Keywords
sinonasal malignancies - radiation side effects - radiation-induced hypopituitarism
- hypopituitarism screening - quality of life - Anterior Skull Base Questionnaire
Introduction
Radiotherapy is a mainstay of definitive or adjuvant treatment for anterior skull
base tumors. Radiation-induced hypopituitarism (RIH) has long been recognized as one
of the deleterious side effects of skull base radiation. RIH can present within the
first few years to as late as 20 years after radiation.[1] RIH not only occurs in patients who received direct sellar radiation but also to
patients with nearby tumors, as the hypothalamic-pituitary unit commonly falls within
the radiation field.[1] These tumors include cranial tumors not involving the hypothalamic-pituitary-axis
(HPA) and even oropharyngeal tumors whose radiation field include the sella.[2]
[3] Higher radiation doses (greater than 50 Gy), as commonly seen for anterior skull
base malignancies and nasopharyngeal carcinoma, are associated with higher incidence
of HPA dysfunction.[1]
[4]
[5]
[6] The estimated cumulative incidence of RIH after high-dose radiation is broad, with
estimates ranging from 37% over 3 years among our own institutional cohort of sinonasal
malignancies,[5] to 41% at 6 years among adult survivors of nonpituitary brain tumors with radiation
exposure,[7] and 82[8] to 93%[9] for long-term survivors (greater than 10 years) of nasopharyngeal carcinoma. There
is a well-characterized dose-dependent sequence of pituitary derangements, with the
somatotrophs more likely to be affected first at even low radiation doses; gonadal,
adrenal, and thyroid axes are more likely to become dysfunctional with higher radiation
doses and with increased time out from radiation.[4]
[10] A more recent publication[11] found a dose threshold for endocrine dysfunction at 30 Gy in their skull base tumor
case–control study, finding that thyroid-stimulating hormone (TSH) and adrenocorticotropic
hormone deficiencies are rare in those receiving less than 50 Gy.
As cancer treatment paradigms have changed for certain sinonasal malignancies, with
an increase in chemoselective and chemoradiation protocols, it is imperative to continue
to measure and evaluate quality of life (QoL), as well as potential known side effects
of treatment. QoL is a multidimensional concept consisting of many different factors,
including mental, social, and emotional well-being, financial burden, pain control,
physical functioning, and cosmesis.[12] There are a variety of different QoL assessment tools. These tools can assess QoL
generally or be disease-specific, allowing a more targeted analysis of specific aspects
pertaining to that disease. The Anterior Skull Base Questionnaire (ASBQ), developed
and validated by Gil et al,[13]
[14] is a patient-reported, disease-specific QoL survey for benign and malignant pathologies.
Radiotherapy has been shown to be an independent negative prognostic factor for QoL
among head and neck cancer patients using multiple disease-specific QoL assessment
tools,[2]
[15] including the ASBQ (specifically in the physical symptoms and impact on emotions
domains).[13]
[14]
[16] Löfdahl et al[2] demonstrated this decreased QoL even in patients without hypopituitarism compared
with healthy controls without radiation, most likely due to nonendocrine side effects
of radiation. Other factors that have been shown to decrease QoL among skull base
surgical patients in at least one domain of the ASBQ are malignancy, older age, female
gender, repeat surgeries, wider resections, nasoseptal flap reconstruction, and classic
surgical approach (compared with combined subcranial approach),[12]
[13]
[14]
[16]
[17] although these covariates may not be predictive for those patients whose surgeries
were done endoscopically.[12]
[16]
[18]
[19]
Hypopituitarism has been associated with decreased QoL, especially if the patient
is also suffering from depression, has negative illness perceptions, and/or lives
in a rural community.[20] Conversations around QoL specifically among patients with RIH have focused on the
deleterious effects of growth hormone deficiency (GHD), and how QoL can be improved
with growth hormone replacement.[20]
[21] It has more recently been speculated that this decrease in QoL from GHD is more
related to supraphysiological treatment with hydrocortisone at doses above 20 mg per
day rather than GHD itself.[15] There have been few studies on the effect of RIH beyond GHD apart from those with
pituitary adenomas and craniopharyngiomas.[20]
[22] To our knowledge, there have not been any investigations into QoL of radiated sinonasal
cancer survivors who develop hypopituitarism compared directly to those who do not
develop hypopituitarism. Furthermore, there have been no formal studies into the effect
of hormonal replacement, outside of growth hormone replacement, on QoL. Untreated
central adrenal insufficiency, hypothyroidism, and hypogonadism can cause severe fatigue,
sexual dysfunction, negative body image perception, and reduced bone density,[6] among other symptoms, all of which can lead to decreased QoL, in surplus of the
already decreased baseline QoL in those with skull base radiation. Given this known
clinical sequalae, this study aimed to formally evaluate QoL among RIH patients compared
with non-hypopituitary patients and the effect of treatment.
