Accuracy of Visual Assessment of Dopamine Transporter Imaging in Early Parkinsonism

Abstract

Dopamine transporter (DaT) imaging may be supportive in the initial clinical diagnostic workup in patients with suspected parkinsonian diseases, given that the method has the potential to detect dopaminergic degeneration. We investigated the diagnostic accuracy of visual assessment of the initial DaT single‐photon emission CT (DaT‐SPECT) with ¹²³I‐FP‐CIT in a large group of early‐stage parkinsonian patients. After inclusion in a long‐term multidisciplinary population‐based prospective study, a baseline DaT‐SPECT was done in 171 incidental, L‐dopa‐naïve, parkinsonian patients (102 men and 69 women) and 37 healthy controls (19 men and 18 women). The results of the DaT‐SPECTs were linked to criteria‐based clinical diagnoses, which were set after a mean follow‐up of 4.6 (±1.7) years. The outcome of the visual assessment was also compared with that of a semiquantitative evaluation method using regions of interest to measure uptake ratios in the caudate and putamen. We found that visual assessment of DaT‐SPECT in clinically diagnosed incidental Parkinson's disease patients had a sensitivity of 94% and a specificity of 92%, rendering a positive likelihood ratio of 11.75 and a negative likelihood ratio of 0.07. The proportion of false positives was 1.4% and false negatives 4.8% at baseline. These figures were comparable to those of the semiquantitative method. This study demonstrates that visual interpretation of presynaptic dopamine imaging with ¹²³I‐FP‐CIT offers reliable support in the diagnostic procedure of early parkinsonian diseases.

Established diagnostic criteria are often used to specify the clinical diagnosis of Parkinson's disease (PD)1 or atypical parkinsonian syndromes (APSs), such as MSA2 and PSP.3 A common feature of PD, MSA, and PSP is progressive degeneration of dopaminergic neurons in the substantia nigra, with subsequent dopamine deficit in the striatal synapses and loss of dopamine transporter (DaT) protein. Imaging of the presynaptic DaT population in the striatum with single‐photon emission CT (SPECT) may therefore be of supportive value in ruling out other causes of parkinsonism.

A common way of assessing DaT‐SPECT examinations is by visual interpretation. A normal uptake pattern is visualized in the transversal plane as a symmetrical comma or bean‐shaped uptake in the striatum, whereas the typical pathological feature in degenerative parkinsonism is a more or less asymmetrical, dot‐shaped uptake. As symptoms gradually progress and the emerging motor symptoms urge the patient to the first contact with the health care system, the degenerative process in the SN has already reached an advanced stage.4, 5 Even with a duration of the self‐reported symptoms shorter than 1 year, the SPECT examination may be clearly pathological, with an obvious bilateral uptake reduction.6

Thus, DaT‐SPECT can contribute to the early diagnostic accuracy when the characteristic clinical picture has not yet fully evolved.7, 8, 9, 10 Benamer et al.11 found a high sensitivity for DaT‐SPECT in the early phase of clinically diagnosed parkinsonism; however, large‐scale prospective studies on this issue are relatively few.

A test has to be compared with a gold standard. In this context, the gold standard is ultimately a postmortem histopathological examination confirming the clinical diagnosis. However, to our knowledge, only one study has validated DaT‐SPECT in parkinsonian diseases against autopsy‐confirmed diagnoses, in 14 patients.12 Another study validated the accuracy of DaT‐SPECT against autopsy‐verified diagnoses of dementia with Lewy bodies (LBs).13 An alternative standard against which DaT‐SPECT can be validated may be a clinical diagnosis that is based on long‐term follow‐up using established clinical diagnostic criteria. The UK Parkinson's Disease Society Brain Bank (UK PDSBB) clinical criteria for PD1 has been reported to have high accuracy.14, 15, 16, 17, 18 In this study, we used the long‐term follow‐up clinical diagnosis of parkinsonian syndromes (i.e., PD, MSA, and PSP), based purely on established clinical criteria and unbiased by other tests (including DaT‐SPECT) as the standard, in a large group of patients, in order to assess the sensitivity and specificity of visual interpretation of a DaT‐SPECT with ¹²³I‐FP‐CIT ([¹²³]I‐N‐ω‐fluoropropyl‐2‐β‐carbomethoxy‐3β‐(4iodophenyl)nortropane,¹²³I‐ioflupane; GE Health BV, Eindhoven, The Netherlands) performed at the first visit to the doctor. The patients in this study were included in a population‐based incidence cohort study consisting of patients with parkinsonism according to the stage 1 UK PDSBB criteria and a group of age‐matched nonparkinsonian healthy controls.

