A prospective study on association of fatigue with dose to central nervous system structures in head and neck irradiation

Abstract

Background

Fatigue is a common symptom in cancer patients accounting for 50–80% of patients and is multifactorial. Although many studies have focused on fatigue in head and neck cancer patients, studies on radiation dosimetric analysis of central nervous system (CNS) structures and their association with fatigue are rare. Hence, we have assessed patient-reported acute fatigue scores and correlated them with radiation dose received by brainstem, posterior fossa and pituitary.

Materials and methods

Forty-two head and neck cancer patients receiving radiotherapy from October 2018 to March 2020 were analyzed for fatigue scores using a questionnaire. Posterior Fossa, Pituitary and Brainstem structures were delineated and dose received by them was correlated with fatigue scores.

Results

Fatigue scores increased from 49 ± 12 at baseline to 78 ± 12 at the 6^th week of radiation treatment and reduced to 56 ± 18 at one month post treatment, but did not reach pre-treatment values. A statistically significant correlation was observed between posterior fossa D max and 6^th week fatigue scores; and brainstem D max and fatigue scores at one month post treatment.

Conclusions

Doses to brainstem and posterior fossa should be assessed and kept as low as reasonably possible to minimize fatigue.

Introduction

Fifty-eight percent of global head and neck cancers (HNC) occur in Asia and it accounts for nearly 30% of all cancers in India [1]. Radiotherapy (RT) forms an integral part in management of HNC with at least 75% of patients receiving radiotherapy as part of their treatment either in radical setting or in adjuvant setting with or without concurrent chemotherapy.

Fatigue is one of the most common symptoms in cancer patients. Around 58% of patients with cancer are affected by fatigue [2, 3]. Fatigue is a multifactorial condition that may result from the malignancy itself, RT or chemotherapy-induced side effects (e.g., anemia, dysphagia, weight loss), psychological factors (e.g., distress, sleep disturbance), or comorbidities [4, 5] Few recent studies also showed that irradiation of brain structures contributed to fatigue. [6]

The PARSPORT (Parotid-Sparing intensity modulated versus conventional radiotherapy in head and neck cancer) was a landmark trial in Head and Neck cancer, which showed a significant reduction in radiation induced toxicities by using the Intensity Modulated Radiotherapy (IMRT) technique [7]. However, an unexpected finding in this study was increased fatigue in patients who had received IMRT compared to 3D conformal technique, suggesting that radiation dose to brain structures could affect fatigue [6]. As fatigue is one of the most common concerns experienced during radiation in HNC, it has to be addressed for better patient well-being, improved quality of life, compliance and treatment outcomes.

Although many studies have focused on fatigue in head and neck cancer patients, few have correlated it with dose to central nervous system (CNS) structures. This study aims to help spare CNS structures and improve outcomes and quality of life [8].

Materials and methods

Study design and participants

This was a prospective observational study conducted at a tertiary cancer care center in India from October 2018 to March 2020. The study protocol was approved by the Institutional Ethics Committee, and all participants provided written informed consent. Adult patients (≥ 18 years) with histologically confirmed head and neck squamous cell carcinoma (HNSCC) were eligible if they were planned for radical or adjuvant radiotherapy, with or without concurrent chemotherapy. Patients with prior radiotherapy, synchronous malignancies, or metastatic disease at diagnosis were excluded.

Sample size calculation

The study was designed to detect a moderate correlation between radiation dose to CNS structures and fatigue scores. Based on standard sample size requirements for correlation analyses, a minimum of 42 patients is sufficient to detect a correlation coefficient of approximately r ≥ 0.45 with 80% power and a 5% significance level. This target is consistent with previously reported associations between radiation dose to brain structures and fatigue in head and neck cancer patients [14, 18]. Therefore, 42 patients were recruited, balancing statistical rigor with feasibility in a resource-limited setting.

Data collection and variables

The primary outcome was cancer-related fatigue, assessed using the validated Multidimensional Fatigue Inventory (MFI-20) at three time points: before radiotherapy (baseline), at the end of the 6^th week of radiotherapy, and one month after completion of treatment. Radiation dosimetric parameters for CNS structures: brainstem, posterior fossa, and pituitary were extracted from treatment planning systems. The following dose parameters were recorded: maximum dose (Dmax), mean dose (Dmean), and the dose received by 2cc of tissue (D2cc). Demographic and clinical covariates such as age, gender, Eastern Cooperative Oncology Group (ECOG) performance status, tumor site, stage, chemotherapy, and surgery were also recorded.

