More extensive hypometabolism and higher mortality risk in patients with right- than left-predominant neurodegeneration of the anterior temporal lobe

We included patients conforming with the following inclusion and exclusion criteria (Fig. 1).

Cerebral [¹⁸F]fluorodeoxyglucose (FDG) PET was carried out at our institution for differential diagnosis of cognitive impairment between 01/2009 and 05/2020 (inclusion criterion 1). These were searched by an experienced clinician (P.T.M.; > 20 years of neuroimaging experience with PET and SPECT) for a metabolic pattern typical for svPPA or SD (inclusion criterion 2), which was defined as predominant hypometabolism of the left, right, or both ATL. This hypometabolism may extend to frontal and (inferior) parietal cortices, but with a clear-cut gradient from the temporal pole to other regions and without clearly suggesting other neurodegenerative diseases (e.g., Alzheimer’s disease, behavioral variant of FTD (bvFTD), dementia with Lewy bodies, or atypical parkinsonian syndromes). FDG PET were read in a highly standardized fashion based on five by six axial slice displays (scaled to brain parenchyma uptake and using an identical hot-metal colorscale and thresholding) and NEUROSTAT 3D-SSP decrease displays compared to age-matched controls [14]. On a 4-point-scale, the rater marked whether a metabolic pattern typical for svPPA or SD was present (not present or present with low, intermediate, or high certainty). We included patients with high certainty of svPPA or SD metabolic pattern and patients with intermediate certainty plus a negative amyloid PET (inclusion criterion 3). PET reads were blind to all clinical details, e.g., type of cognitive deficits. Patients with metabolic disturbance of other than neurodegenerative origin and those without available survival data were excluded (exclusion criteria 1 and 2).

Notably, in contrast to several previous studies, we did not use a clinical syndrome (e.g., SD, svPPA, or bvFTD) as an inclusion criterion in order to be able to characterize the biological entity of right ATL neurodegeneration.

The institutional Ethics Committee approved this retrospective study (application No. 351/16). All patients gave written informed consent to retrospective analysis of their data.

PET images were acquired on a Siemens ECAT PET scanner (40–60 min p.i.; N = 9), Philips Gemini TF64 (N = 21), or Philips Vereos (N = 5) PET/CT systems (50–60 min p.i.) after injection of 292 ± 29, 219 ± 31, or 213 ± 7 MBq FDG, respectively. For details on data acquisition and reconstruction, please see [15, 16]. In order to account for different spatial resolutions of the scanners, reconstructed and spatially normalized images were smoothed with Gaussian filters of 5 (ECAT), 7 (TF64), or 8 mm FWHM (Vereos) using SPM12 (Version 7219; www.​fil.​ion.​ucl.​ac.​uk). After scaling to individual brain parenchyma uptake, PET images were subjected to statistical analyses with SPM12.

The German version of the Consortium to Establish a Registry for Alzheimer’s Disease Neuropsychological Assessment Battery (CERAD-NAB [17]) was used as part of the routine clinical work-up in a memory clinic setting. Analyzed test scores were semantic fluency (animals), Boston Naming Test (number of correct responses), word list learning (sum of three trials), word list recall (absolute number of recalled words and relative recall, i.e., percent of words recalled in the third learning trial), word list recognition, figure drawing, and figure recall (absolute and percent of previously drawn). Raw test scores were converted to age-, sex-, and education-corrected z scores with reference to normative test data [18].

Survival data of individual patients were acquired by automated queries of the state resident registry in 08/2021 or from clinical records of the hospital information system, yielding either confirmation of survival or the date of death. For the main survival analysis, we used survival durations from the time of FDG PET. We only secondarily used survival durations from symptom onset, as the latter is commonly only a rough estimate in retrospect.

For the definition of symmetric ATL metabolism and for comparison of regional FDG uptake, we used FDG PET data from 35 healthy elderly persons and 45 age-matched control patients scanned under identical conditions from our database for statistical comparisons, in whom somatic CNS diseases were carefully excluded. Controls were imaged on Philips TF64 (N = 35) or Philips Vereos (N = 45) PET/CT systems (Table 1).

Patients fulfilling all inclusion and no exclusion criteria were assigned to either of the three categories of patients with asymmetric right-predominant, left-predominant, or symmetric ATL involvement. The normal range of ATL metabolic asymmetry was defined based on the control group. Normalized regional glucose metabolism of the middle temporal pole region of the automated anatomical labeling atlas, version 3 (AAL3) [19], was read out, and an asymmetry index (AI) was calculated: AI = 200 × (right − left)/(right + left). Patients with an AI that fell within the 95% confidence interval were assigned to the group of patients with symmetric ATL neurodegeneration (“BILATL patients”). All other patients were assigned to the respective groups with asymmetric right- (“RATL patients”) or left-predominant neurodegeneration of the ATL (“LATL patients”).

Demographic variables were compared between patient and control groups by t tests for continuous or Fisher’s exact tests for categorical variables. The significance threshold for pairwise comparisons was set to p = 0.05.

Normalized regional FDG uptake was compared between groups by ANCOVA with factor “group,” controlling for PET scanner. Groups were contrasted in a pairwise fashion, such that contrast images depicted decreased FDG uptake in each of the patient groups compared to controls and each of the remaining patient groups (FWE-corrected p < 0.05; cluster extent k > 500 voxels of 1 mm isotropic size).