The aim of this study was to assess the sensitivity, specificity, and negative and positive likelihood ratios (LRs) for visual interpretation of ¹²³I‐FP‐CIT SPECT in an early stage of disease, including a comparison with a semiquantitative analysis of the uptake.

Patients and Methods

Subjects

An ongoing longitudinal (8 years) population‐based research project on parkinsonian disorders19 included 185 incidental cases fulfilling the step 1 UK PDSBB clinical criteria for parkinsonism1 and who gave their informed oral and written informed consent to participate in the project (inclusion time 1 January 2004–30 April 2009).

The study was approved by the regional ethics committee and the local radiation safety committee. Inclusion criteria were (1) informed consent, (2) residence within the catchment area of the University Hospital of Umeå (Umeå, Sweden), and (3) previously nondiagnosed and recent‐onset parkinsonism. Exclusion criteria were (1) refusal, (2) relocation outside the catchment area, (3) dementia or a mini‐mental test score less than 24 of 30, and (4) evidence of essential tremor or secondary parkinsonism. Inclusion was based on clinical assessment preceding any imaging or other auxiliary tests.

The study population of this article comprised the 171 patients included in the main research project (102 men and 69 women) who had a baseline DaT‐SPECT soon after inclusion. The mean age was 71.3 (±9.3) years and mean self‐reported duration of symptoms was 2.1 (±1.7) years (median, 1.6 years) at baseline SPECT. The mean UPDRS, motor part (UPDRS‐III; maximum subtotal score: 108 points) score at baseline was 27.1 (±11) points, and the median UPDRS‐III score was 26 points (data missing in 2 patients). In this study population, 13 patients fulfilled criteria for MSA and 16 patients fulfilled criteria for PSP after a median follow‐up period of 4 years.

At baseline, 50 cases (29.2%) had the tremor‐dominant motor phenotype and 102 patients (59.6%) had a predominantly postural impairment and gait disorder phenotype of parkinsonism, whereas 19 cases (11.1%) had an indeterminate type phenotype.20 In the protocol of the prospective study, a number of different auxiliary examinations and tests are included, such as pre‐ and postsynaptic SPECT, structural and MRI of the brain, and neurophysiological and ‐psychological tests. However, the clinical diagnoses were assessed and reassessed solely based on established clinical criteria, by specialists in movement disorders at the Department of Neurology at the University Hospital of Umeå.19 The first consensus diagnostic criteria for MSA2 were used, where the result of a DaT‐SPECT was not taken into account (which is the case in the second consensus criteria for MSA21).

The mean follow‐up time of the patients in this study, reflecting the maturity of the diagnoses, was 4.6 (±1.7) years (median, 4 years); in 7 cases, the follow‐up time was only up to 1 year. The baseline DaT‐SPECTs were linked to the updated diagnosis at the latest clinical follow‐up occasion.

Healthy volunteer controls (HCs) were recruited by newspaper announcements. Thirty‐seven elderly volunteers (19 men and 18 women) were included after informed consent was obtained. A structured neurological examination was performed by two of the authors (J.L. and L.F.), specialists in movement disorders, to rule out any signs of parkinsonism or other neurological disorders. HCs were age matched to the first 50 included patients, with a mean age of 68.2 (±6.7) years. All HCs had a baseline DaT‐SPECT as part of the research protocol described above. Twelve were only participating at baseline, 4 were followed for 3 years (of which 1 was then excluded because of incurrent dementia without signs of LB disease), and the remaining 21 HCs were followed for 5 years. None of the included HCs was later diagnosed with parkinsonian or LB diseases.