Radiotherapy planning and dosimetry

Patients were immobilized with a thermoplastic mask and underwent contrast-enhanced CT simulation with 3-mm slice thickness (coverage: vertex to T4 vertebra). Organs at risk (OARs), including the brainstem, posterior fossa, and pituitary gland, were delineated per standardized contouring atlases. Intensity-modulated radiotherapy (IMRT) was delivered with a simultaneous integrated boost (1.8–2.2 Gy/fraction, 5 fractions/week). Dose constraints were pituitary Dmax < 50 Gy and Dmax < 54 Gy for the brainstem and posterior fossa.

Bias control measures

To minimize reporting bias, the MFI-20 questionnaire was self-administered in the patient’s native language. A uniform protocol for contouring and treatment planning was followed to reduce measurement bias.

Statistical analysis

Descriptive statistics were used to summarize patient characteristics and fatigue scores. Pearson correlation was employed to assess the association between fatigue and radiation doses. To explore the independent association between radiation dose to CNS structures and fatigue scores, multiple linear regression analysis was performed using the MFI-20 fatigue scores as the dependent variable.

Separate linear regression models were constructed for fatigue scores measured at the 6^th week of radiotherapy and at one month post-treatment. Independent variables included dosimetric parameters — maximum dose (Dmax), mean dose (Dmean), and dose to 2cc (D2cc) — for each of the delineated CNS structures (brainstem, posterior fossa, and pituitary). In addition, clinical covariates such as age, sex, ECOG performance status, chemotherapy (yes/no), surgery (yes/no), and tumor site were included to adjust for potential confounders. For each model, unstandardized regression coefficients (β), p-values, and 95% confidence intervals were reported. A p-value < 0.05 was considered statistically significant. Assumptions of linearity, normality of residuals, and absence of multicollinearity (evaluated using variance inflation factor) were checked to ensure model validity. One-way ANOVA evaluated differences in fatigue based on chemotherapy or surgical intervention. All analyses were performed using SPSS software (Version 25)

Results

Patient characteristics

Of 42 patients enrolled, one patient was excluded due to a cardiovascular event during treatment, leaving 41 patients for analysis. The median age was 55 years (range: 27–75), with a male predominance (71%). ECOG performance status was 1 in 52.3% and 2 in 35.7% of participants. Most patients (71.4%) had no comorbidities. Primary tumor sites included the oral cavity (45.2%), oropharynx (19%), hypopharynx (14.3%), and larynx (14.3%). Stage IV disease was present in 64%.

Treatment characteristics

Seventeen patients (40.5%) underwent surgery followed by radiation, 7 (16.7%) had surgery followed by chemoradiation, and 18 (42.8%) were treated with chemoradiation alone. All patients received IMRT, with a mean treatment duration of 43 ± 8 days. Chemotherapy was administered at 45 mg/m² weekly.

Fatigue scores

Mean fatigue scores increased from 49 ± 12 at baseline to 78 ± 12 by the sixth week of radiotherapy, and then declined to 56 ± 18 at one-month post-treatment. The average increase in fatigue from baseline to the sixth week was +38 ± 13; the change from baseline to post-treatment was +15 ± 19.

Radiation dose to CNS structures

Table 1 presents Dmax, Dmean, and D2cc for the brainstem, posterior fossa, and pituitary gland. Posterior fossa Dmax showed a significant correlation with fatigue at week 6 (p < 0.001). Brainstem Dmax was significantly associated with fatigue at one month post-treatment (p = 0.007).

Table 1.

Dose received by central nervous system (CNS) structures

Structure	Mean [Gy]	Median (IQR) [Gy]
BS Dmax	41.36	42.23 (34.89–53.79)
BS D2cc	31.63	32.30 (23.22–41.28)
BS Dmean	13.75	14.28 (6.18–20.12)
Pituitary Dmax	10.77	3.60 (1.69–18.8)
Pituitary Dmean	7.40	2.50 (1.32–11.1)
Posterior fossa Dmax	45.83	48.45 (39.99–55.5)
Posterior fossa D2cc	35.84	36.37 (30.51–43.50)

Dmax — maximum dose received by the organ; Dmean — mean dose received by the organ, D2cc — dose received by 2cc of the organ; BS — brainstem, IQR — interquartile range, Gy — Gray

Table 2.

Treatment protocols

Treatment protocol used	Number of patients
Surgery + radiation	17
Surgery + chemoradiation	7
Chemoradiation	18

Table 3.

