Keywords world trade center responders - mild cognitive impairment - PET/MRI - β-amyloid -
tau - neurodegeneration
Introduction
The World Trade Center (WTC) responders endured extreme emotional and neurotoxic insults
following the terrorist attacks on September 11, 2001. Ongoing work has identified
increased incidence of multiple physical and psychiatric diseases such as posttraumatic
stress disorder (PTSD) and increased incidence of mild cognitive impairment (MCI).[1 ] Prior studies investigating the chemical composition of the dust cloud at ground
zero revealed that it contained known neurotoxins[2 ]
[3 ]
[4 ] which could potentially have contributed to the observed early-onset MCI through
an underlying neurodegenerative dementia pathological process. Indeed, recent evidence
has demonstrated that WTC responders display cortical thinning,[5 ] evidence of Alzheimer's disease (AD) peripheral biomarkers,[6 ] and that posttraumatic stress disorder (PTSD) among responders may contribute to
neurocognitive dysfunction.[7 ] These observations have raised concerns that a disproportionate number of responders
presenting with MCI may develop neurodegenerative cerebral pathology. The nature of
this medical uncertainty has prompted investigations into identifying the underlying
etiology for the observed risk and to which neurodegenerative dementia subgroup they
might subscribe to.
The most common causes of neurodegenerative dementia is AD, whose neuropathological
cascade is the common cause of MCI and as is clear in the new research ATN framework,[8 ]
[9 ] is characterized by brain infiltration of two amyloid-β (Aβ) isoforms; Aβ1–42 , and Aβ1–40 , leading to Aβ deposition and the formation of two different types of Aβ plaques
(dense-core and diffuse plaques [A]), resulting in increased phosphorylation of tau
protein and accumulation of neurofibrillary tangles (T), followed by irreversible
neurodegeneration (N). Therefore, in our first investigation as to which neurodegenerative
dementia subgroup WTC responders with early-onset MCI may subscribe to, we examined
PET/MRI markers of ATN as a small case series of 12 responders. The uptake, distribution,
and retention of two PET ligands, that is, [18 F] florbetaben (FBB) which is used to detect the presence of Aβ amyloid fibrils (A),
along with [18 F] flortaucipir (FTP) which is used to detect the presence of tauopathy (T), and structural
brain MRI which can be used to measure cortical volume and loss of which indicates
the presence of cerebral atrophy due to neurodegeneration (N), were employed for the
current study. This is the first study to report preliminary data for ATN neuroimaging
biomarkers of AD in a small subset of WTC responders presenting with early-onset MCI.
Materials and Methods
Setting and Participants
The Stony Brook University (SBU), together with Centers for Disease Control and Prevention
(CDC), has monitored WTC responders since July 2002. Using the Montreal Cognitive
Assessment (MoCA) as a cognitive screening tool,[1 ]
[10 ] 12 WTC responders at midlife diagnosed with MCI (MoCA<23) or CI (MoCA<20) were recruited
into two neuroimaging studies utilizing the same Siemens mMR 3T positron emission
tomography (PET)/magnetic resonance imaging (MRI) scanner (Siemens Healthcare, Erlangen,
Germany). The MoCA is a widely used measure of cognitive impairment developed to identify
age-related CI objectively and reliably.[11 ] Six responders were recruited at the SBU site for PET FTP tau scans with MCI and
six separate responders were recruited at Icahn School of Medicine at Mount Sinai
(ISMMS) for PET FBB Aβ scans with CI. All 12 responders underwent the same MRI sequences
for determining gray matter volumes. Inclusion criteria were as follows: (1) 45 to
65 years of age; (2) MoCA ≤ 23 within 3 months of scan; (3) body mass index (BMI)
≤ 40kg/m2 . Exclusion criteria were as follows: (1) history of psychosis, (2) drinking/substance
abuse, (3) stroke, (4) head trauma, (5) epilepsy, (6) brain tumor, (7) renal failure/dialysis,
(8) liver disease/hepatitis, (9) diabetes, (10) major depressive disorder, (11) heart
failure, (12) metal implants, (13) claustrophobia, (14) current pregnancy or breastfeeding,
(15) current anticoagulant medications, and (16) neurotropic medications (antidepressants,
antipsychotics, antiparkinsonian drugs, etc.).