SPECT Procedures

The DaT‐SPECT was performed 3 hours after an intravenous 185‐MBq bolus dose of ¹²³I‐FP‐CIT, according to the manufacturer's instructions. All the baseline DaT‐SPECT examinations were conducted before treatment was initiated (i.e., in a levodopa‐naïve state), and any other possible interfering medication was temporarily withdrawn before the SPECT. For thyroid protection, an oral dose of 200 mg of potassium perchlorate was routinely administered before and after the DaT‐SPECT. The effective dose from 185 MBq of ¹²³I‐FP‐CIT is approximately 4.4 mSv.

Gamma Cameras and Image Acquisition

The SPECT acquisition was made with two different gamma cameras at the Umeå University Hospital. Sixty‐four of the DaT‐SPECTs were done with a three‐headed brain‐dedicated SPECT camera (the Neurocam [NC]; General Electric, Milwaukee, WI). Image data were acquired in a 128 × 128 matrix with a pixel size of 2.0 × 2.0 mm. The camera had a fixed rotation radius of 12.2 cm, and by a 360‐degree step‐wise rotation in 128 equally spaced angles, the image data were collected for 45 seconds per angle. Image data were reconstructed with the filtered back‐projection technique without attenuation correction, using a Butterworth prefilter (cutoff, 0.45 cm⁻¹; power, 7).

The other 144 DaT‐SPECTs were done using a multipurpose two‐headed hybrid system equipped with a low‐dose CT, allowing reconstruction with correction for scatter and attenuation correction. Image data were acquired in a 128 × 128 image matrix with a zoom factor of 1.5. The resulting pixel size was 2.95 × 2.95 mm.

Image data were collected by a step‐wise rotation of 360 degres in 120 equally spaced angles for 30 seconds per angle. The distance between the detector and the center of the subject's head (i.e., the radius of rotation) was set to 15 cm and was kept as close to this distance as possible in each instance. The low‐dose CT images were used for calculation of individual attenuation correction. The additional effective dose from this low‐dose CT scan of the head is estimated to 0.1 mSv.

Image data were reconstructed using ordered subset expectation maximization reconstruction (two iterations and 10 subgroups), and with scatter correction, using the triple‐energy‐window method. Postfiltering was made using a Butterworth postfilter (cutoff, 0.45 cm⁻¹; power, 8).

Low‐energy general‐purpose collimators were used in both cameras, rendering a resolution (i.e., full width at half maximum) of 8.5 mm (NC; General Electric) and 9.0 mm (Infinia Hawkeye [INF]; General Electric), respectively, at a distance of 10.0 cm from the surface of the collimator. A 20% energy window was centered at the ¹²³I photon energy peak of 159 keV.

Image Analysis

Visual Interpretation

The visual interpretation was made in a Picture Archiving and Communication System workstation using the “GE color” scale as the standard, and the “Step‐10” color scale was sometimes used in addition. The visual assessment was made by the pattern and intensity of the uptake in the striatal region in the transversal plane and all slices were evaluated. The visual evaluation of all examinations was made blinded to the clinical status and diagnosis of the individual cases. Two of the authors (S.J.M. and K.R.), specialists in nuclear medicine, performed the visual assessment and reporting of each examination. The reporting was done according to the visual appearance of the uptake in the putamen and caudate in each hemisphere, as either normal or slightly, moderately, or severely reduced compared to a normal scan. The uptake in the putamen was decisive for the judgment of the examination as pathological or normal. Both observers evaluated each set of images, and consensus was met in all but 1 case. In 15 cases, the uptake in the putamen was reported to be only slightly reduced and was interpreted as being between normal and pathological.

In a total of 27 cases, including the 15 cases mentioned above, the DaT‐SPECT was either reported as intermediate between normal and pathological or there was an interobserver discrepancy, or the original interpretation of the examination was judged normal in a patient or vice versa. For the purpose of this study, a reassessment of these DaT‐SPECTs were made separately by the two observers, blinded to former reporting and the clinical details. After reassessment, a discrepancy remained between the observers in 10 of these cases. Therefore, in a second step, these were evaluated again by the two observers together, in order to reach a consensus.

