Keywords nasopharyngeal neoplasms - hearing loss - radiotherapy - proton - otitis media with
effusion
Introduction
Nasopharyngeal carcinoma (NPC) is common in the Asian population, particularly in
the southeast region, including Thailand. According to the Global Cancer Observatory
(GLOBOCAN) 2020, there were 2,316 new NPC cases and 1,482 deaths, with an annual incidence
of 2.6/100 thousand for men and 1/100 thousand for women, in 2020.[1 ]
[2 ] According to the National Comprehensive Cancer Network guideline, the current gold
standard for treatment is radiation with or without chemotherapy.[3 ] Intensity-modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT)
techniques are considered as the standard treatment for NPC.[4 ]
[5 ]
[6 ] A type of photon or X-ray beam therapy, IMRT has gained popularity because of its
ability to deliver radiation more precisely to the tumor while sparing the nearby
radiosensitive tissues, such as those of the brain, orbits, cochlea, and spinal cord.[7 ] Volumetric modulated arc therapy is another IMRT technique, in which the machine
rotates around the patient delivering radiation beams in an arc-like pattern. The
radiation dose intensity was determined by the amount of radiation left after passing
through tissues, type of radiation particles, volume of the irradiated organ, and
type of organ.[8 ] Patient factors, such as underlying disease, smoking behavior, and concurrent chemotherapy
treatment, all had considerable impact.[9 ] Proton beam therapy is a novel technique that uses the proton particle for energy
delivery. Current research supports the use of proton beam therapy since it is associated
with less side effects when compared with photon beam therapy.[8 ]
[10 ] However, due to its cost and limited availability, its usage should be reserved
for individuals who are expected to experience significant side effects from regular
IMRT or VMAT.
Despite modern techniques, radiotherapy (RT) can cause adverse effects on surrounding
tissues, such as hypothyroidism, nasopharyngeal fibrosis, chronic rhinosinusitis,
and hearing loss.[11 ]
[12 ] Approximately 73% of patients who underwent RT to the head and neck developed conductive,
sensorineural, or mixed hearing loss.[13 ] Conductive hearing loss commonly manifests at 3 months after treatment due to middle
ear effusion (MEE) (8–29%), tympanic membrane stiffness (15–32%), and ossicular fibrosis
(5%), whereas sensorineural hearing loss develops later due to cochlea or auditory
nerve injury.[14 ] The relationship between radiation dose to auditory structures and toxicity has
been previously reported.[15 ]
[16 ] A mathematical model to predict the risk of radiation-induced ototoxicity in NPC
patients was mostly based on radiation dose to the cochlea, but did not consider other
clinical factors.[17 ]
[18 ]
[19 ] However, the risk factors of ototoxicity in NPC patients include radiation to the
auditory pathway (internal acoustic canal, middle ear, cochlea, and Eustachian tube)
and the use of cisplatin chemotherapy based on several studies in the non-IMRT[17 ]
[18 ]
[20 ]
[21 ] and IMRT eras.[22 ]
[23 ] Patients aged 50 years and older often experience a higher incidence of age-related
hearing loss, which can exacerbate the impact of radiation-induced hearing impairment,
making these findings more prominent and troublesome in this population.[24 ]
Therefore, the purpose of the current study is to develop a multivariable normal tissue
complication probability (NTCP) model that can predict the risk of hearing impairment
using clinical and radiation dosimetry features. We also aimed to estimate the likelihood
of audiologic complications following RT, to identify the factors that cause hearing
loss in NPC patients who have undergone IMRT or VMAT, and to report the pattern and
severity of hearing loss in NPC patients following RT.
Methods
Through a chart review, this retrospective cohort, prognostic study included patients:
1) whose NPC diagnosis was confirmed by histopathology and who were treated with IMRT
or VMAT at a dose of 66 to 70 Gy in 30 to 35 fractions, with or without chemotherapy,
at King Chulalongkorn Memorial Hospital (KCMH) between December 2008 and December
2019; 2) with available radiotherapy dose–volume histogram (DVH) data; and 3) with
the audiogram and computed tomography (CT) scan reported prior to and following treatment.