Materials and Methods
Data collection: A retrospective review of patients prospectively enrolled in an Institutional Review
Board (IRB)-approved anterior skull base registry (IRB HUM003673) at a tertiary care
center between the years 2011 and 2020 was performed. The anterior skull base registry
is hosted on our institution's REDCap[23] platform. Included patients were those with a history of anterior skull base tumor,
underwent at least one round of radiation to the primary site in the skull base, and
had filled out at least one ASBQ at some point after their radiation treatment. Radiation
therapy could include either intensity-modulated radiation therapy (IMRT), photon
therapy, or stereotactic radiosurgery (e.g., SRS or gamma knife). Subjects with pituitary
adenomas and without pituitary serologies, unless they had a known diagnosis in their
chart relating to their pituitary function, were excluded. A total of 145 patients
met inclusion and exclusion criteria. Comprehensive retrospective data was extracted
from the patient chart using manual review and DataDirect software,[24] including tumor type, date of initial radiation, dates of subsequent biochemical
and clinical follow-up, age, gender, type of skull base radiation, radiation dose,
history of systemic antitumor therapies (chemotherapy, immunologics, or hormonal therapies),
length of follow-up, and death.
Evaluation of QoL: Our primary outcome was the ASBQ score. The ASBQ contains 35 questions, each on
a 5-point Likert scale, with an option of choosing 1, 2, 3, 4, or 5. Note that 1 indicates
poor QoL, whereas 5 indicates excellent QoL. Total scores thus range from 35 to 175.
There are multiple domains within the ASBQ including performance, physical function,
vitality, body pain, impact on emotions, and specific skull base symptoms. Although
only 145 patients were included in this study, most of the subjects had filled out
multiple ASBQ surveys at various time points after radiation, for a total of 330 surveys
analyzed. For the purposes of this study comparing postradiation scores, preradiation
surveys were not analyzed. For each survey the patient took, we assigned a “pituitary
function status,” which could be one of four levels: (1) never-hypopit, (2) pre-hypopit,
(3) hypopit, and (4) treated hypopit. The never-hypopit status was assigned to subjects
that did not have any evidence of pituitary dysfunction at the time of their survey
and never went on to develop pituitary dysfunction at any point in their treatment
course. Pre-hypopit status was assigned to a patient that did not have any evidence
of pituitary dysfunction at the time of the survey but went on to develop hypopituitarism
in the future. Pre-hypopit also included those subjects who ultimately developed hypopituitarism,
but it was unclear at the time of that specific survey if they had yet developed any
pituitary dysfunction based on serological tests (although clinically they had no
pituitary symptomology). The third level, “hypopit,” was assigned to subjects who
demonstrated clear biochemical evidence of pituitary dysfunction at the time of their
survey. “Treated hypopit” was assigned to subjects who were on some type of hormone
replacement for their hypopituitarism (including hydrocortisone or other formulations
of synthetic cortisol, testosterone injections or gel, estradiol therapy, or levothyroxine)
at the time of their survey. Those who were on a hormone replacement therapy but still
had serological evidence of hypopituitarism were assigned to the hypopit category.
Patients often completed their ASBQ at the time of their pituitary screening laboratories,
or during one of their cancer surveillance or endocrine follow-up visits. However,
there were some asynchronous survey dates. In these cases, pituitary function status
was assigned based on the most recent laboratory values.
Evaluation of endocrinopathy: As there has been increasing understanding of RIH over the past few decades, our
institution has strived to achieve comprehensive pituitary screening profiles annually
after completion of radiation. These screening serologies include 8 a.m. cortisol,
prolactin, growth hormone (GH), insulin-like growth factor 1, free T4, TSH, luteinizing
hormone (LH), follicle-stimulating hormone (FSH), total testosterone, and bioavailable
testosterone. Although annual screening is our gold standard, some patients opted
to receive their endocrine follow-up closer to home or were not able to come in to
draw laboratories every year. In these cases, we used the laboratory values and clinical
notes that were available to us, including those from accessible outside records.
