Meta-analysis of the literature on diagnostic accuracy of SPECT in parkinsonian syndromes

Abstract

Background: Parkinson's disease (PD) is the second most common neurodegenerative disorder. One of the most widely used techniques to diagnose PD is a Single Photon Emission Computer Tomography (SPECT) scan to visualise the integrity of the dopaminergic pathways in the brain. Despite this there remains some discussion on the value of SPECT in the differential diagnosis of PD. We did a meta-analysis of all the existing literature on the diagnostic accuracy of both pre- and post-synaptic SPECT imaging in the differential diagnosis of PD.

Methods: Relevant studies were searched in Medline, EMBASE and Cochrane databases with back-searching of their reference lists. We limited our analysis to studies with a clinically relevant methodology: i.e. when they assessed the ability of the SPECT to provide 1. diagnosis of PD in an early phase vs. normalcy; 2 diagnostic differentiation between PD and essential tremor (ET); 3. distinguishing between PD and vascular parkinsonism (VP); 4. delineation of PD from atypical parkinsonian syndromes (APS). Gold standard was, dependent on the study type, clinical examination at initial visit or follow-up, and/or response to dopaminergic agents.

Results: The search gave 185 hits, of which we deemed 32 suitable for our analysis. From these we recalculated the diagnostic odds ratio of SPECT for the clinical questions above. The pooled odds ratio (with 95%CI) for presynaptic SPECT scan's ability to distinguish between early PD and normalcy was 60 (13 - 277). For the ability to differentiate between PD and ET this ratio was 210 (79-562). The ratio for presynaptic SPECT's ability to delineate PD from VP was 105 (32 - 348). The mean odds ratio for the presynaptic SPECT scans to differentiate between PD and the two APS was 2 (1 - 4), and for the postsynaptic SPECT imaging this was 19 (9-36).

Conclusion: SPECT with presynaptic radiotracers is relatively accurate to differentiate patients with PD in an early phase from normalcy, patients with PD from those with ET, and PD from VP. The accuracy of SPECT with both presynaptic and postsynaptic tracers to differentiate between PD and APS is relatively low.

Figure 4

Diagnostic differentiation between PD and APS (MSA & PSP). Presynaptic radiotracer. [28, 30, 36, 38, 39, 46-48, 50-52]. * follow-up trial. # The pooled odds ratio after excluding the study of Messa '98 (sensitivity 100%, specificity 0%) remains unchanged.

Figure 5

Diagnostic differentiation of patient with PD vs. APS (MSA & PSP). Postsynaptic radiotracers [18, 28, 34, 35, 37, 46, 50, 53-59]. * follow-up trial. # The pooled odds ratio (95%CI) after excluding the study of Pirker '97 (sensitivity 100%, specificity 0%) did not change significantly: 19 (10–33). Schwarz 1993. this trial can be seen as follow-up trial (long-term l-dopa is golden standard) but also as a cross-sectional trial. In the last case the results will be different (sensitivity = 79%, specificity = 100%). Schwarz 1997: We took signs (in) compatible with PD as golden standard. If taken long-term l-dopa as golden standard sensitivity is 100% and specificity is 67%. Schwarz 1998: We took signs (in) compatible with PD taken as golden standard, if taken long-term-l-dopa as golden standard sensitivity is 100% and specificity is 50%.

Figure 6

Diagnostic differentiation of patients with PD vs. APS (MSA & PSP). Doctors Prediction [28, 34, 38, 39]. * follow-up trial. # The pooled odds ratio (95%CI) after excluding the studies with a sensitivity of 100% and a specificity of 0% (Stoffers '05 and Schwarz '98) did not change significantly: 6 (1–40). Both trials diagnosed at the beginning of the follow-up all patients as having PD.

Figure 2

Patients with PD vs ET. Presynaptic radiotracer [36, 38, 45-48, 84]. * = follow-up trial.

Figure 3

Diagnostic differentiation of patients with PD versus VP. Presynaptic tracer. [28, 36, 38, 47, 49]. * = follow-up trial.

Figure 1

Diagnostic differentiation of patients with PD in an early phase vs. normalcy. Presynaptic radiotracer [33, 40-44]. # all studies have a specificity of 100% so the pooled odds ratio should therefore be infinite. The odds ratio of 60 is caused by STATA software's procedure for handling zero cells in the 2 × 2 table. abnormal SPECT definition: > 2 standard deviations below the bindingrate of healthy controls. outcome: sensitivity, specificity, odds ratio 95% confidence interval. surface square is based on the weight of the study. n = numbers of subjects. FP-CIT: Iodine-123-Ioflupane. TRODAT: 99mCT-TRODAT-1.