We also excluded patients: 1) who previously received radiation or chemotherapy in
the head and neck region; 2) with incomplete prescribed treatment plan; 3) with insufficient
information to identify the relevant factors; for instance, the middle ear was not
visible on the CT scan, image artifacts, or no audiogram report; and 4) who previously
had MEE prior to treatment.
Data was collected from the electronic medical records of the NPC patients who received
treatment at the KCMH otolaryngology and radiation oncology department. The demographic
data included age at the initial RT administration, sex, Tumor, Node, and Metastasis
(TNM) staging,[3 ] chemotherapy (in line with the National Comprehensive Cancer Network (NCCN) guidelines,[3 ] the chemotherapy regimens in this study were all cisplatin-based) and dose-volume
histogram (DVH) data (mean and maximum radiation dose [Gy] corresponding to the following
volumes under the supervision of an RT specialist: [1] gross tumor volume [GTV], [2]
clinical target volume [CTV], [3] planning target volume [PTV], and [4] each cochlea).
Posttreatment audiogram and CT scan were collected at least 3 months following the
last treatment session.
Treatment
All patients were immobilized in the supine position with a tailored head–shoulder
thermoplastic mask, then a CT simulation was performed. Magnetic resonance simulation
was performed on every patient and co-registration with the CT images. Two or three
PTVs were designated as follows: PTV-high risk (PTV-HR) was defined as gross tumor
and gross pathologic lymph nodes (LNs) and received doses of 70 Gy; PTV-intermediate
risk was defined as the subclinical disease and received prophylactic doses of 60
to 70 Gy; and PTV-low risk (PTV-LR) was defined as the elective LN region (bilateral
cervical LN levels II–V, VII) and received doses of 50 to 56 Gy. A simultaneous integrated
boost (SIB) or sequential cone-down boost of 20 Gy to the PTV-HR was selected by the
physician's decision. IMRT or VMAT was applied. Radiotherapy planning was performed
using the Eclipse treatment planning system (Varian Medical Systems Inc., Palo Alto,
CA, USA) version 6.5–15.0. Radiation was delivered using a linear accelerator (Varian
Medical System Inc.) Treatment verification was performed regularly with daily electronic
portal images and weekly cone-beam CT.
All patients received concurrent chemoradiotherapy, with or without neoadjuvant or
adjuvant chemotherapy. Cisplatin was administered weekly or tri-weekly concurrently
with definitive radiotherapy at a dose of 70 Gy in 33 to 35 fractions. Neoadjuvant
or adjuvant chemotherapy regimens, including platinum-based (cisplatin/carboplatin),
infusion fluorouracil (5 FU), paclitaxel, or gemcitabine, were given at 3- or 4-week
intervals for 3 cycles.
Outcome Measurements
The primary outcome of the present study was the incidence of hearing loss, which
was determined by using an audiogram. Pure tone average (PTA) of both air and bone
conductions (3 frequency protocols) as well as the duration of the threshold worsening
by > 10 dB HL were collected as a secondary outcome of the study. Middle ear effusion
was determined by the presence of fluid between −5 and 20 Hounsfield units (HUs) in
the middle ear cavity on the CT scan.
Statistical Analysis
Categorical data were analyzed using the Chi-squared or Fisher's exact test. Continuous
data were analyzed using the Wilcoxon signed rank-sum test or Student t-test. The
significant factors for toxicity were identified using binary logistic regression
with significance set at a p -value of < 0.05. Multivariate logistic regression (forward stepwise selection/bootstrapping)
was used to create an NTCP model:
with S(x) = β0 + β1 x1 + β2 x2 + … + βn xn ,
in which β0 is a constant and β1 , …, βn are the logistic regression coefficients of the variables x1 , x2 , …, xn , respectively.
The performance of the model was assessed by the area under the receiver operating
characteristic curve (AUC) analysis and the Hosmer-Lemeshow goodness-of-fit test,
whereas a non-significant p -value of > 0.05 indicated good predictive ability. Finally, internal validation was
performed with the 10-fold cross validation. The data were analyzed using STATA (StataCorp
LLC, College Station, TX, USA), version 15.1.
