Semi-quantitative analysis of visually normal 123I-FP-CIT across three large databases revealed no difference between control and patients

Characteristics of ScanOnlyDB

A total of 764 subjects matched for gender and age were included in the ScanOnlyDB; the descriptive statistics are summarized in Table 1. The main clinically suspected diagnoses before the 123I FP-CIT study of the included subjects were: essential tremor (n = 181, 23.7%), post-neuroleptic Parkinsonism (n = 177, 23.2%), atypical Parkinsonism (n = 95, 13.5%), and dementia without dopaminergic denervation (n = 103, 12.4%). For the complete list, see Additional file 1: Table S1. Briefly, the striatum, the caudate, and the putamen mean SBR values of women were 6.6%, 7.1%, and 8.4% higher, respectively, than those of men (P < 0.001). The caudate mean SBR value was higher than the putamen mean SBR value for both genders (P < 0.05). There was no significant asymmetry for the striatum as the mean ± SD SBR was 2.64 ± 0.50 for the left side and 2.62 ± 0.49 for the right side (P = 0.53). CT AC-corrected SBR values were lower than Chang AC-corrected SBR values for all striatal regions (P < 0.0001).

Table 1 Characteristics according to genderData from ScanOnlyDB are presented

Mean and standard deviation of age, injected dose, and Chang AC Specific Binding Ratio (SBR) values for the striatum, the caudate and the putamen, for the database composed of visually normal scans (suspected Parkinsonism and no diagnostic follow-up, ScanOnlyDB).

ScanOnlyDB GLM with age as independent variable

The striatum, the caudate, and the putamen SBR decreased with age (P < 0.001; Table 2). For example, the mean ± SD SBR variation for the striatum (Chang AC) was − 4.94 ± 0.27% per decade. For the posterior putamen (CT AC) the estimated intercept ± SE was 3.21 ± 0.06% and slope 0.020 ± 0.0009. In addition, the SBRputamen to SBRcaudate ratio indicated a clear covariance with age (P < 10–5).

Table 2 linear regression between Specific Binding Ratio according to attenuation correction methods and age for ScanOnlyDB

Linear regression parameters derived from Specific Binding Ratio (SBR)of visually normal scans (suspected Parkinsonism and no diagnostic follow-up, ScanOnlyDB).

Chang AC: a numerical method for attenuation correction, CT AC: computed tomography-based attenuation correction, standard error: SE,  All p-values < 10–5

ScanOnlyDB: GLM with age, gender as an independent variable, and age-gender interaction

The relationship between age and SBR in men and women is presented in Fig. 2 for the striatum (Fig. 2a), caudate (Fig. 2b), and putamen (Fig. 2c). The R2 values of the GLM were 0.33, 0.24, and 0.33 for CT AC and 0.31, 0.23, and 0.32 for Chang AC. Striatum Chang AC CoV was 15% after correction for mean age, whereas the SD was 0.40. Table 3 details the estimates of the GLM analysis for each variable. There was an age-gender interaction. For example, for Chang AC, estimates for interaction parameters per year were − 0.027, − 0.029, and − 0.027 (women) and − 0.022, − 0.021, and − 0.023 (men) for the striatum, the caudate, and the putamen, respectively. We computed adjusted SBR values according to age and gender at three different representative ages, 31 years, 62.5 years, and 94 years (Fig. 2d). At 31 years, female SBR values were 8.8% higher than male SBR values. This difference decreased with age (6.4% at 62.5 years and 0.9% at 94 years).

Fig. 2figure 2

Scatterplots of ScanOnlyDB 123I FP-CIT SBR (Chang AC) values as a function of age for both male (open circle) and female (blue circle) for the three VOIs: Striatum (a), Caudate (b), and Putamen (c). Data relevant to each gender were independently fitted by linear regression lines (solid black line for males and solid blue line for females) and superimposed on each respective scatterplot. Age-gender interaction of striatal SBR for a mean age of 31 years (orange), 62.5 years (yellow), and 94 years (purple) is shown in (d)

Table 3 Estimates of the general linear model analysis of SBR from the database ScanOnlyDB

General linear model estimates of Specific Binding Ratios (SBR) from the database composed of visually normal scans (suspected Parkinsonism and no diagnostic follow-up, ScanOnlyDB) as a function of age, gender, and age-gender interaction (for the ‘male’ effect) for both attenuation correction methods: Chang and CT.

