Since the first report by Becker et al in 1995, SN hyperechogenicity in patients with IPD has been confirmed by many researchers [9, 16, 25–28]. These investigators have suggested that SN-TCD is an accurate instrument in discriminating IPD patients from normal healthy subjects, and from patients with other types of parkinsonism. However, most studies on TCD of the SN involve groups of well-diagnosed IPD patients, in the later stages of the disease. The diagnostic accuracy of SN-TCD in early-stage patient populations, with less well-defined clinical syndromes, thus remains to be determined. In an effort to do this, we did SN-TCD in 82 undiagnosed parkinsonian patients. As the gold standard for the patient's final diagnosis, we chose the clinical diagnosis after follow-up. This is, of course, not as good as post-mortem neuropathological analysis, but the best possible alternative at present.
In our study, the diagnostic accuracy of presynaptic SPECT is superior to that of SN-TCD. Especially the sensitivity and NPV in differentiating patients with parkinsonism with nigrostriatal degeneration (IPD, MSA, PSP, DLBD, CBD), from other causes of parkinsonism, are remarkably higher in presynaptic SPECT than in SN-TCD. However, the PPV of a positive SN-TCD for parkinsonism with nigrostriatal degeneration is high. This means that, although not all patients with nigrostriatal degeneration have a positive SN-TCD, a positive SN-TCD is a good predictor in diagnosing parkinsonism with nigrostriatal degeneration.
In our study, we also investigated the similarity in results between SN-TCD and presynaptic SPECT. We found a high PPV (88%) of a positive SN-TCD for an abnormal SPECT result, confirming other reports . For clinical practice, this would imply that a positive SN-TCD in a patient with an early-stage, recently diagnosed parkinsonian syndrome, would reduce the added diagnostic value of a presynaptic SPECT.
A striking difference between ours and other studies, is the percentage of false-negative IPD patients diagnosed by SN-TCD. Our 50% is much higher than the 0–20% reported by other investigators [9, 16, 25–28]. There are several explanations for our lower sensitivity rate. Firstly, we used a broad spectrum of parkinsonian patients, which is representative for the diagnostic problem that one wants to solve. Other studies with similarly mixed patient groups also found a lower accuracy of the SN-TCD [22, 23, 30]. Secondly, our sonographers were blinded to the results of the clinical diagnosis. This might also have led to a decreased sensitivity of the SN-TCD, since we considered the SN-TCD negative in cases where we had doubts about the level of hyperechointensity. Thirdly, our patient population had a relatively large number of patients in the early stages of their disease. Although Berg et al. reported SN echointensity to be stable during follow-up, the question remains whether this is also the case in early-stage patient populations . Fourthly, the quality of the ultrasound system is a non-neglectable variable, since in our pilot experiments, we found that the newest ultrasound systems will reveal hyperechointensity of the SN in more patients . Finally, the TCD technique itself can provide an explanation. One needs considerable personal expertise to perform and interpret a TCD correctly. Since the pioneering research group in Tübingen has built up an enormous amount of expertise, it will be difficult for other groups to reproduce their excellent diagnostic results [25, 32–37].
In summary, our study with early-stage parkinsonian patients, shows that the specificity and the PPV of SN-TCD for the final clinical diagnosis is just as high as it is for the SPECT scans. The sensitivity of the SN-TCD in our patient population is significantly lower than that of presynaptic SPECT scans. So in this population, the SN-TCD might be used as screening instrument: one could argue that in patients where the SN-TCD is compatible with parkinsonism with nigrostriatal degeneration, a presynaptic SPECT is no longer necessary in the diagnostic work-up. Applying this strategy, in our group of patients, 35% SPECT scans could have been omitted, resulting in a significant reduction in costs. Besides this, SN-TCD is also a more patient-friendly technique than SPECT scans.