If a patient had an abnormal serology, this often prompted a comprehensive evaluation
in our Pituitary Endocrine Clinic. For these patients, the diagnosis given via clinical
evaluation was used in our analysis. For those remaining few patients, available serologies
were reviewed in detail. To be considered “hypopit” a subject must have demonstrated
serologic evidence of at least one of the following: central hypogonadism, central
hypothyroidism, and/or secondary adrenal insufficiency. Secondary/central hypogonadism
was documented in premenopausal women by the finding of amenorrhea or oligomenorrhea
in the setting of low LH and FSH concentrations, and in postmenopausal women with
inappropriately low LH and FSH. In men, secondary hypogonadism was documented by finding
low serum bioavailable testosterone in the setting of inappropriately nonelevated
LH and FSH. Secondary hypothyroidism was defined as low free T4 concentration with
nonelevated TSH. Secondary adrenal failure was defined by a morning cortisol less
than 5 μg/dL.[25] At our institution, abnormal levels of GH are rarely treated as isolated clinical
entities, and isolated GHD was not used in our hypopituitarism criteria.
Statistical analysis: Data were tabulated in an Excel spreadsheet (Excel 2007, Microsoft, Redmond, Washington,
United States) and analyzed in SAS 9.3 (SAS Institute, Cary, North Carolina, United
States). Pooled t-test and Wilcoxon signed rank test were used for bivariate analysis comparing continuous
demographic variables by overall binary hypopituitarism status for parametric and
nonparametric variables, respectively. Chi-squared tests and simple logistic regression
were used for categorical variables. Ideally, we would compare each patient to themselves.
However, because not everyone had ASBQ scores at all time points, with multiple surveys
done per patient over time, we employed expert statistical consultation (E.B.) who
suggested and performed three different statistical models with the primary outcome
of ASBQ score. For each model, Likert scores were averaged per domain and overall.
Model 1 tested group differences using a nonparametric Wilcoxon signed rank test with
pairwise comparisons. This model takes all ASBQ scores as unique snapshots in time.
Model 2 used a linear mixed effects model to account for intraperson correlation with
a random intercept and random slope included for time since radiation. Model 3 was
an adjusted linear mixed effects model that, in addition to accounting for intraperson
correlation, adjusted for years out from radiation, age, sex, tumor pathology, radiation
dose, type of radiation, and history of systemic antitumor therapy (including chemotherapy).
Model 3 is particularly useful given our data set's lack of robust paired data (surveys
done at the pre-hypopit vs. hypopit time points for the same person) by adjusting
for variables that vary from person-to-person and likely have an impact on QoL. Statistical
significance level was set to α < 0.05.
The minimal clinically important difference (MCID) was calculated using methodology
by Norman et al[26] of half of the standard deviation of sample in question.
Results
[Table 1] demonstrates the demographic characteristics of our sample population. Overall,
145 patients were included in this study, with 35.2% (n = 51) of them developing hypopituitarism at some point in their treatment course.
This aligns with our previously published institutional cohort hypopituitarism rate
of 37%.[5] There were no significant differences between the non-hypopituitarism and hypopituitarism
groups except for in length of follow-up and overall incidence of death as of the
time of final analysis (January 2022). [Table 2] shows the types of tumor pathologies included, with the most patients having either
squamous cell carcinoma or esthesioneuroblastoma. The vast majority of patients had
malignant tumors, although we did include six with meningiomas and eight with craniopharyngiomas.
Table 1
Demographic characteristics, stratified by overall binary pituitary function status
|
Mean (SD)
Percent (count)
|
All
(n = 145)
|
Non-hypopit
(n = 94)
|
Hypopit
(n = 51)
|
p-Value
|
|
Age at time of radiation (y)
|
52.2 (14.8)
|
53.0 (15.5)
|
50.6 (13.3)
|
0.3624[a]
|
|
Sex
|
|
Male
|
59.3% (86)
|
56.4% (53)
|
64.7% (33)
|
0.3300[b]
|
|
Female
|
40.7% (59)
|
43.6% (41)
|
35.3% (18)
|
|
Type of skull base radiation
|
|
|
|
0.7643[b]
|
|
IMRT
|
89.7% (130)
|
88.3% (83)
|
92.2% (47)
|
ref
|
|
Proton
|
7.59% (11)
|
8.51% (8)
|
5.88% (3)
|
0.8692[d]
|
|
Gamma/stereotactic
|
2.76% (4)
|
3.19% (3)
|
1.96% (1)
|
0.7886[d]
|
|
Skull base total radiation dose
|
71.4 (27.6)
n = 128
|
73.0 (29.6)
n = 83
|
68.3 (23.5)
n = 45
|
0.1997[c]
|
|
Antitumor therapy
|
|
|
|
0.3467[b]
|
|
Nothing
|
44.8% (65)
|
42.6% (40)
|
49.0% (25)
|
Ref
|
|
Chemotherapy
|
49.7% (72)
|
50.0% (47)
|
49.0% (25)
|
0.3388[d]
|
|
Other (immunologic, hormonal)
|
5.52% (8)
|
7.45% (7)
|
1.96% (1)
|
0.1980[d]
|
|
Length of follow-up (y)
|
6.47 (5.29)
|
5.76 (4.99)
|
7.78 (5.61)
|
0.0098[c]
|
|
Death
|
25.5% (37)
|
34.0% (32)
|
9.8% (5)
|
0.0014[b]
|
Abbreviations: IMRT, intensity-modulated radiation therapy; SD, standard deviation.
