Evolution of radiation-induced temporal lobe injury after intensity-modulated radiation therapy in nasopharyngeal carcinoma: a large cohort retrospective study

Radiotherapy remains the mainstay of treatment for nasopharyngeal carcinoma (NPC) due to its complicated anatomic location and unique radiotherapy-sensitivity [1]. Since NPC often represents close proximity and infiltration to skull base, temporal lobes are inevitably included into the target volume despite the use of the more advanced intensity-modulated radiotherapy (IMRT) [2‐4]. Radiotherapy-induced temporal lobe injury (RTLI) is one of the most serious complications of NPC [1]. The incidence of RTLI among patients receiving IMRT ranged from 4.6 to 8.5% [5]. RTLI can lead to various neurological symptoms including headache, seizures, focal neurological deficits, and cognitive dysfunction [5, 6]. A clear understanding of the early pattern and evolution of RTLI on MRI is crucial for the early treatment of RTLI and the selection of optimal timing for long-term follow-up of RTLI survivors.

According to the timing of radiotherapy, the pathophysiological response of normal brain tissue to radiation can be divided into three stages: the acute reaction period (days to weeks), early delayed radiation period (1–6 months), and late delayed radiation period (6 months to a few years) [7]. Magnetic resonance imaging (MRI) has been considered a useful and noninvasive method for evaluating RTLI and its evolution [8, 9]. White matter lesions (WMLs) and contrast-enhanced lesions (CELs) are the two main MRI manifestations of late radiation injury and can occur separately or together [10]. However, the natural course of these two common patterns of RTLI is still not fully understood, and there are conflicting findings regarding the initial MRI patterns and the evolution of these two patterns of lesions.

Morphological characteristics of RTLI on MRI have been reported in several studies [7‐14]. However, these previous studies had limitations, including small sample sizes and inconsistent results. Wang et al. evaluated RTLI evolution in 124 patients with 192 temporal lobe injuries and concluded that RTLI may be reversible, with WMLs appearing first, followed by CELs [10]. In contrast, Zhou et al. [14] investigated the natural course and evolution of RTLI in 105 patients with NPC after radiotherapy and found that solid-enhanced nodular lesions were the earliest MRI abnormalities of RTLI. Therefore, further research with a larger sample size is required to confirm the initial MRI patterns and evolution of RTLI.

This study aimed to enhance understanding of the initial MRI patterns and evolution of RTLI with a large sample size, which may provide a robust theoretical foundation for early and accurate diagnosis, the selection of appropriate follow-up intervals, and timely treatment options for RTLI.

To the best of our knowledge, this study is the largest cohort investigation of the evolution of RTLI in NPC. We found that both WMLs and CELs may manifest as the earliest and sole MRI abnormalities in RTLI. Notably, the evolution of bilateral temporal lobe brain injuries could be different within the same patient, and the evolution of unilateral temporal lobe injury combined with WMLs and CELs may not occur synchronously. In addition, our results suggest that the time interval between the initial detection of RTLI and the last negative MRI examination is an independent predictor of the earliest MRI pattern of RTLI.

Regarding the earliest MRI pattern in RTLI, the results of this study do not completely align with those of previous studies [10, 11]. Wang et al. [10] and Chan et al. [11] reported that WMLs were the earliest abnormal MRI findings of RTLI, with CELs occurring after or concurrently with WMLs. Animal studies have demonstrated that WMLs are the earliest forms of radiation injury [6, 15]. However, Zhou et al. [14] reported that solid-enhanced nodular lesions were the earliest and only initial abnormalities of RTLI. A study by Chan et al. [11] evaluating the morphological characteristics of RTLI on MRI reported a result similar to that of Zhou et al. [14] in which two temporal lobes with blood–brain barrier disruption showed no evidence of white matter lesions. In our study, we found that both WMLs and CELs were the earliest and only MRI patterns associated with RTLI, with isolated CELs having a much higher incidence than isolated WMLs (15.8% vs. 0.6%). The discordance between these studies can be attributed to several reasons. Firstly, the mechanism of radiation-induced brain injury remains unclear and may involve microvascular injury [16], neuronal and neural stem cell (NSC) injury [17‐19], glial cell injury [20], inflammation and free radical production [21, 22]. These changes can lead to acute disruption of the blood–brain barrier, increased permeability, and edema [23‐25]. Different pathological changes in RTLI may result in different initial manifestations on MRI. Secondly, the follow-up intervals were significantly different among the different studies. In our study, the median time interval between the last previous negative MRI examination and subsequent detection of RTLI on MRI was 10.6 months, which was very close to the 10.5 months reported by Zhou et al. [14], but much shorter than the 20.5 months reported by Wang et al. [10]. Additionally, the time intervals from the first MRI detection of isolated WMLs, isolated CELs, and combined WMLs and CELs to the last previous negative MRI examination in our study were 8.6, 8.9, and 11.0 months, respectively, similar to Zhou et al.’s [14] results demonstrating a shorter time interval between the first MRI detection of CELs and the last negative MRI compared to that in those with multiple MRI components at the first detection of RTLI (5.5 vs 10.5 months, respectively). Moreover, the time interval between the first detection of RTLI and the last previous negative MRI examination was identified as an independent factor influencing the earliest MRI patterns of RTLI based on the results of the Kruskal–Wallis test and Poisson regression. In theory, shorter intervals between MRI examinations may provide more precise information about the natural history of RTLI. However, in practice, some patients did not adhere to regular follow-up, which resulted in prolonged intervals and delayed detection of lesions. Therefore, regular follow-up and MRI examinations, as recommended by guidelines, are crucial for early detection of RTLI and understanding the early MRI pattern of RTLI.