Magnetic Resonance Imaging Acquisition and Cortical Volumes of Interest
The Siemens mMR scanner was used at both sites to acquire structural brain MRI sequences
during the PET scan, for anatomical delineation including dedicated T1-weighted 3D
MPRAGE (TR/TE/TI=1,900/2.5/900ms, matrix size=256×256, voxel size=0.87×0.87×0.87mm3 ). Cortical volumes of interest (VOI's) investigation included frontal and temporal
cortices, brain regions typically associated with cognitive domains such as learning,
memory, executive functions, and information processing speed[12 ]
[13 ]
[14 ]
[15 ]
[16 ]; limbic system (including the hippocampus, amygdala, parahippocampal regions including
the entorhinal cortex), brain regions typically associated with processing and consolidation
of traumatic, and stressful memories[17 ]
[18 ]; and the occipital lobes, brain regions typically associated with reexperiencing
of traumatic memories through mental imagery, states of depersonalization, and dissociation.[19 ]
[20 ]
[21 ]
Magnetic Resonance Imaging Neurodegeneration Analysis
Quantitative volumetric analysis of cortical gray matter volumes was performed utilizing
the T1 MPRAGE scans and NeuroQuant (CorTechs Labs, Inc, San Diego, CA), a Food and
Drug Administration (FDA) approved software package for automatic labeling, visualization,
and volumetric quantification of structural brain VOI's (as percentage of total intracranial
volume) which then compares them to an age- and sex-matched normative cohort, generating
a brain atrophy report of the responders' volumetric percentiles as compared with
norms.[22 ] We compiled bilateral volumetric percentile data from the 12 study participants
and set the cut-off for identifying VOI neurodegeneration at the 5th percentile for
age, as per NeuroQuant guidelines.[23 ]
Positron Emission Tomography Image Acquisition
Six responders with CI received FBB (5–8 millicurie) and six others with MCI received
FTP (3–5 millicurie) followed by a list-mode acquisition during concomitant MRI scans.
Sinograms were binned covering a period of 20minutes starting 90 (FBB) or 80minutes
(FTP) of postinjection according to established protocols.[24 ]
[25 ] Images were reconstructed using ordered subset expectation maximization (OSEM) algorithm
with point spread function modeling, six iterations and 21 subsets, and 3-mm full-width
half-maximum (FWHM) Gaussian postfiltering. Quantitative corrections included attenuation,
scatter, random, detector efficiency, decay, and deadtime. Acquired PET data were
scaled to injected dose and body weight to produce standardized uptake value (SUV)
images. Attenuation correction was estimated with a “pseudo-CT” approach customized
for brain imaging[26 ] which is based on T1-MPRAGE sequence and accounts for skull attenuation.
Positron Emission Tomography Data Analysis
For each dataset, cortical reconstruction and segmentation of the associated T1-MPRAGE
images was performed using FreeSurfer[27 ] (V.6.0.0) followed by regional parcellation according to the Desikan–Killiany brain
atlas. Using the PETSurfer tools within FreeSurfer,[28 ]
[29 ] we extracted the VOI data from the PET images which also allowed for partial volume
correction (PVC) of the VOIs using the Symmetric Geometric Transfer Matrix method.
For FBB data, we implemented previously described protocols and cut-offs.[30 ] We constructed SUV ratios (SUVRs) for four regions (frontal cortex [0.93], lateral
temporal cortex [0.93], parietal cortex [0.98], and posterior cingulate cortex [1.10]),
as well as 0.96 for a composite SUVR (cSUVR) which was the mean of frontal, occipital,
parietal, lateral temporal, anterior, and posterior cingulate cortex SUVRs. PVC was
not used for these SUVRs, whole cerebellum was the reference region, and their published
cut-offs for AD were used.[30 ] Finally, we performed centiloid (CL) analysis of FBB data in which we used PMOD
software (PMOD Technologies, Zurich, Switzerland) instead of FreeSurfer and followed
the protocol for this tracer to calculate a CL value for each responder where CL values
can fall on a 0 to 100 scale and where a CL value of over 25 can be used as a cut-off
for AD.[31 ] Briefly, the CL method provides a universal scaling metric that can account for
differences involving quantitative values obtained from using different tracers for
Aβ, thus permitting for the integration and interpretation of various clinical and
research studies with a standard measurement. The calculation to derive CL in this
study was performed as previously described[31 ] with the following formula: 153.4×SUVRFBB −154.9.
For FTP data, we implemented previously described protocols and cut-offs[25 ] in which the cSUVR was defined as the mean of entorhinal cortex, amygdala, lateral
occipital cortex, and inferior temporal gyrus, and used the established cut-off of
1.22 to identify preclinical AD. Per that protocol, PVC data were used, and cerebellar
cortex was the reference region. We also estimated Braak's staging of the FTP data
(PVC used and cerebellar cortex as reference) as previously described.[32 ] Briefly, Braak's staging is classified as the buildup of neurofibrillary tangles
with neurodegeneration in stages I and II (transentorhinal region), stages III and
IV (limbic regions, including the hippocampus), and stages V and VI (neocortex). We
classified each responder into Braak's staging by calculating the volume-weighted
average Z -score of the composite regions corresponding to each image-based tau stage, with
values greater than 2.5 considered as positive in each respective stage.