a Pooled t-test.
b Chi-squared test.
c Wilcoxon rank-sum test.
d Simple logistic regression.
Table 2
Types of tumor pathologies included ordered by decreasing sample incidence
|
Count
|
All
(n = 145)
|
Non-hypopit (n = 94)
|
Hypopit
(n = 51)
|
|
Squamous cell carcinoma
|
32
|
26
|
6
|
|
Esthesioneuroblastoma
|
30
|
19
|
11
|
|
Sinonasal undifferentiated
|
11
|
5
|
6
|
|
Chordoma
|
10
|
8
|
2
|
|
Other sarcoma
|
10
|
7
|
3
|
|
Melanoma
|
8
|
7
|
1
|
|
Craniopharyngioma
|
8
|
0
|
8
|
|
Neuroendocrine
|
7
|
3
|
4
|
|
Adenocarcinoma
|
7
|
4
|
3
|
|
Meningioma
|
6
|
5
|
1
|
|
Adenoid cystic carcinoma
|
6
|
5
|
1
|
|
Metastasis
|
3
|
2
|
1
|
|
Chondrosarcoma
|
3
|
1
|
2
|
|
Nasopharyngeal
|
1
|
0
|
1
|
|
Plasmacytoma
|
1
|
1
|
0
|
|
Lymphoepithelioma
|
1
|
1
|
0
|
|
Acinic cell carcinoma
|
1
|
0
|
1
|
[Table 3] summarizes the results of our three statistical models looking at the primary outcome
of average ASBQ score per domain and overall. In model 1, each of the categories of
the pituitary function status was compared with each other. Models 2 and 3 compared
the never-hypopit to the three other groups (pre-hypopit, hypopit, and treated hypopit).
Never-hypopit was chosen over pre-hypopit as the reference group due to a much higher
total number of surveys in the never-hypopit category, and there were no statistical
differences between the never-hypopit and pre-hypopit groups. Across all models, there
was a statistically significant difference in overall QoL scores between never-hypopit
and hypopit. Those with hypopituitarism at the time of the survey had overall average
QoL scores ranging from 0.24 to 0.45 points (on a 5-point Likert scale) lower than
those without hypopituitarism at the time of the survey (model 1 p 0.0004 compared with never-hypopit; p 0.049 compared with pre-hypopit). Model 1 surpassed the MCID. [Fig. 1] shows the changes in overall mean QoL across the four pituitary function status
groups using model 1. The hypopit group had significantly lower scores than the non-hypopit
group in at least two of three models across every ASBQ domain except for impact on
emotions and physical functioning.
Fig. 1 ASBQ mean total score by pituitary function status using model 1. *p-value < 0.05, ‡clinically significant. ASBQ, Anterior Skull Base Questionnaire.