Semiquantitative Analysis

Images were also analyzed semiquantitatively by one of the authors (S.J.M.) using a semiautomatic method using template regions of interest (ROIs), as described in earlier publications.22, 23 The semiquantitative evaluation was made blinded to clinical diagnosis and symptoms of the subject, and separate from the visual interpretation.

In short, the mean counts measured by separate ROIs fitted to the caudate, putamen, and entire striatum, respectively, were divided by the mean counts measured in the occipital region, used as the reference region. This ratio thus represents the specific uptake in each of these basal ganglia regions. In images from the INF gamma camera, the uptake ratios were corrected according to the method described in an earlier publication24 if the rotation ratio deviated from 15 cm.

Statistical Analysis

The statistical analysis was performed using IBM Statistics SPSS 20 (IBM Corp., Armonk, NY). Continuous data are presented as mean ± 1 standard deviation (SD). Averages of UPDRS‐III scores are presented as mean and median values. Sensitivity, specificity, and LRs were calculated based on the visual assessment of each examination. Mann‐Whitney's U test was used for comparison of ordinal data. A two‐sided value of P > 0.05 was considered statistically significant.

Results

The visual assessment of DaT‐SPECT images with uptake in the lower normal semiquantitatively measured range (i.e., 1–2 SDs below the normal mean) were subject to assessment uncertainty (as illustrated in Fig. 1). In this study, 10 of 171 DaT‐SPECT examinations done in parkinsonian patients were judged visually normal and 3 of 37 DaT‐SPECT examinations in HCs were judged as pathological. Among the 10 patients (8 PD and 2 APS) with a visually normal DaT‐SPECT, all but 1 had semiquantitative uptake ratios in the putamen that were within 2 SDs of the mean uptake in the control group (as presented in Fig. 2A,B). In this group of 10 patients, the median baseline UPDRS‐III score was 25.5 (minimum, 18; maximum, 42), without any statistically significant difference, compared to patients whose DaT‐SPECT was judged as pathological (median, 26.0; minimum, 5; maximum, 64; data missing in 2 cases; P > 0.05). Seven of these patients had more than 2 years of self‐reported duration of symptoms and 3 had less than 1 year of self‐reported symptoms. Nine of these patients had a clinical follow‐up time of 2 to 7 years, whereas 1 PD patient had a follow‐up time of less than 1 year (as shown in Table 1). In 6 of these 10 patients, 1 had a follow‐up SPECT after 1 year, 2 had a follow‐up SPECT after 1 and 3 years, 1 had follow‐up SPECTs after 1, 3, and 5 years, and 2 also had a DaT‐SPECT after 8 years. The remaining 4 patients were not followed up with SPECT. The follow‐up DaT‐SPECTs did not show any progress in 5 cases. In one of the patients, the follow‐up DaT‐SPECT after 1 year deteriorated, but then, after 3 years, the image was normalized. In 4 PD patients, the DaT‐SPECT was visually interpreted as abnormal, but the semiquantitative evaluation showed uptake ratios in the putamen bilaterally that were within the normal range (mean, −2 SDs).

Figure 1.

Baseline DaT‐SPECT in parkinsonian patients and HCs, illustrating the semiquantitative ratios in the putamen. Red triangles indicate indeterminate visual assessment or judgment discrepancy between observers. Green circles indicate visual assessment without uncertainty or discrepancy.

Figure 2.

Visually assessed DaT‐SPECTs (normal, green circles; pathological, red triangles) plotted against semiquantitative uptake evaluation in the left and right putamen. Each dot represents an individual, indicated by diagnostic entity. (A) Neurocam. (B) Infinia Hawkeye. APS, atypical parkinsonian syndrome (i.e., multiple system atrophy or progressive supranuclear palsy); HC, healthy control; PD, Parkinson's Disease.