The present study was approved by the Institutional Review Board of the Faculty of
Medicine at Chulalongkorn University (COA No. 303/63). The need to obtain patient
informed consent was waived by the institutional review board because of the retrospective
nature of the study.
Results
Altogether, 835 NPC patients receiving photon-based IMRT were identified from the
database, but 587 cases were removed based on the exclusion criteria. The data of
the remaining 248 patients were reviewed. Additionally, 19 patients were further excluded
because they were not given full radiation treatment as planned. Finally, 229 participants
with a total of 458 ears were included in the analysis, as demonstrated in [Fig. 1 ]. The demographic data, TNM classification, and treatment characteristics are shown
in [Table 1 ]. Most patients were men. The patients' mean age was 49.43 years (standard deviation
[SD] = 14.03). Nasal symptoms were the most common presentation (53%) followed by
neck mass (45%), auditory symptoms (44%), such as hearing loss and aural fullness.
The other signs and symptoms were neural involvement (11%), such as diplopia, facial
numbness, or headache. The maximum dose (Dmax ) to GTV was 74.25 Gy (SD 2.48); CTV, 72.61 Gy (SD 2.84); PTV-70, 71.48 Gy (SD 3.22);
and CV, 45.45 Gy (SD 13.19). The median duration of hearing follow-up after therapy
was 130 days. Most patients received VMAT rather than the conventional IMRT and were
concurrently submitted to chemotherapy during RT (93.01%).
Fig. 1 Flow diagram of patient recruitment.
Table 1
Demographic data according to the patients' baseline and treatment characteristics
Baseline characteristics at presentation
Patients (n = 229)
Percentage (%)
Treatment characteristics
Patients (n = 229)
Percentage (%)
Gender
Female
81
35.37
Radiation technique
IMRT
24
10.49
Male
148
64.63
VMAT
205
89.51
Age
< 50 years
102
44.54
Concurrent chemotherapy
No
16
6.99
≥ 50 years
127
55.46
Yes
213
93.01
Signs and symptoms
Nasal
122
53.27
Follow-up audiogram (days)
< 31
3
1.31
Aural
102
44.54
31–90
74
32.31
Neck mass
103
44.97
91–180
55
24.02
Neural
26
11.35
181–365
48
20.97
TNM stage
T (%)
N (%)
M (%)
> 365
49
21.39
0
50 (21.9)
222 (96.9)
Average follow-up (days)
254.86
1
66 (28.8)
61 (26.6)
7 (3.1)
Median follow-up (days)
130
2
58 (25.3)
92 (40.2)
3
55 (24.0)
26 (11.4)
4
50 (21.8)
Abbreviations: IMRT, intensity-modulated radiotherapy; M, metastasis; N, node; T,
tumor; TNM, tumor, node, and metastasis staging; VMAT, volumetric modulated arc therapy.
Sensorineural Hearing Loss
Hearing loss was observed in 96 out of 229 (41.92%) patients. The demographic data
were divided into 2 groups based on post RT hearing status as presented in [Table 2 ]. There was a statistically significant difference in age, T stage, and mean radiation
exposure (Dmean ) to the cochlea of > 43 Gy between the two groups. The cochlea dosage threshold,
or cut-off value, of 43 Gy was calculated using the maximum AUC of 0.663. The results
of the univariate and multivariate logistic regression analyses are depicted in [Table 3 ]. In the univariate analysis, age > 50 years, stages 3 and 4, and cochlea dose > 43
Gy were related to odds ratios (ORs) of 1.68, 1.60, and 2.58, respectively. The final
NTCP model comprised age and cochlea dose with an AUC of 0.644 (precision of 70.1%).
The calculated risk of hearing loss ranged between 15.84 and 44.52%, as illustrated
in [Table 4 ].