Table 3
Pairwise comparisons of ASBQ QoL scores between pituitary function status groups by
ASBQ domains, comparing statistical models 1 to 3
|
Model 1: Actual difference (p-value)
Model 2 and 3: Estimated difference (p-value)
|
Never-hypopit versus pre-hypopit
|
Never-hypopit versus hypopit
|
Never-hypopit versus treated hypopit
|
Pre-hypopit versus hypopit
|
Hypopit versus treated hypopit
|
|
Performance
|
|
Model 1
|
–0.168 (0.368)
|
–0.325 (0.012)ab
|
–0.202 (0.215)
|
–0.157 (0.721)
|
0.123 (0.883)
|
|
Model 2
|
–0.138 (0.139)
|
–0.230 (0.023)[a]
|
–0.132 (0.256)
|
|
Model 3
|
–0.195 (0.062)
|
–0.293 (0.009)[a]
|
–0.209 (0.100)
|
|
Physical function
|
|
Model 1
|
0.002 (0.994)
|
–0.610 (0.002)ab
|
–0.323 (0.117)[b]
|
–0.612 (0.028)ab
|
0.280 (0.727)
|
|
Model 2
|
–0.059 (0.730)
|
–0.277 (0.096)
|
–0.183 (0.282)
|
|
Model 3
|
–0.136 (0.463)
|
–0.332 (0.064)[b]
|
–0.245 (0.184)
|
|
Vitality
|
|
Model 1
|
–0.165 (0.685)
|
–0.634 (< 0.0001)ab
|
–0.312 (0.096)[b]
|
–0.469 (0.052)[b]
|
0.322 (0.388)[b]
|
|
Model 2
|
–0.166 (0.216)
|
–0.385 (0.006)ab
|
–0.255 (0.080)
|
|
Model 3
|
–0.174 (0.281)
|
–0.357 (0.012)ab
|
–0.270 (0.083)
|
|
Body pain
|
|
Model 1
|
–0.342 (0.372)[b]
|
–0.724 (0.003)ab
|
–0.482 (0.126)[b]
|
–0.381 (0.561)[b]
|
0.242 (0.767)
|
|
Model 2
|
–0.175 (0.409)
|
–0.447 (0.041)ab
|
–0.388 (0.103)[b]
|
|
Model 3
|
–0.413 (0.065)[b]
|
–0.700 (0.003)ab
|
–0.655 (0.010)ab
|
|
Impact on emotions
|
|
Model 1
|
–0.085 (0.768)
|
–0.151 (0.564)
|
–0.002 (0.979)
|
–0.066 (0.996)
|
0.149 (0.888)
|
|
Model 2
|
–0.053 (0.644)
|
–0.104 (0.332)
|
–0.071 (0.467)
|
|
Model 3
|
–0.098 (0.439)
|
–0.164 (0.175)
|
–0.100 (0.368)
|
|
Specific symptoms
|
|
Model 1
|
–0.041 (0.999)
|
–0.331 (0.021)ab
|
–0.256 (0.174)
|
–0.290 (0.115)
|
0.075 (0.683)
|
|
Model 2
|
–0.012 (0.921)
|
–0.201 (0.064)
|
–0.203 (0.037)[a]
|
|
Model 3
|
–0.077 (0.539)
|
–0.227 (0.041)[a]
|
–0.256 (0.016)[a]
|
|
Overall QoL
|
|
Model 1
|
–0.110 (0.882)
|
–0.454 (0.0004)ab
|
–0.253 (0.078)
|
–0.344 (0.049)ab
|
0.200 (0.663)
|
|
Model 2
|
–0.089 (0.377)
|
–0.240 (0.018)[a]
|
–0.177 (0.095)
|
|
Model 3
|
–0.154 (0.182)
|
–0.278 (0.012)[a]
|
–0.236 (0.046)[a]
|
|
MCID
|
0.297
|
0.314
|
0.300
|
0.317
|
0.308
|
Abbreviations: ASBQ, Anterior Skull Base Questionnaire; MCID, minimal clinically important
difference; QoL, quality of life.
Note: Negative differential values indicate worsening QoL.
a
p-Value < 0.05.
b Clinically significant.
In regard to treatment, the treated (i.e., hormone-replaced) patients in general had
higher QoL than the hypopit group ([Fig. 2]); however, this increase did not reach statistical significance, but did surpass
the MCID in the vitality domain. When comparing the treated group to the never-hypopit
controls, treated patients had overall lower scores which reached statistical significance
mostly in model 3 in the specific symptoms and pain domains, as well as overall (although
only reaching a clinically detectable difference in the pain domain), reflecting that
treatment may not fully restore baseline postradiation levels of QoL looking at aggregate
data. However, this treatment effect seems to be modulated by the amount of time that
has passed since radiation. We divided the ASBQ surveys into the following three subcategories,
0 to 2, 2 to 5, and 5+ years out from radiation, to better assess this longitudinal
effect of QoL over time. Using model 1, comparing to the hypopit group, treatment
seemed to improve QoL scores, at least clinically in the 0- to 2-year range out from
radiation (estimated 0.22 point average increase, p 0.876), decrease scores in the 2- to 5-year range (estimated 0.32 point average decrease,
p 0.572), and significantly increase scores in the 5+ year range out from radiation
(estimated 0.42 point average increase, p 0.041) ([Tables 4] and [5]). [Fig. 3] shows the boxplots comparing the QoL scores between the pituitary function statuses
across these three time-based subgroups.
Fig. 2 Comparing mean ASBQ QoL scores per domain across the pituitary function status groups
using model 1. ASBQ, Anterior Skull Base Questionnaire; QoL, quality of life.
Fig. 3 Comparisons of mean ASBQ QoL scores between pituitary function status groups across
years out from radiation using model 1. *p-value < 0.05, ‡clinically significant. ASBQ, Anterior Skull Base Questionnaire; QoL, quality of life.