Chang AC: numerical method for attenuation correction, CT AC: computed tomography-based attenuation correction, standard error: SE, * p-values < 10–5

HvDB: GLM analysis with age and gender as independent variables as well as age with gender interaction

For the HvDB, the GLM (R2 = 0.30) showed a covariation of SBR with age (estimate ± SE:− 0.026 ± 0.0032, P < 10–5). However, no significant covariation with gender (P = 0.4), nor age-gender interaction (P = 0.34) was found. The mean ± SD decrease in striatal SBR per decade was − 5.92 ± 0.24%. The linear regression equation of SBRputamen to SBRcaudate ratio was estimate ± SE: 0.86 ± 0.03 + 0.0005*age (± 0.0006) but without a significant covariation with age (P = 0.39). After correction for mean age, the CoV for striatal SBR was 15% and the SD was 0.42.

NoDG5YearsDB: GLM analysis with age and gender as independent variables and age-gender interaction

For the NoDG5YearsDB, the model (R2= 0.35) showed a covariation of SBR (estimate ± SE): with age − 0.022 ± 0.002; P < 10–5 and gender − 0.5 ± 0.175; P = 4*10–3). However, there was no significant age-gender interaction (P = 0.14). The mean ± SD change in striatal SBR per decade was − 4.65 ± 0.22%. After correction for mean age, the CoV for striatal SBR was 16% and the SD was 0.38.

Comparison with ENC-DAT and SiemensDB

The R2 of the linear regressions of SBR with age were compared between the ScanOnlyDB, the HvDB, the NoDG5YearsDB, the ENC-DAT databases using BRASS analysis of uncalibrated ACSC data [6]and the SiemensDB [12](coefficients published for each VOI and gender in Table 4). No significant difference was found.

Table 4 SBR estimated parameters from ScanOnlyDB and those from the ENC-DAT and the SiemensDB

Comparison of slopes (estimate ± standard error) and R2 parameters from ScanOnlyDB and those from the ENC-DAT study using BRASS analysis of uncalibrated ACSC data [6] and the SiemensDB [12]. ACSC, attenuation correction and scatter correction; ENC-DAT, European Normal Control Database of DaTscan™

Combination of ScanOnlyDB and HvDB: GLM analysis with age and gender as independent variables and age-gender interaction

Figure 3 presents the SBR values of ScanOnlyDB and HvDB as a function of age after harmonization [SBR ScanOnlyDB harmonized = SBR ScanOnlyDB − 0.25]. The GLM analysis (R2= 0.27) showed a significant covariation with age (P < 10–5) and gender (P = 6.10–3). The slope ± SE was − 0.026 ± 0.001 and the intercept ± SE (male) was − 0.345 ± 0.13. There was no age-gender interaction (P = 0.08). The database factor was not significant (P = 0.76). After correction for mean age, the CoV for striatal SBR was 15% and the SD was 0.42.

Fig. 3figure 3

Scatterplot of harmonized ScanOnlyDB striatum VOI Specific Binding Ratio (SBR) values (black open circles) and HvDB SBR values (green open circles) as a function of age. Linear regression lines for age: ScanOnlyDB (green solid line) and ScanOnlyDB + NoDG5YearsDB ± 2 SD (black solid and black dotted lines) were superimposed

Combination of ScanOnlyDB and NoDG5YearsDB: GLM with age and gender as independent variables and age-gender interaction

Figure 4 presents the SBR values of the harmonized SBR ScanOnlyDB harmonized = SBR ScanOnlyDB + 0.33] as a function of age. The GLM analysis (R2= 0.35) showed a significant covariation with age (P < 10–5) and gender (P < 10–5). The slope ± SE was − 0.021 ± 0.0009 and the intercept ± SE (male) was − 0.577 ± 0.09. The age-gender interaction slope ± SE was 0.0057 ± 0.001 (P = 6.10–6). The database factor was not significant (P = 0.9). After correction for mean age, the CoV for striatal SBR was 15% and the SD was 0.39.

Fig. 4figure 4

Scatterplot of harmonized ScanOnlyDB striatum VOI Specific Binding Ratios (SBR; black open circles) and NoDG5YearsDB (blue open circles). Linear regression lines for age: ScanOnlyDB (magenta solid line) and ScanOnlyDB + NoDG5YearsDB ± 2 SD (black solid and black dotted lines)

留言 (0)

沒有登入
gif