBP in patients treated with normofractionated radiotherapy for nasopharyngeal carcinoma, despite relatively high doses delivered to the brachial plexus in cases of advanced nodal disease. Significant correlations were observed between nodal stage, involvement of lower cervical nodal levels, and increased brachial plexus dosimetric parameters, particularly for the Dmax and D0.03 cm ³. Despite the relatively high volumes of the brachial plexus receiving doses ≥70 Gy, no confirmed cases of plexopathy were detected within the median follow-up period, although subclinical or late manifestations cannot be excluded. This suggests a possible dose-volume threshold below which the risk of symptomatic RIBP remains low in nasopharyngeal carcinoma patients. Future research should focus on prospective, multicenter studies with standardized brachial plexus delineation protocols, ideally incorporating MRI imaging to improve anatomical accuracy. Long-term follow-up will be essential to better assess the true incidence of late RIBP and to validate safe dose thresholds. Additionally, the development of adaptive radiotherapy strategies and advanced planning techniques, such as proton therapy or dose-painting IMRT, could further minimize brachial plexus exposure while maintaining tumor control. Finally, integrating clinical, imaging, and dosimetric biomarkers may help identify patients at higher risk for neurological toxicity and personalize radiotherapy plans accordingly.

This study was approved in advance by the Ethics Committee of [SALAH AZAIEZ INSTITUTE TUNIS TUNISIA]. The reference number: ISA/2025/28.

Verbal informed consent was obtained from all participants prior to inclusion in the study. Verbal consent was chosen instead of written consent because the study involved routine telephone follow-up of symptomatic patients. During these follow-up calls, patients who required EEG monitoring were informed that their EEG recordings and related clinical data could be used anonymously for retrospective research purposes. No identifying personal data were collected, and the geographically dispersed nature of participants made written consent impractical. The procedure for obtaining and documenting verbal consent was approved by the ethics committee.