Table 4
Averaged total ASBQ QoL scores per pituitary function status aggregated by years out
from radiation groups (0–2 years, 2–5 years, and 5+ years)
|
Years out from radiation
|
Pituitary function status
|
N
|
Mean
|
Standard deviation
|
Minimum
|
Maximum
|
|
0–2 y
|
Never-hypopit
|
88
|
3.38746
|
0.6109
|
1.91429
|
4.68571
|
|
Pre-hypopit
|
24
|
3.32393
|
0.44015
|
2.45714
|
4.02857
|
|
Hypopit
|
15
|
3.08825
|
0.42199
|
2.54286
|
3.82857
|
|
Treated
|
4
|
3.30714
|
0.3471
|
3
|
3.74286
|
|
2–5 y
|
Never-hypopit
|
76
|
3.43118
|
0.58777
|
2
|
4.62857
|
|
Pre-hypopit
|
10
|
3.29353
|
0.89032
|
1.77143
|
4.20588
|
|
Hypopit
|
14
|
3.19659
|
0.45146
|
2.37143
|
3.77143
|
|
Treated
|
23
|
2.97107
|
0.50629
|
1.97143
|
3.65714
|
|
5+ y
|
Never-hypopit
|
47
|
3.55913
|
0.58295
|
2.23529
|
4.6
|
|
Pre-hypopit
|
2
|
3.60882
|
0.01248
|
3.6
|
3.61765
|
|
Hypopit
|
16
|
2.71098
|
0.85008
|
1.6
|
4.31429
|
|
Treated
|
11
|
3.59901
|
0.57522
|
2.57143
|
4.42857
|
Abbreviations: ASBQ, Anterior Skull Base Questionnaire; QoL, quality of life.
Table 5
Actual differences in averaged total ASBQ QoL scores and p-values for two-sided multiple pairwise comparisons between the pituitary function
status groups aggregated by years out from radiation
|
Actual difference
(p-value[c])
|
0–2 years out from radiation
|
2–5 years out from radiation
|
5+ years out from radiation
|
|
Never-hypopit versus pre-hypopit
|
–0.0635 (0.9428)
|
–0.138 (0.9994)
|
0.050 (0.9993)
|
|
Never hypopit versus hypopit
|
–0.299 (0.1858)[b]
|
–0.235 (0.4940)
|
–0.848 (0.0045)ab
|
|
Never hypopit versus treated hypopit
|
–0.080 (0.9765)
|
–0.460 (0.0101)ab
|
0.040 (0.9999)
|
|
Pre-hypopit versus hypopit
|
–0.236 (0.3852)[b]
|
–0.097 (0.6811)
|
–0.898 (0.4962)[b]
|
|
Hypopit versus treated hypopit
|
0.219 (0.8764)[b]
|
–0.322 (0.5729)[b]
|
0.888 (0.0412)ab
|
Abbreviations: ASBQ, Anterior Skull Base Questionnaire; QoL, quality of life.
a
p-Value < 0.05.
b Clinically significant.
c Wilcoxon signed rank test.
Overall, the never-hypopit and pre-hypopit groups had similar QoL. The fully adjusted
model 3 revealed a difference in QoL between tumor pathologies across all domains,
with plasmacytoma and adenocarcinoma having consistently worse QoL scores across most
domains. On average, adenocarcinoma patients had a mean QoL score 0.685 points lower
than squamous cell carcinoma patients (p 0.0045); plasmacytoma patients had an estimated mean QoL score 0.910 points lower
than squamous cell carcinoma patients (p < 0.0001) holding all other variables constant. Interestingly, older age at the time
of radiation was a protective factor only in the impact on emotions domain, with an
increase of 0.01 points for every additional year of age (p 0.0012) holding all other variables in the model constant. There were no significant
differences in QoL between males and females except in the impact on emotions domain,
where females on average score 0.19 points lower (p 0.0288).
Discussion
These data confirm our hypothesis that patients with RIH beyond GHD have significantly
lower QoL compared with those who do not develop hypopituitarism after radiation.