Cognitive performance of each of the patient groups (z-scores based on normative data) was tested for a difference from zero (i.e., from normative control data) by one-sample t tests (p < 0.05 after Bonferroni-correction for 9 test variables). Furthermore, cognitive performance was compared between patient groups by ANOVA and post-hoc Tukey HSD (p < 0.05 after Bonferroni-correction for 9 test variables).

Furthermore, we searched clinical records for documented behavioral symptoms: four symptoms reported to be typical for right SD (social awkwardness, job loss, loss of insight, difficulties with person identification [2]) and the symptoms represented in the 12 items of the short version of the Neuropsychiatric Inventory (delusions, hallucinations, agitation/aggression, dysphoria/depression, anxiety, euphoria/elation, apathy/indifference, disinhibition, irritability/lability, aberrant motor behaviors, nighttime behavioral disturbances, and appetite/eating disturbances) [20]. We exploratively compared groups regarding the presence or absence of each of these symptoms using separate chi-square tests (significance threshold set to p = 0.05).

Group comparisons of PET images were performed with SPM12 and Matlab R2018b (The MathWorks, Inc.); all other computations were performed with R version 4.0.3 (www.​R-project.​org).

We included 35 patients in the analyses: 10 patients with right-, 17 with left-predominant, and 8 with symmetric neurodegeneration of the ATL. Amyloid PET was performed in 19/35 patients (negative according to clinical reports). Neither age nor sex differed between groups (p > 0.1, ANOVA and Fisher’s exact test, respectively). There were no significant differences between patient groups in symptom duration, education years, or MMSE (p > 0.1, ANOVA; Table 1). Information about handedness was available from 24/35 patients, of whom 23 were right-handed (7 patients with right-, 11 with left-predominant, and 5 with symmetric neurodegeneration of the ATL), one patient with symmetric ATL involvement was ambidextrous. Interestingly, two of the included patients were brothers, one with left-, the other with right-predominant ATL involvement. Please see Supplementary Fig. 1 for example of patient images.

Decreased normalized regional glucose metabolism was most extensive in patients with right-predominant neurodegeneration of the ATL (Fig. 2; please see also Supplementary Table 1). Compared to control subjects, it was observed in the right ATL, as expected, but also in the right orbitofrontal and anterior cingulate cortex. To a lesser extent, it was also decreased in the left ATL and left orbitofrontal cortex. Compared to patients with left-predominant ATL neurodegeneration, metabolism was decreased in the right ATL and right frontal and right parietal cortex. Compared to patients with symmetric ATL neurodegeneration, FDG uptake was reduced in two small clusters in the right inferior ATL and the right medial temporal lobe.

Patients with left-predominant ATL neurodegeneration showed significantly reduced FDG uptake when compared to controls in the left ATL and orbitofrontal cortex (Fig. 2).

The group with symmetric ATL neurodegeneration showed decreases compared to controls in both ATL and, compared to left-predominant patients, in a small cluster in the right inferior ATL (Fig. 2).

Neither the left-predominant nor the symmetric ATL neurodegeneration groups had significantly decreased metabolism compared to patients with right-predominant ATL neurodegeneration.

Subtest scores of the CERAD-NAB were available from 28 to 31 of the 35 patients. One-sample t tests revealed significantly impaired cognitive performance with reference to normative data (p < 0.05 after Bonferroni correction, Fig. 3): patients with right-predominant neurodegeneration of the ATL were impaired in naming, verbal learning, and figure recall. Patients with left-predominant neurodegeneration of the ATL showed deficits in naming, verbal learning, recall, recognition, and semantic fluency. Patients with symmetric ATL neurodegeneration were impaired in naming, verbal recall, figure recall (absolute and relative), and semantic fluency. None of the comparisons between patient groups reached significance, neither at Bonferroni-corrected nor uncorrected levels.

Thirteen of the 35 included patients died (4 with right-, 6 with left-predominant, 3 with symmetric ATL neurodegeneration) within the median follow-up duration of 7.7 years (95% C.I.: 5.5 to undefined). The age- and sex-adjusted Cox model (overall model fit: LR = 7.3 on 4 degrees of freedom, p = 0.12) revealed a significantly higher mortality risk in patients with right- compared to those with left-predominant ATL neurodegeneration (p = 0.042; HR = 6.1 [95% C.I. 1.1 to 34.7]). No other pairwise comparisons reached significance. Median survival durations were 5.7 for patients with right-, 8.3 years for patients with left-predominant, and 6.7 for patients with symmetric neurodegeneration of the ATL (Fig. 4).

Sample sizes were small in this retrospective, mono-centric study. Owing to the long time span during which FDG PET were acquired, different PET scanners were used. We tried to cope with this with different smoothing filters in image preprocessing and by including PET scanner as a covariate in statistical analyses. Patient selection in this study was based on imaging findings. It was neither based on clinical diagnoses, nor based on neuropathological findings (although all available amyloid PET were negative). It is hence possible that patients with non-neurodegenerative etiologies have been included. From all data presented here, we assume, however, that most included patients fall in the clinical spectrum of FTD and have a TDP-43 neuropathology, most likely TDP-C.