Data from 310 patients were collected from four tertiary centers and analyzed. Table 1 provides a summary of the whole study cohort characteristics. A total of 128 patients underwent lumbar punctures to assess Alzheimer’s biomarkers, which we were able to obtain in 126 of them. No significant differences were observed between the groups concerning demographic characteristics, cardiovascular risk factors, initial cognitive scores, and AD CSF biomarkers. The average age at clinical onset was 69.8 years, while the mean age at diagnosis was 73.5 years, with a male predominance of 67.7% (210 out of 310). The most frequently reported initial symptoms were cognitive impairment (60.6%), followed by motor disorders (27.1%), and finally, psychiatric disturbances (12.3%). Notable differences in the prevalence of clinical symptoms such as parkinsonian syndrome (p < 0.01), visual hallucinations (p < 0.001), and the occurrence of falls (p < 0.01) were found between the groups. Patients with motor onset exhibited parkinsonian syndrome and falls more frequently, whereas those with cognitive onset experienced visual hallucinations less often. We also observed significant differences in the proportion of performed FDG PET scan and polysomnography that are increased in the psychiatric group, and DAT scan results are significantly more frequently abnormal in the motor group. In the same line, some treatments are differently prescribed according to the group (Table 1).

The average follow-up duration in the study was 7.25 years (SD 3.6), spanning from the first symptoms to death or the end of the follow-up period. During this time, 40% (124 patients) of the participants passed away. Patient characteristics, based on whether they experienced death, are summarized in Table 2.

In the whole study cohort, the calculated median survival (until death) was 10.0 years. The median survival for the cognitive group was 10.3 years, 9.5 years for the motor group, and 15.2 years for the psychiatric group. Figure 1 illustrates the Kaplan–Meier survival curve as a function of the initial predominant clinical symptoms. Upon visually analyzing the curve, a faster decline in the survival curve of subjects with motor clinical onset is observed compared to the other two dementia with Lewy bodies (DLB) clinical phenotypes from 6 years of follow-up. The evolution of these other phenotypes appears to be similar over time. No significant difference was detected using the log-rank test (p = 0.11).

Kaplan–Meier survival curve (death) according to clinical entry points. Survival curve according to the Kaplan–Meier method for patients with motor (blue line), cognitive (red line), or psychiatric (green line) predominant clinical onset. The numbers below the curve indicate the number of patients at risk at each time point. Differences between the two groups were evaluated using a log-rank test (p = 0.11). Data were censored at the date of last observation (crosses) or death

Multivariate analyses employing a Cox model for mortality, poor prognosis, and full clinical picture are presented in Table 3. The first model includes age at onset, sex, and site of inclusion as explanatory variables, while the second model incorporates cardiovascular risk factors. Cox1 model for mortality shows a non-significantly higher instantaneous mortality risk in patients with motor or psychiatric onset compared to cognitive onset). Age at clinical onset (HR 1.10 [1.06–1.10], p < 0.001) and male gender (HR 1.70 [1.12–2.60], p = 0.01) were associated with a significant excess risk of instantaneous mortality. The second model yielded similar but non-significant results toward a higher instantaneous mortality risk in patients with motor or psychiatric onset compared to cognitive onset. Age at clinical onset and male gender continued to be associated with a higher risk of instantaneous mortality (HR 1.09 [1.06–1.12], p < 0.001 and HR 1.71 [1.12–2.62], p = 0.01). No significant risk difference was observed for the presence of hypertension (HR 0.91 [0.50–1.65], p = 0.76), diabetes (HR 1.69 [0.80–3.59], p = 0.17), dyslipidemia (HR 0.61 [0.32–1.14], p = 0.12), or active smoking (HR 1.59 [0.37–6.88], p = 0.54). A significant decrease in the risk of instantaneous mortality was found in patients with obstructive sleep apnea syndrome.

In the whole study cohort, the median survival until a poor prognosis event or censorship is 5.7 years. This duration is 5.5 years in the cognitive group, 6.2 years in the motor group, and 6.3 years in the psychiatric group. The first Cox model revealed that age is significantly associated with a higher risk of poor prognosis (HR 1.08 [1.06–1.1], p < 0.001), while male gender is not in this model (HR 0.93 [0.70–1.2], p = 0.63). Additionally, there was a non-significant increased instantaneous risk of poor prognosis in the cognitive and motor groups compared to the psychiatric group (HR 1.50 [0.98–2.3], p = 0.06 and HR 1.38 [0.85–2.2], p = 0.19, respectively) as well as no excess risk of needing help at home (HR 1.83 [0.9–3.7], p = 0.1) or institutionalization (HR 1.06 [0.6–1.2], p = 0.9) for the cognitive onset group compared with the psychiatric onset group.

The second Cox model shows that a biological AD CSF biomarker profile, was significantly associated with a short-term risk of poor prognosis (HR 2.66 [1.46–34.8] p = 0.001) as well as active smoking (HR 3.38 [1.33–8.6] p = 0.01). We did not find any other significant associations regarding the phenotype or the comorbidities (hypertension, dyslipidemia, obstructive sleep apnea, diabetes).

Out of the 310 patients included in the study, 246 (79.4%) exhibited a complete clinical picture (cognitive, psychiatric, and motor symptoms) during their follow-up, with an average delay of 3.41 years (SD 3.1). A complete clinical picture was observed in 133 of the 188 patients with a predominantly cognitive clinical onset (70.7%), with an average delay of 3.7 years (SD 2.9). In contrast, 79 of the 84 patients with a predominantly motor onset (94%) presented a complete clinical picture with an average delay of 2.6 years (SD 2.0), and 34 of the 38 patients with a psychiatric entry point (89.5%) exhibited a complete clinical picture with an average delay of 4.2 years (SD 4.8). Statistical comparisons between groups revealed a significantly increased risk of a full clinical picture in the motor predominant onset group compared to the cognitive predominant onset group (HR 1.45 [1.08–1.9] p = 0.014 in Cox1 model, HR 1.41 [1.05–1.9] p = 0.02 in Cox2 model). Additionally, there was a non-significant increase in risk compared to the psychiatric predominant onset group (HR 1.35 [0.86–2.1] p = 0.20 in Cox1 model, HR 1.45 [0.91–2.3] p = 0.12 in Cox2 model).