This was true across most domains (except impact on emotions) and statistical models,
even among those with the same age, sex, time out from radiation, type of tumor, type
of radiation, and radiation dose (model 3). Our average ASBQ QoL score, no matter
the hypopituitarism status, was relatively high, at 3.13, corresponding to “good”
QoL, but this decreased to a greater degree than the MCID (greater than a 7.9% decrease)
when going from non-hypopit to hypopit, at least in model 1. The largest average drop
in scores comparing non-hypopit to hypopit was in the body pain domain (–0.72), which
correlates to an 18% decrease in QoL. Though not dramatic, it does surpass the MCID,
meaning patients are able to perceive a difference in their QoL. Model 3 showed a
statistically significant decrease in overall QoL, although it did not surpass the
MCID except for in the physical function, vitality, and body pain domains. This suggests
that hypopituitarism most affects functioning, vitality, and pain after controlling
for other variables that could be contributing to QoL. This aligns with the symptom
profile of overwhelming fatigue and subsequent lack of daily life functioning often
seen in patients with central hypothyroidism or secondary adrenal insufficiency. The
specific symptoms domain, which asks about specific symptoms related to skull base
patients (changes in vision, appetite, taste, smell, appearance, and secretions),
had the lowest average scores (2.70). Factors such as age and gender seemed to impact
the emotions domain more than pituitary function status among these radiated patients,
perhaps due to the development of beneficial coping mechanisms to chronic illness
as patients age.
As for the differences in the demographics between the hypopituitarism and non-hypopituitarism
groups, those in the hypopituitarism group were followed on average for 7.78 years,
compared with 5.76 years for those in the non-hypopituitarism group (p 0.0098). This can most likely be attributed to the potential higher motivation among
hypopituitarism patients to maintain endocrine follow-up given their symptoms and
need for treatment, particularly opting for follow-up at our institution's pituitary
clinic, making their records more easily accessible. This length of follow-up discrepancy
could also be explained by the higher rate of death in the non-hypopituitarism group
(34%) compared with the hypopituitarism group (9.8%) (p 0.0014). This seemingly contradicts the previous finding that hypopituitarism is
associated with a 55% increase in mortality, particularly in nonreplaced patients
with history of radiation.[27] However, in our cohort, we hypothesize that patients in the non-hypopituitarism
group were not living long enough to develop or be diagnosed with hypopituitarism.
There is a concern that this shorter follow-up time and higher incidence of death
in the non-hypopituitarism group could be artificially enhancing for more aggressive
tumors, and thus death occurred before hypopituitarism status was fully realized.
However, if this was the case, we still found that the non-hypopituitarism patients,
even if they had more aggressive tumors, still had higher QoL than the hypopituitarism
group, which further supports that hypopituitarism itself can predict lower QoL.
We found that the dose of skull base radiation (maximum dose to the primary tumor)
did not significantly differ between the non-hypopituitarism and hypopituitarism groups
(p 0.20). This seemingly contradicts the long-standing existing literature[5]
[6]
[28]
[29]
[30]
[31] that suggests a positive dose–responsive relationship between radiation dose and
incidence of hypopituitarism, especially in the three axes we analyze in this study,
although the data are mixed.[29] However, almost all of our patients received relatively high doses of radiation
(average 71.4 Gy) due to the aggressive nature of these tumors, which is well above
the thresholds discussed for developing hypopituitarism. Therefore, dose could be
playing less of a role at these high radiation levels, and instead there are other
factors contributing. Alternatively, our sample size could potentially not be large
enough to reach significance. Type of skull base radiation (IMRT vs. photon vs. stereotactic)
did not seem to affect hypopituitarism status. Although there is a theoretical advantage
to stereotactic and proton radiation compared with conventional radiotherapy, studies
have not suggested a long-term difference in hypopituitarism between these groups
in certain populations.[29]
[30]
[32]
[33]
[34] It is hypothesized that development of hypopituitarism after stereotactic radiation
may be more dependent on tumor volume rather than radiation dosage.[35] Also, the vast majority (130, 90%) of patients underwent IMRT in this study, with
only 15 who underwent either proton or stereotactic radiation therapy.
Treatment improved patients' QoL across all domains, though only surpassing the clinically
noticeable difference in the vitality domain when analyzing all time points. Treatment
often improves energy levels and thus the perception of vitality. However, these improvements
in QoL did not reach a statistically significant level unless looking only at those
patients greater than 5 years out from radiation. We hypothesize that these patients
who were farther out from their radiation were not being recognized early for having
hypopituitarism, and thus were not treated early on. They felt the full symptomatic
burden of hypopituitarism without replacement (as evidenced by the large differential
scores between never-hypopit and hypopit among these long-term survivors), and thus
felt much better once treated. Whereas those who were recognized earlier were able
to be treated quickly before dramatic symptoms developed. We hypothesize that those
in the 2- to 5-year range may have experienced worsening QoL after treatment due to
side effects of medications outweighing the perceived benefits of more subclinical
hypopituitarism, although more research needs to be done to investigate this theory.