Visual Image Interpretation and Qualitative Assessment
Image interpretation was conducted by a fellowship-trained, board-certified neuroradiologist
with dedicated PET/MRI training and 8 years clinical and research experience in the
field. The neuroradiologist reader interpreted the brain MRI and FBB or FTP PET images
independently, followed by adjudication of PET images with a board-certified nuclear
medicine physician with 25 years of clinical and research experience in brain PET
imaging. Qualitative assessment for scan positivity was visually assessed following
post-processing with MIMneuro (V.6.9.5; MIM Software, Inc. Cleveland, Ohio, United
States).
Statistical Methods
Means with standard deviations (SDs) across responders were calculated. Box and whisker
plots depict median and interquartile ranges (IQR) for VOI's as volumetric percentiles
with a 5th percentile cut-off line representing neurodegeneration. Box and whisker
plots depict median and interquartile ranges (IQR) for VOI FBB and FTP SUVR activity.
Analyses were performed in GraphPad Prism [V.9] GraphPad Software, San Diego, California,
United States).
Results
Participants
Twelve WTC responders' mean age was 54.9 (±5.4) years, mean education was 14.4 (±1.8)
years, mean BMI was 30.3 (±4.0) kg/m2 , 75% were male, mean MoCA was 18.9 (±2.4), and they were tested between March 2018
and April 2019 as part of a larger parent study conducted jointly by SBU and ISMMS.
Their occupations during WTC search and rescue efforts included New York Police Department
(NYPD), mechanics, technicians, private business owners, and construction specialists.
Magnetic Resonance Imaging
MRI volumetric analyses with NeuroQuant age- and sex-matched normative parcellation
displayed composite mean gray matter volumetric percentiles for the occipital, temporal,
inferior temporal, and entorhinal cortices and temporal pole that were below the 5th
percentile, as shown in [Fig. 1 ]. Specifically, the percentage of 12 responders displaying regional volumes below
the 5th percentile was as follows: 50% frontal lobe, 92% temporal lobe, 67% temporal
pole, 75% entorhinal cortex, and 92% inferior temporal cortex. The hippocampus displayed
the least volumetric reduction across these 12 cases.
Fig. 1 Box and whisker plot showing age/sex/cranial volume-matched percentiles against a
normative cohort (NeuroQuant) in each cortical volume regions of interest (VOI) for
12 World Trade Center (WTC) responders with cognitive impairment (CI). Cut-offs for
neurodegeneration at the 5th percentile are indicated by a red shaded area.
Florbetaben Positron Emission Tomography
Out of the six WTC responders with CI who received the FBB tracer during their PET/MRI
scan, three responders displayed visually excess FBB activity on qualitative assessment.
Quantification of FBB SUVRs revealed that among the three responders visually assessed
as positive, one had a positive CL value, with cSUVR and individual SUVR positivity
(i.e., above published threshold[30 ]) in all four VOIs (frontal cortex, lateral temporal cortex, parietal cortex, and
posterior cingulate cortex). The remaining two responders who were visually assessed
as positive had below cut-off CL values and cSUVRs, except for one responder who displayed
above cut-off SUVR in the frontal cortex. The three responders qualitatively read
as negative were quantitatively below cut-offs, except for one responder who quantified
above SUVR cut-off in the posterior cingulate cortex and another responder who quantified
above SUVR cut-off for the lateral temporal cortex. Combined FBB SUVR activity for
each VOI is shown in [Fig. 2 ].
Fig. 2 Box and whiskers plot showing standard uptake volume ratio (SUVR) retention for (18 F)-florbetaben (FBB) for each cortical volume of interest (VOI) in six WTC responders
with mild cognitive impairment at midlife. WTC, World Trade Center.
Flortaucipir Positron Emission Tomography
Six separate WTC responders with MCI received the FTP tracer during their PET/MRI
scan and three responders displayed visually excess FTP retention on qualitative assessment.
Quantification of FTP SUVRs revealed two out of the three responders visually assessed
as positive with above cutoff cSUVRs with the third slightly subthreshold at 1.21.
Of the three responders qualitatively read as negative, one responder displayed a
positive cSUVR. Combined FTP SUVR activity for each VOI is shown in [Fig. 3 ]. In addition, an example of a qualitatively assessed positive and negative PET/MRI
FTP scan is shown in [Fig. 4 ].