Correlation of Multidimensional Fatigue Inventory (MFI) score at 6^th week with Dose to central nervous system (CNS) structures

CNS Structures	Correlation coefficient (r)	Sig 2 tailed ( p)
Brainstem D max	0.393	0.005
Brainstem Dmean	0.518	< 0.001
Brainstem D2cc	0.392	0.006
Pituitary Dmax	0.507	< 0.001
Pituitary Dmean	0.494	0.001
Posterior fossa Dmax	0.557	< 0.001
Posterior fossa Dmean	0.358	0.011
Posterior fossa D2cc	0.423	0.003

Table 4.

Linear regression analysis of correlation of dose to central nervous system (CNS) structures and Multidimensional Fatigue Inventory (MFI) fatigue scores at 6^th week

CNS structures	B	Sig
Brainstem D max	0.099	0.597
Brainstem Dmean	0.299	0.100
Brainstem D2cc	0.108	0.539
Pituitary Dmax	0.597	0.067
Pituitary Dmean	0.100	0.128
Posterior fossa Dmax	0.524	< 0.001*
Posterior fossa Dmean	0.539	0.900
Posterior fossa D2cc	−0.197	0.507

^*95% confidence interval (CI): −0.268–0.824

Table 5.

Correlation of multidimensional fatigue inventory (MFI) score at 1 month with dose to central nervous system (CNS) structures

CNS Structures	Correlation coefficient (r)	Sig 2 tailed (p)
Brainstem D max	0.511	0.001
Brainstem Dmean	0.85	0.595
Brainstem D2cc	0.327	0.037
Pituitary Dmax	0.104	0.518
Pituitary Dmean	0.087	0.589
Posterior fossa Dmax	0.289	0.067
Posterior fossa Dmean	0.207	0.193
Posterior fossa D2cc	0.343	0.028

Table 6.

Linear regression analysis of correlation of dose to central nervous system (CNS) structures and multidimensional fatigue inventory (MFI) fatigue scores at 1 month

CNS Structures	B	Sig
Brainstem Dmax	0.413	0.007*
Brainstem D2cc	−0.373b	0.124
Posterior fossa Dmax	−0.130b	0.515

^*95% confidence interval (CI): −0.185–1.108

Figure 1.

Scatter plot showing correlation of posterior fossa Dmax and fatigue at 6^th week Fatigue score at 6^th week10 20 30 40 50 60 70604020PF Dmax doseR2 linear = 0.275

Figure 2.

Scatter plot showing correlation of posterior fossa Dmax and fatigue at 1 month Fatigue score at 1 month post treatment–20 0 20 40 606050403020100PF Dmax doseR2 linear = 0.171

Regression analysis

Linear regression revealed that each 1 Gy increase in posterior fossa Dmax was associated with a 0.524-point increase in fatigue score at week 6. Similarly, each 1 Gy increase in brainstem Dmax predicted a 0.413-point increase in fatigue score at one-month post-treatment. These relationships are illustrated in Graphs 1 and 2, respectively.

Subgroup analysis

Fatigue was significantly higher in the chemotherapy group at baseline compared to the non-chemotherapy group. Although a similar trend was observed at 6 weeks and 1 month, the differences at these time points were not statistically significant.

Discussion

Cancer related fatigue (CRF) is extremely common, with most studies reporting prevalence ranging from 60% to 90% [1]. It is also one of the most important factors affecting quality of life in oncology today and yet poorly understood. Unlike acute fatigue, which comes on quickly, lasts a short time, and is relieved by rest, CRF is prolonged, debilitating, and not relieved by rest [4, 12]. However, fatigue as a common symptom was often unrecognized until recently. As the emphasis on quality of life of oncology patients have been increasing, more studies are being undertaken to assess and reduce fatigue in cancer patients.

Various organizations have defined CRF, and the most commonly used definition is that given by the National Comprehensive Cancer Network. CRF is defined as a persistent subjective sense of physical, emotional, and cognitive tiredness or exhaustion related to cancer or its treatment that is not proportional to recent activity and that significantly interferes with usual functioning [13]. To analyze this CRF, many questionnaires such as Brief fatigue inventory scale, Piper fatigue scale, The European Organization for the Research and Treatment of Cancer Quality of Life Questionnaire version 3.0 and The Functional Assessment of Chronic Illness Therapy-Fatigue Scale, have been used. The MFI-20 questionnaire used in this study takes into account the major contributing factors of fatigue such as reduced activity, reduced motivation, physical and mental fatigue [10, 11].

Jereczek et al. and Ferris et al. also used the MFI-20 questionnaire and observed an increase of 17 and 10 points from baseline at the 6^th week and decrease by 11.5 and 3.5 point at one month post radiation completion, respectively [12, 14]. Fatigue scores in our study at the 6^th week increased by 29 points from baseline and decreased by 22.5 points from the 6^th week to one month post radiation completion.