Table 2
Comparison between the normal hearing and hearing loss groups
Analyzed factors
Group I: normal hearing (N = 133)
Group II: hearing loss (N = 96)
p -value
Gender: male – n (%)
85 (63.9)
63 (65.6)
0.79
Age in years (IQR)
49 (39–56)
55.5 (43.5–61)
0.002*
Age groups: n (%)
• < 50 years
• ≥ 50 years
68 (51.1)
65 (48.9)
34 (35.4)
62 (64.6)
0.04*
Symptoms at presentation: n (%)
Aural
55 (41.4)
47 (49)
0.25
Nasal
66 (49.6)
56 (58.3)
0.19
Mass
59 (44.4)
44 (45.8)
0.83
Neural
11 (8.3)
15 (15.6)
0.11
Staging: n (%)
T3–4
53 (39.6)
52 (54.2)
0.03*
N2–3
64 (47.8)
54 (56.3)
0.20
M1
2 (1.5)
5 (5.2)
0.11
Treatment protocol
Dmax : GTV-NP (IQR) in cGy
7,428 (7,301–7,529)
7,458 (7,306–7,578)
0.42
Dmax : CTV-NP (IQR) in cGy
7,274 (7,156–7,431)
7,302 (7,107.5–7,455.5)
0.53
Dmax : PTV-70 (IQR) in cGy
7,192 (6,997–7,325)
7,252 (6,990–7,376.5)
0.32
Using chemotherapy: n (%)
119 (89.5)
94 (97.9)
0.28
Analyzed factors
Group I: post-RT normal hearing, ears (
N
= 321)
Group II: post-RT hearing loss, ears (
N
= 137)
p
-value
Dmean : cochlear dose (IQR) in cGy
4,157 (3,588–5,017)
4,817 (3,893–5,810)
< 0.001*
Abbreviations: CTV, clinical target volume; Dmax , maximum dose; GTV, gross tumor volume; IQR, interquartile range; M, metastasis;
N, node; NP, nasopharynx; RT, radiotherapy; T, tumor.
Note: Bold values represent p -value < 0.05.
Table 3
Univariate and multivariate analyses of hearing loss
Analyzed factors
Univariate analysis
Multivariate analysis
Factors
Reference
OR (95%CI)
p -value
aOR (95%CI)
p -value
Male
Female
1.12 (0.73–1.71)
0.06
Age group ≥ 50 years
< 50
1.68 (1.11–2.53)
0.01*
1.67 (1.09–2.55)
0.02*
T 3–4
0–2
1.60 (1.07–2.39)
0.02*
N 2–3
0–1
1.42 (0.95–2.13)
0.29
M1
0
1.79 (0.61–5.27)
0.29
Cochlear dose≥ 4,300 cGy
< 4,300
2.58 (2.69–3.92)
< 0.001*
2.57 (1.68–3.92)
< 0.001*
Abbreviations: 95%CI, 95% confidence interval; aOR, adjusted odds ratio; OR, odds ratio; M, metastasis;
N, node; T, tumor.
Table 4
Normal tissue complication probability model and calculated risk factor for hearing
loss
Age at RT ≥ 50 years
Cochlear dose > 43 Gy
Modifying factor S (x)
Risk factor for hearing loss (%)
Yes
Yes
−0.22
44.52
Yes
No
−1.16
23.86
No
Yes
−0.73
32.51
No
No
−1.67
15.84
Abbreviation: RT, radiotherapy.
[Fig. 2 ] depicts the pre- and post-treatment audiologic characteristics of air conduction
pure tone audiometry (PTA), bone conduction PTA, and bone conduction specific frequencies.
Pure tone audiometry for air and bone conductions decreased by 8.77(± 14.93) and 7.7(± 11.82)
dB after RT, respectively. The hearing loss was detected in every frequency and was
more pronounced at high frequencies.
Fig. 2 Boxplot diagrams of audiograms (dB HL) between pure-tone audiometry and specific
frequencies of bone conduction.
MEE
Middle ear effusion had been newly diagnosed in 92 patients (42.79%) out of the 215
patients. The demographic data for groups with and without MEE are presented in [Table 5 ]. There was a statistically significant difference between the 2 groups in terms
of presenting symptoms (aural and neural), T-stage, and radiation exposure to cochlea.
The cochlea dosage threshold, or cut-off value, of 44 Gy was calculated using the
maximum AUC of 0.659. The results of the uni and multivariate logistic regression
analyses are depicted in [Table 6 ]. In the univariate analysis, T-stages 3 and 4 and mean radiation exposure (Dmean ) > 44 Gy showed ORs of 2.95 and 1.80, respectively. The final NTCP model was established
utilizing T-stage and cochlea dose with an AUC of 0.658 (precision of 66.1%). The
calculated risk of developing MEE ranged between 20.42 and 51.99%, as illustrated
in [Table 7 ].