Fig. 3 Box and whiskers plot showing standard uptake volume ratio (SUVR) retention for (18 F)-flortaucipir (FTP) for each cortical volume of interest (VOI) in six WTC responders
with mild cognitive impairment at midlife. WTC, World Trade Center.
Fig. 4 Standard PET/MRI images showing axial and coronal cortical distribution of: (18 F) flortaucipir (FTP) in two World Trade Center (WTC) responder with MCI (first panel
shows FTP positive responder and second panel shows FTP negative responder; (18 F) florbetaben (FBB) in two WTC responders with MCI (third panel shows FBB positive
responder and fourth panel shows FBB negative responder). MCI, mild cognitive impairment;
MRI, magnetic resonance imaging; PET, positron emission tomography.
Results from the above analyses in all 12 responders are summarized in [Table 1 ].
Table 1
Summary table of findings from six WTC responders receiving FBB and six separate WTC
responders with receiving FTP
Characteristic
FBB1
FBB2
FBB3
FBB4
FBB5
FBB6
FTP1
FTP2
FTP3
FTP4
FTP5
FTP6
Age (y)
50
52
53
64
62
46
53
57
50
55
56
61
MoCA (out of 30)
18
17
15
15
20
18
21
21
21
21
22
18
PTSD status
−
−
−
−
+
−
−
−
−
+
−
−
WTC exposure duration (d)
11
> 30
288
148
289
94
28
1
19
92
168
155
Family history of dementia/CI
−
−
−
+
−
−
−
−
+
−
+
−
Qualitative/clinical read
+
−
+
+
−
−
+
+
−
+
−
−
Whole brain SUVRFBB /SUVRFTP
0.97
1.04
0.95
1.29
1.02
0.96
1.24
1.1
1.22
1.2
1.26
1.21
cSUVRFBB (0.96)/cSUVRFTP (1.22)
0.87
0.92
0.93
1.1
0.93
0.87
1.23
1.08
1.22
1.23
1.26
1.18
CL/Braak's stage
−6.5
4.1
−8.6
42.7
2
−7.8
I and II
−
I and II
−
I and II
−
Neuroquant ROIs<5th percentile
6
6
22
31
43
14
34
10
15
31
26
23
Abbreviations: CI, cognitive impairment; CL, centiloid; cSUVR, composite standardized
uptake value ratio; FBB, florbetaben; FTP, flortaucipir; MoCA, Montreal Cognitive
Assessment; ROI, region of interest; PTSD, posttraumatic stress disorder; WTC, World
Trade Center.
Discussion
WTC responders are presenting with early-onset neurocognitive dysfunction with an
undetermined etiology. To address this, our first investigations utilized a simultaneous
PET/MRI brain scan, employing PET tracers for either Aβ-type cerebral amyloidosis
(FBB) or tauopathy (FTP) and MRI for gray matter volume analysis of neurodegeneration
to examine evidence for ATN biomarkers of AD in a small sample of WTC responders presenting
with CI, since AD is the most common neurodegenerative dementia.
Analyzing MRI data in these 12 WTC responders and matching it to age- and sex-matched
normative data revealed lower gray matter volumes across many cortical regions. Specifically,
neurodegeneration was evident in the temporal lobe, entorhinal cortex, and inferior
temporal gyrus regions that are essential to memory formation and recall.[12 ] In addition, half of these 12 WTC responders had frontal lobe volumes below the
5th percentile which is a cortical area affected in advanced AD and functionally involved
in cognitive executive functions and information processing speed.[9 ] However, we observed the least neurodegeneration in the hippocampus, a feature that
may have implications for clinical prognosis. A relatively unusual hippocampal-sparing
form of AD has been reported to occur at a younger age which may apply to WTC responders
now at midlife presenting with early-onset MCI or CI, as this early-onset form of
AD is marked by a more pronounced longitudinal cognitive decline and a shorter disease
duration as well as earlier mortality.[33 ]
[34 ]
[35 ] Moreover, half of these 12 WTC responders displayed neurodegeneration and high FTP
retention in the occipital lobe which has been associated with visual hallucinations
in AD as a result of atrophy[36 ] and is a region typically affected in advanced AD.[37 ] The occipital lobes and/or occipitoparietal border zones are involved in posterior
cortical atrophy,[38 ] as well as in PTSD patients,[39 ] the latter possibly related to recurring visual “flashbacks” of mental imagery from
chronically reexperiencing traumatic events[19 ]
[20 ]
[21 ]
[40 ] which may pertain to those experienced by WTC responders at September 11, 2001.