This highlights the importance of repeated and consistent serological testing in these
skull base radiation patients so that replacement can occur if needed at appropriate
doses. Though the patient may not perceive an immediate improvement in QoL with hormonal
replacement early on, early detection and appropriate replacement serves to prevent
the significant potential decline in QoL if you wait greater than 5 years out from
radiation (a 22% average drop). Those greater than 5 years out from radiation, when
treated, on average were able to regain their QoL to at or above their non-hypopituitarism
baseline after radiation. Currently, there is great variability in the follow-up of
skull base malignancy patients after radiation, including timing and frequency of
screening serologies and imaging. The National Comprehensive Cancer Network guidelines
do not include comprehensive pituitary serologies in their follow-up recommendations
for head and neck cancers after radiation except for TSH every 6 to 12 months after
radiation.[36] This screening with serum TSH alone could mislead the clinician to believe the patient
is euthyroid rather than central hypothyroid and thus in need of replacement. Similar
confusion comes from the possible misinterpretation of gonadal dysfunction. Many of
these patients also receive chemotherapy, which is known to reduce gonadal function,
and therefore can hide a secondary diagnosis of radiation-induced hypogonadism.[37] Ideally, we need specific guidelines for pituitary screening serologies on irradiated
sinonasal malignancies and skull base tumors.
There are certain limitations to our study. First, we had a lack of data. Although
we aimed to have an annual comprehensive pituitary serology and ASBQ survey on each
patient, this was not always possible. The realities of the coronavirus disease 2019
pandemic limited patient interactions, including laboratory draws and survey administration.
Some patients would elect to get their hypopituitarism follow-up from an endocrinologist
or primary care physician closer to home and would only come to our institution for
cancer surveillance. Therefore, it was sometimes unclear what their pituitary function
status was on the exact date of their ASBQ, or whether they had confirmatory gold
standard testing for specific pituitary deficiencies. We tried to mitigate this by
looking at outside records. In addition, model 3 analyzed a reduced number of surveys
(ranging from n = 294–299) due to lack of certain demographic or clinical data that were adjusted
for in the model. Second, given the inconsistency of ASBQ administration, we did not
require patients to have a baseline ASBQ before radiation, since the effect of radiation
on QoL was not within the scope of our study, and we did not feel it would be scientifically
sound to ask patients to retrospectively recall their QoL. We also did not require,
again due to lack of data, every hypopituitarism patient to have both a pre- and post-hypopituitarism
ASBQ to directly compare. Only 16 patients in our data set had at least one survey
taken after radiation when they were pre-hypopit and again when hypopit. Among these
patients, a paired Wilcoxon signed rank test with continuity correction showed no
significant overall difference between pre-hypopit and hypopit with an estimated difference
of 0.084 (95% confidence interval = –0.113, 0.177, p 0.229). However, this interpretation using paired data are limited given the low
number of patients precluding a robust statistical analysis. Alternatively, we were
able to include more patients by extrapolating data from the entire sample population
through statistical models (namely models 2 and 3) that accounted for person-to person
variability in QoL via intraperson correlation between the multiple ASBQ surveys that
each individual did have, even if they did not have surveys specifically at the pre-/post-hypopituitarism
or pre-/posttreatment time points. Furthermore, model 3 attempted to adjust for person-to-person
variability of key demographic and tumor-specific variables finding that QoL scores
among those with the same age, sex, years out from radiation, tumor pathology, radiation
type, radiation dose, and chemotherapy status but had hypopit were lower than the
QoL among those without hypopituitarism. Third, almost all patients in the anterior
skull base registry had one or multiple surgeries before radiation, so there is a
possibility that their hypopituitarism could have been due to direct surgical damage
as opposed to radiation-induced damage. Regardless of etiology, the trend still holds
for decreased QoL among hypopituitary patients after radiation compared with non-hypopituitary
patients after radiation. Hypopituitarism seems to most impact the vitality and body
pain domains across all models to a clinical and statistically significant degree.
We acknowledge the need for more studies that can prospectively and comprehensively
collect a longitudinal data set to examine intraperson correlation directly or be
able to better match non-hypopituitary controls to hypopituitary cases. Although a
preliminary study, the advantages were that this was a longitudinal study over the
course of 9 years with multiple data points per individual. We had a relatively large
sample size given the rarity of these tumors.
Conclusion
This is the first study to formally confirm the hypothesis that RIH causes decreased
QoL among skull base and sinonasal tumor patients. The data supports that awareness,
early detection of hypopituitarism via consistent and comprehensive pituitary serologies,
and appropriate medical treatment and psychosocial interventions can prevent longer
term detrimental impact on QoL.