Table 5
Comparison between the no MEE and the new MEE groups
Analyzed factors
Group I: no MEE post-RT (N = 123)
Group II: new MEE post-RT (N = 92)
p -value
Gender: male – n (%)
77 (62.6)
61 (66.3)
0.58
Age in years (IQR)
50 (40–59)
52.5 (41.5–58)
0.55
Age groups: n (%)
• < 50 years
• ≥ 50 years
49 (39.8)
74 (60.2)
50 (54.3)
42 (45.7)
0.72
Symptoms at presentation: n (%)
Aural
46 (37.4)
47 (51.1)
0.04*
Nasal
67 (54.5)
48 (52.2)
0.74
Mass
51 (41.5)
48 (52.2)
0.12
Neural
8 (6.5)
16 (17.4)
0.01*
Staging: n (%)
T3–4
45 (36.6)
53 (57.6)
0.002*
N2–3
62 (50.4)
49 (53.3)
0.68
M1
6 (4.9)
1 (1.1)
0.12
Treatment protocol
Dmax : GTV-NP (IQR) in cGy
7,458 (7,294–7,540)
7,411.5 (7,302.5–7,577.5)
0.92
Dmax : CTV-NP (IQR) in cGy
7,272 (7,143–7,438)
7,293.5 (7,165.5–7,453.5)
0.45
Dmax : PTV-70 (IQR) in cGy
7,190 (6,982–7,357)
7,234 (7,007.5–7,341)
0.42
Using chemotherapy, n (%)
113 (91.9)
89 (96.7)
0.55
Analyzed factors
Group I: no MEE post RT, ears (
N
= 123)
Group II: new MEE post RT, ear (
N
= 92)
p
-value
Dmean : cochlear dose (IQR) in cGy
4,255 (3,591–5,314)
4,601 (3,765–5,687)
< 0.001*
Abbreviations: CTV, clinical target volume; GTV, gross tumor volume; IQR, interquartile
range; MEE, middle ear effusion; M, metastasis; N, node; NP, nasopharynx; PTV-70,
planning target volume receiving 70 Gy; RT, radiotherapy; T, tumor.
Note: Bold values represent p -value < 0.05.
Table 6
Uniand multivariate analyses of middle ear effusion
Analyzed factors
Univariate
Multivariate
Factors
Reference
OR (95%CI)
p -value
aOR (95%CI)
p -value
Male
Female
1.05 (0.67–1.67)
0.80
Age group ≥ 50 years
< 50
1.09 (0.71–1.69)
0.69
T3–4
0–2
2.95 (1.88–4.63)
< 0.001*
2.76 (1.75–4.35)
< 0.001*
N2–3
0–1
1.04 (0.67–1.61)
0.86
M1
0
0.21 (0.03–1.66)
0.14
Cochlear dose (cGy)
< 4,400
1.80 (1.16–2.81)
0.01*
1.54 (0.98–2.43)
0.06
Abbreviations : 95%CI, 95% confidence interval; aOR, adjusted odds ratio; OR, odds ratio; M, metastasis;
N, node; T, tumor.
S(x) = −1.36 + [1.01“if T stage 3-4] + [0.43 “if cochlear dose ≥ 4,400”]
Table 7
Normal tissue complication probability model and calculated risk factor for middle
ear effusion
T-stages 3 and 4
Cochlear dose > 44 Gy
Modifying factor S (x)
Risk factor for MEE (%)
Yes
Yes
0.08
51.99
Yes
No
−0.35
41.33
No
Yes
−0.93
28.29
No
No
−1.36
20.42
Abbreviation : MEE, middle ear effusion; T, tumor.
Discussion
It is acknowledged that hearing loss is a common adverse effect following head and
neck RT, particularly when the inner ear is included in the irradiated field and the
radiation dose is high.[25 ] Hearing loss is progressive and irreversible,[14 ] often occurring at least 3 months after the last treatment. Typically, the ability
to hear high-frequency sounds is affected first, followed by the ability to hear low-frequency
sounds. The reason remains unknown. However, patients are more likely to report hearing
difficulties even if only their ability to hear low-frequency sounds is affected,
as this is the frequency of speech.[26 ]
The treatment intensity of the photon or X-ray beam ranged from 8 to 18 megavoltage
(MV), applied directly to the tumor, and decreased as it traveled to the tumor's periphery.