Recent evidence has linked such PTSD states with underlying neuroinflammation,[41 ]
[42 ] leading to subsequent neurodegeneration,[36 ]
[37 ]
[39 ]
[43 ]
[44 ] thereby warranting a closer investigation and monitoring of WTC responder PTSD symptomatology
as a possible contributor to the observed neurocognitive dysfunction which was recently
suggested in one of our studies.[7 ]
Our preliminary PET results using quantitative and qualitative approaches identified
three WTC responders with excess FBB retention, supportive of the possible presence
of Aβ-type cerebral amyloidosis of which one responder had an abnormal CL in addition
to an abnormal qualitative read suggestive of AD neuropathology. However, we also
discovered multiple occurrences of excess FBB SUVR activity in regions outside of
that used in the AD CL calculation, suggesting that the neuropathological etiology
of the observed cognitive decline in WTC population may not necessarily subscribe
to AD. In addition, the observed volumetric hippocampal sparing among these 12 WTC
responders with neurocognitive dysfunction suggests the possibility that they may
subscribe to a non-AD neurodegenerative dementia subgroup. Moreover, qualitative and
quantitative analyses of the six FTP scans with responders presenting with MCI revealed
three scans visually read as positive of which two had above cut-off cSUVR and met
criteria for early Braak's stages I and II. Another responder who was visually read
as negative displayed above cut-off cSUVR and met criteria for Braak's stages I and
II tauopathy. While there was more evidence of A instead of T in all 12 responders,
we surmise that this may simply reflect the early stages of the ATN cascade, as A
can precede T before leading to N, thereby urging the need for follow-ups with responders
with CI who demonstrated higher FBB activity.
These results are preliminary and while suggestive of an underlying neuropathological
etiology for the observed early-onset neurocognitive dysfunction observed in WTC responders,
they nevertheless highlight the importance for continuing investigations of this phenomenon.
WTC responders are presenting with an alarming incidence of neurocognitive and neurobehavioral
dysfunction and our group is actively working on identifying which neurodegenerative
dementia subgroup, they subscribe to or if we are faced with a unique and emerging
WTC-specific neurodegenerative dementia because of their unique exposures. This study
represents our first PET modality investigation for the most common neurodegenerative
dementia subgroup that is AD. Though there is evidence that one WTC responder with
CI may subscribe to AD with an above threshold CL value, along with half of the other
responders in this study at varying levels of risk for AD, the MRI data from this
study, along with a cortical thickness study previously reporting cortical thinning
in WTC responders,[5 ] strongly support ongoing and future investigations of the neurobiological etiology
underlying WTC-related neurocognitive dysfunction.
Limitations
While being a pivotal study indicating the need for larger replication efforts, this
study has limitations. First and foremost, this study was limited by the small sample
size and lack of cognitively intact case controls. Instead, we relied on MRI age-
and sex-matched normative data and published PET cut-offs for FBB and FTP and CL scores,
as these are well-validated quantitative methodologies to inform our conclusions.
Furthermore, the small sample size meant that we were unable to examine potential
explanations and/or correlates from the PET/MRI data with neurocognitive, neurobehavioral,
and neuromotor variables which we collected at the WTC Health and Wellness Program
and at research participant screening sessions. Moreover, since data on both FBB and
FTP PET radiotracers were not available across the same 12 responders presented in
this study, it was not possible to compare FBB and FTP retention and MRI gray matter
volumes within patients to further probe the ATN model of AD. Finally, we have no
postmortem neuropathology on any WTC responders with MCI, and clinicopathological
correlations will be key to defining the molecular and cellular underpinnings and
to identify membership to a potential neurodegenerative dementia subgroup.
Conclusion
Taken together, PET/MRI assessment of 12 WTC responders with neurocognitive dysfunction
suggests that there is evidence for neurodegeneration across cortical regions. All
12 WTC responders in this study exhibited neurodegeneration in cortical regions related
to AD, and half of the responders exhibited evidence for abnormal cortical aggregation
of AD biomarkers. Future cross-sectional and longitudinal imaging studies performing
concurrent PET/MRI assessments of both Aβ and tau in the same responder are needed
to further the feasibility of the ATN model in WTC responders experiencing neurocognitive
dysfunction. Future studies investigating cerebrospinal and plasma biomarkers of AD
or ADRD are warranted, as are detailed quantitative postmortem examinations from potential
brain donors. Identifying which neurodegenerative dementia subgroup WTC responders
subscribe to or are at risk of as a result of their exposures at September 11, 2001,
will be key in informing policy makers and their future efforts to develop appropriate
clinical and patient care interventions.
