1. Clinical demographic results
The 172 PD patients were divided into three groups based on the time of appearance of CI: CI occurring at the first treatment (F-CI, n=58), CI occurring at the last treatment (L-CI, n=66), and no CI during the entire duration of follow-up (N-CI, n=48). In this study, the age of onset was equal to the age of first treatment minus the duration of disease. There was significant difference in the age of onset, age of first treatment, and smoking history between the N-CI and CI groups (F-CI and L-CI), but no difference was noted between the F-CI and L-CI groups (Table 1).
We used single-factor ANOVA analysis for age of onset, age of first treatment, and duration of disease. The area under the ROC curve (AUC) of age of first treatment was the largest (Figure 2). When the age of first treatment ≥61 years was considered as the threshold value for PD diagnosis in patients with CI, the sensitivity and specificity were 77.40% and 66.70%, respectively.
2. Comparison of MS and NMS among subgroups
2.1 Comparison of F-CI and N-CI, and Non-FCI (L-CI and N-CI) clinical symptoms at the first treatment
We carried out detailed statistical analysis on the clinical symptoms of PD patients at the first treatment and compared the MS and NMS of F-CI and N-CI, F-CI and Non-FCI (Which refers to PD patients without CI at the first treatment, including: L-CI and N-CI). As shown in Table 2, there was a significant difference in depression between F-CI and N-CI, which shows that the depression is more common in patients with CI at the first treatment than in those without CI during treatments.
Upon comparing F-CI with N-CI, the P values of depression and anxiety were 0.047 and 0.087, respectively, which were included in the logistics regression analysis.
2.2 Comparison of clinical symptoms at the first treatment and cumulative and newly added symptoms at the last treatment between L-CI and N-CI
We compared the clinical symptoms of L-CI and N-CI at the first treatment. There was a significant difference in depression and olfactory dysfunction between the L-CI and N-CI groups (Table 3). Accordingly, we analyzed the new or aggravated symptoms of L-CI and N-CI patients at the last treatment and compared the differences in the total clinical symptoms. The results suggest that there is a significant difference in total symptoms between L-CI and N-CI in hypomnesia and reduced computing power (RCP). New or aggravated hypomnesia is more common in L-CI than in N-CI (P=0.016).
Statistically significant clinical symptoms (depression, olfactory dysfunction, hypomnesia, RCP, all P<0.05) and sexual dysfunction and anxiety (0.05≤P<0.1) were included in the logistics regression analysis.
3. Comparison of TCS results between the first- and last- examination
TCS was performed in PD patients during the first and last hospitalization. In almost all patients, the midbrain area, hyperechoic area, and echo grading of the SN were recorded. Compared with the first TCS, the increased grade of SN in PD patients was significantly more than the decreased grading (P=0.001) (Figure3, Table 4).
There was no statistically significant difference in the ultrasonic examination time interval among the three groups of PD patients. Interestingly, there was a difference in the midbrain area between the last and the first examination in patients with F-CI, but not in patients with L-CI and N-LC. The grading of the last TCS examination of L-CI was higher than that of the first (P=0.023).
4. Logistics regression analysis
The demographic differences of 172 patients with respect to age of onset, age of first treatment, and smoking history were statistically significant, which were included in the logistics regression analysis. Although the difference in the change of the midbrain area within the group at the first- and last examination of F-CI was statistically significant, there were differences in the application of the midbrain area in our cross-sectional study. Therefore, it is difficult to consider the midbrain area as an index to measure the disease progress of patients. SNH is a relatively qualitative imaging index. As a typical ultrasound manifestation of PD patients, it can represent the value of TCS in PD examination. Therefore, we only included the different ultrasound grading of SN in the logistics regression analysis.
4.1 Logistics regression analysis of the demographics and clinical symptoms of F-CI and N-CI
The results showed that the age of onset, age of first treatment, smoking history, anxiety, and depression were not independent risk factors for CI in PD patients at the first treatment (Table 5).
The ROC was established according to the logistics regression analysis results of F-CI and N-CI (Fig. 4), and the AUC was 0.770. The sensitivity and specificity of identifying PD patients with CI at the first treatment were 83.10% and 68.70%, respectively.
4.2 Logistics regression analysis of demographics, first clinical symptoms, and first ultrasound grading of L-CI and N-CI
There were statistically significant differences in depression, sexual dysfunction, and olfactory dysfunction between L-CI and N-CI patients, indicating that these signs are independent risk factors in identifying PD patients without CI at the first treatment and whether CI will occur later (Table 6).
The ROC was established according to the logistics regression analysis results of L-CI and N-CI (Fig. 5), and the AUC was 0.793. The sensitivity and specificity of predicting CI in PD patients were 81.8% and 64.6%, respectively.
5.3 Logistics regression analysis of cumulative clinical symptoms, last ultrasound grading, and demographics of patients with L-CI and N-CI at the last treatment
Only hypomnesia was an independent risk factor for CI in PD patients, while the last ultrasound grading was not an independent risk factor (Table 7).
The ROC was established according to the logistics regression analysis results of L-CI and N-CI (Fig. 6), and the AUC was 0.783. The sensitivity and specificity of identifying PD patients with cognitive impairment at the last treatment were 59.1% and 85.4%, respectively.
5. Risk factors of PDD in patients with cognitive impairment
Although it is reported that patients with CI will eventually meet the diagnostic criteria of dementia9, only some patients were finally diagnosed with PDD or dementia during the follow-up period in our study. Therefore, we analyzed the clinical and ultrasound odds ratio (OR) of PDD and non-PDD to identify the risk factors of dementia in patients with CI. An OR=1 indicates that this factor has no effect on the occurrence of disease, OR>1 indicates that the factor is a risk factor, OR<1 indicates that the factor is a protective factor (Table 8).
In PD patients with CI, the risk of dementia increased 5.882 times with RCP. Moreover, rigidity and olfactory dysfunction are also risk factors for dementia, 2.091- and 1.987-times higher than in N-PDD patients, respectively. However, drinking history, somnipathy, and sexual dysfunction were protective factors for dementia in PD patients with CI.