Despite the efforts to avoid essential organs, such as the inner ear and brainstem,
the amount of radiation reaching this undesirable area can be between 1.8 and 2.0 Gy
for each exposure.[27 ] Theories that explain how radiation affects the inner ear are described as follows:
Direct damage to the deoxyribonucleic acid (DNA) in the mitochondria of the inner
hair cells and indirect damage from the formation of reactive oxygen species and reactive
nitrogen species, resulting in DNA breakage. This occurs 1 hour after irradiation.[28 ]
The activation of proinflammatory cytokines produced by macrophages, such as tumor
necrosis factor (TNF) α, IL-1, IL-6, and IL-8, causes mitotic arrest or leads to hair
cell apoptosis via the p53-dependent or independent pathways. This occurs 6 hours
after irradiation.[29 ]
Injury to the endothelial cells of the stria vascularis impairs the K+ recycling channel,
rendering it unable to maintain the endolymphatic membrane potential. This also happens
to the myelin nerve sheath and connective tissue cells of the auditory nerve., and
it occurs 24 to 72 hours after irradiation.[30 ]
Both regenerative (epithelial resting and parenchymal cells) and non-regenerative
cells (auditory hair and spiral ganglion cells) have lost some of their proliferative
reverse capacity. This occurs 7 to 14 days after irradiation.[31 ]
In our study, age > 50 years, T-stages 3 and 4, and radiation dose to the cochlea
of > 43 Gy were associated with an increased risk of hearing loss in the univariate
analysis, which was concordant with the results of previous studies.[22 ]
[23 ] According to our NTCP model, a maximum hearing loss risk was predicted to be 44.52%
when both risk variables were present, and a minimum risk of 15.84% when neither risk
factor was present. It is not surprising that a patient has some risk of hearing loss
even with low-dose RT to the cochlea due to the well-documented dose–effect relationship
for radiation damage to organized tissues.[32 ] Additionally, the aging process also had a major influence due to a decline in cellular
repair capacity.[33 ]
In our study, T-stages 3 and 4 and doses of radiation to the cochlea > 44 Gy were
factors related to MEE. If both factors were present, the NTCP model indicated a maximum
risk of MEE of 51.99%, whereas their absence was associated with a minimum risk of
20.42%. However, MEE was shown to have lower prevalence when compared with hearing
loss due to its transient nature and tendency to fluctuate over time. Most of the
patients had MEE prior to treatment because the disease itself obstructed the Eustachian
tube.[34 ] During the follow-up period, incomplete physical examinations and medical records
were found, resulting in the exclusion of a substantial number of patients with MEE
from the study.
The present study has some limitations. First, it is possible that hearing loss detection
during the follow-up period was underestimated. Prior to the establishment of the
recommended guidelines in 2012, a follow-up audiogram was not indicated unless the
patient reported hearing problems. Second, the result may not be possible to generalize
to certain patient categories, such as stage-I patients who solely had definitive
RT without concurrent chemotherapy, and patients with distant metastases. Third, our
model was developed based on IMRT or VMAT patients; thus, its use in proton therapy
patients might be limited. Future prospective multicenter studies are needed for external
validation of the model and to increase the generalizability of the study's findings.
Conclusion
The main risk factors associated with the occurrence of hearing loss and MEE were
age, radiation dose to the cochlea, and tumor stage. Our model is useful in determining
the risk factors, hence facilitating treatment decision-making.
Bibliographical Record Prem Wungcharoen, Anussara Prayongrat, Napadon Tangjaturonrasme. Hearing Loss and
Middle Ear Effusion in Nasopharyngeal Carcinoma Following Radiotherapy: Dose–Response
Relationship and Normal Tissue Complication Probability Modeling
[* ]
. Int Arch Otorhinolaryngol 2025; 29: s00451805045.
DOI: 10.1055/s-0045-1805045