In this study, we observed that the evolution of bilateral temporal lobe injuries varied within the same patient, which supports the evolutionary pattern previously reported by Wang et al. [10]. Additionally, contrary to the commonly accepted principle, we observed that the evolution of WML and CEL may be not synchronized in cases of unilateral temporal lobe injury combined with WML and CEL. Specifically, an increase in CELs may not be accompanied by a corresponding increase in WMLs, and WMLs may decrease or remain static in some cases. Previous studies have reported a positive correlation between edema volume and the incidence of enhancement and necrosis [3, 10], however, our study is the first to demonstrate that the evolution of WMLs and CELs may not occur synchronously. The exact reason for this phenomenon is not fully understood.

The median time interval between the completion of radiotherapy and the first MRI detection of RTLI showed significant variation across studies. In our study, the median interval time was 34.1 months (range, 5.7–101.9 months), which aligns with the findings of Zeng et al. [4] who reported a median interval of 33 months. However, our results indicate a slightly shorter interval (34.1 months) than the 36 and 37 months reported by Wang et al. [10] and Zhou et al. [14], respectively, and a much shorter interval than the 55.9 and 44.5 months reported by Norris et al. [26] and Mao et al. [27], respectively. According to the NCCN guidelines, regular follow-up and MRI examinations should be performed at 3-month intervals during the first year, 6-month intervals during the second year, and yearly intervals thereafter for patients with NPC [28]. Nevertheless, a small number of patients in our study did not undergo routine follow-up MRI. Patients with RTLI typically do not exhibit symptoms in the early stages, and the majority of temporal lobe injuries are incidentally discovered during later reviews. As a result, initial MRI detection of RTLI can occur at any stage of disease progression. This limitation hinders the accurate assessment the exact timing of the initial onset of RTLI and its subsequent evolution.

In this study, we investigated the relationship between the dosimetric parameters and the initial RTLI patterns. However, our findings revealed no significant differences in the dosimetric parameters, including D_min, D_max, and D_mean, among the three RTLI patterns. Radiation dose is considered a direct causal factor for RTLI [6, 29‐32]. A previous study confirmed significant differences in dosimetric parameters between RTLI-positive and RTLI-negative lobes in patients with NPC [33]. Additionally, a study using a rat model concluded that the time interval from radiotherapy completion to the onset of RTLI was dose-dependent; however, once the initial onset occurred, the rate of injury progression and total volume generated remained constant across different doses [34]. Therefore, we speculate that the radiation dose impacts the occurrence and timing of radiation-induced brain injury but does not influence the pattern of initial MRI manifestations of RTLI.

This study had several limitations. Firstly, due to the retrospective nature of the study, routine MRI follow-up was not conducted in all patients, which prevented the continuous observation of the entire dynamic process of RTLI evolution. Secondly, less common RTLI patterns, such as gray matter lesions, hemosiderin deposits, and microbleeding foci, were not included in the current study. The main objective of this study was to evaluate the two common MRI manifestations of RTLI, WMLs and CELs, and their temporal changes. Including an analysis of all types of radiation-induced brain injuries would divert the focus of this study. Additionally, the above-mentioned less common patterns are not the primary imaging manifestations of RTLI and typically appear in the later stages.

In conclusion, we found that both WMLs and CELs could be the earliest and only MRI abnormalities associated with RTLI. In addition, to the best of our knowledge, our study indicated for the first time that the evolution of WMLs and CELs may not be synchronized in cases of unilateral temporal lobe injury combined with WMLs and CELs. Furthermore, the time interval between the initial detection of RTLI and the last MRI examination was identified as an independent factor influencing the earliest MRI patterns of RTLI. Regular follow-up intervals can provide more accurate information about the true nature of RTLI. These results are of great significance for the accurate diagnosis of RTLI and timely treatment options. Moreover, the mechanisms underlying the different initial MRI patterns of RTLI are also distinct. Studying the mechanisms behind the initial onset of RTLI may provide possibilities for early mechanism-based interventions. From this perspective, this study provides valuable insights for endeavors aimed at unraveling the mechanisms underlying the occurrence of RTLI.