The higher fatigue scores in the present study may be attributable to certain patient factors like poor nutritional status, lower socioeconomic status; and tumour factors like advanced stage of disease at presentation. Even though the points were higher in our study, the trend remained the same and we observed a decrease in fatigue scores post treatment; however, they did not return to baseline fatigue levels, similar to findings in multiple other studies [15].

In our study the more pronounced fatigue observed in the chemotherapy group at baseline likely reflects pre-existing disease burden and associated systemic effects such as inflammation and nutritional compromise. These patients may have entered the study already fatigued due to both the physical impact of advanced disease and the psychological burden of an anticipated intensive treatment regimen.

However, by the 6^th week and 1 month post-treatment, the difference in fatigue between the chemotherapy and non-chemotherapy groups was no longer statistically significant. This could be attributed to several factors:

fatigue in non-chemotherapy patients increased during treatment, narrowing the gap;

patients in the chemotherapy group may have experienced partial recovery after the initial chemotherapy course, or adapted physiologically and psychologically to treatment-related fatigue;

supportive care measures during treatment (e.g., symptom management, nutritional support) may have ameliorated fatigue across all groups.

While chemotherapy patients started with a higher fatigue burden, the converging fatigue levels over time may reflect a combination of increased treatment-related fatigue in non-chemo patients and recovery or adaptation in chemo-treated patients.

Leighton et al. studied magnetic resonance imaging (MRI) changes such as volume loss of midbrain and disrupted homeostasis in BS, cerebellum and hypothalamus in patients with fatigue as opposed to those unfatigued. They concluded that chronic fatigue involves insult to midbrain, which suppresses motor and cognitive activity and via feedback mechanism disrupts the CNS homeostasis in other areas and disrupts the hypothalamic–pituitary axis [16]. The dosimetric evaluation of PARSPORT trial by Gulliford et al. concluded that excessive fatigue reported in the IMRT arm of the trial may be attributed to the dose distribution to the PF, cerebellum and BS — which forms the basis for this study. Gulliford et al. found that both maximum and mean doses were significantly higher for the PF, BS and cerebellum for the patients who reported acute fatigue of Grade 2 or higher compared to those that did not. This may also be attributed to the fact that they used only physician-reported fatigue grading, which is different in comparison to the present study [17]. In this study Dmean values to the serial structures considered, did not show any correlation with fatigue scores at any point in time but Dmax was found to be correlated.

A statistical significant correlation was observed in the present study between BS Dmax, PF Dmax, and MFI 20 scores at the 6^th week and one-month post-radiation, similar to the findings by Ferris et al. [14].

In a study by Anderson group and Powell et al., a significant relationship was observed between acute fatigue scores and Dmax, Dmean and D2cc to the pituitary in patients with nasopharyngeal carcinoma [18]. In contrast, we found no correlation between pituitary dose and fatigue scores in our study. This is likely due to the minimal number (2%) of nasopharyngeal cancer cases in our cohort and the small size of pituitary (mean volume 1.5 cc), which prevented D2cc dose recording in these cases.

Other than PF, BS and pituitary, the basal ganglia and hypothalamus are additional CNS structures are known to be associated with CRF [19]. However, these were not analyzed in the present study, as majority of the patients had oral cavity tumors — in whom doses received by these structures are expected to be very low. Additionally, MRI-based delineation was not logistically feasible in our patient population.

Although linear regression was used to analyze associations between radiation dose to central nervous system structures and patient-reported fatigue scores, potential methodological limitations must be acknowledged. First, linear regression assumes a linear relationship between variables, which may not accurately reflect complex biological interactions. Second, multicollinearity among related CNS structures — such as the brainstem, cerebellum, and posterior fossa — can reduce the reliability and interpretability of regression coefficients. Third, the model’s sensitivity to outliers and the assumption of normally distributed residuals may affect the robustness of the results. These statistical limitations are well recognized in literature and should be considered when interpreting our findings (Kim, 2019; Schneider et al., 2010). [20, 21].

This is the only prospective study on Indian HNC patients correlating dose to CNS structures with patient-reported fatigue using the MFI questionnaire, which addressed all aspects of fatigue. This method of assessment is superior to physician-reported grading. Furthermore, all patients were treated with IMRT technique and this uniformity in the radiation method is an added strength of our study.

Conclusions

When evaluating IMRT plans for head and neck cancer patients treated with definitive intent, attention must be paid to PF and BS doses. Even a 1 Gy increase leads to 0.4–0.5 point rise in fatigue score. Although fatigue decreases post-RT, it does not normalize. Thus, stringent constraints should be applied to PF and BS.