Table 1.

Clinical and demographic summary for healthy controls with a pathological DaT SPECT and patients with normal DaT SPECT

Clinical Diagnosis (at Latest Follow‐up)	Baseline DaT‐SPECT Visual Evaluation		Duration of Clinical Follow‐up, Years	Symptom Duration at Baseline, Years	UPDRS III at Baseline	Follow‐up DaT‐SPECT Visual Evaluationa
	Normal	Pathological
HC		1	5	—	0	Normal
HC		1	5	—	0	Varying between normal and abnormal
HC		1	Only baseline	—	0	No follow‐up SPECT
PD	1		7	14.9	25	Varying between normal and abnormal
PD	1		7	5.2	28	Normal
PD	1		0.5	5.2	20	No follow‐up SPECT
PD	1		6	5.2	38	No follow‐up SPECT
PD	1		4	3.6	20	Varying between normal and abnormal
PD	1		3	5.2	39	Normal
PD	1		2	2.3	26	No follow‐up SPECT
PD	1		3	0.8	18	Varying between normal and abnormal
APS	1		5	0.8	21	No follow‐up SPECT
APS	1		4	0.7	42	Normal
Median			4.0a	4.4a	25.5a
Mean (±SD)			4.15 (±2.1) a	4.39 (±4.18) a	27.7 (±8.86) a
Min–Max			0.5–7a	0.7–14.9a	18–42a

The protocol‐based SPECTs were scheduled at baseline and 1, 3, and 5 years after baseline in patients and at baseline and 3 and 5 years after baseline in HCs.

^aPatients only.

HC, healthy control; PD, Parkinson's Disease; APS, atypical parkinsonian syndrome; Min–Max, minimum to maxium.

Among the 37 HCs, the DaT examination was judged as pathological in 3 subjects (as shown in Table 1). One of these had a follow‐up scan after 3 years and 1 had a scan after both 3 and 5 years; all follow‐up scans were judged as normal. The third HC only participated at baseline. In all 3 of the HCs with a pathological baseline DaT‐SPECT, the semiquantitative uptake ratio was within 2 SDs of the normal mean; 2 had ratios in the lower range and 1 in the higher normal range (as shown in Fig. 2B).

The resulting sensitivity for visual assessment of DaT‐SPECT in this group of early‐stage parkinsonian patients was 94% and the specificity 92%. This rendered a positive LR (LR⁺) of 11.75 and a negative LR (LR⁻) of 0.07. The proportion of false positives was 1.4%, and the proportion of false negatives was 4.8%.

Discussion

The accuracy of visual interpretation of the baseline DaT‐SPECT in PD was shown to be high in this work, but a small number of patients with clinical PD had normal findings on DaT‐SPECT. The term false negative may be disputable in this context, given that in 5 of the 10 parkinsonian patients with a visually normal baseline DaT‐SPECT, the follow‐up SPECT did not show any deterioration, and these may thus represent the so‐called subjects without evidence of dopaminergic deficit (SWEDDs).25 It is unlikely that the DaT scan would be normal in the majority of these cases if these symptoms were related to a presynaptic dopaminergic deficit, according to the theory that motor symptoms arise at the point when approximately 50% of the presynaptic neurons have degenerated in the SN.4 Nevertheless, these patients actually fulfilled the established clinical criteria for clinical PD (8 cases), possible PSP (1 case), and probable MSA (1 case) at the latest follow‐up clinical examination.

In an elderly population, the risk of cerebral small‐vessel disease is high, and this might be a cause of false‐positive ratings among cases without parkinsonism in this study, as well as in an unselected clinical setting. Findings in an earlier publication on structural MRI, conducted as a substudy within the same main research project, confirmed that degenerative changes were common.26

Our findings of 94% sensitivity and 92% specificity for the visual interpretation of 171 parkinsonian patients and 37 nonparkinsonian HCs are in line with earlier publications: sensitivity and specificity of 100% and 89% to 96%, respectively, in a one‐site study;27 a mean sensitivity for visual interpretation of ¹²³I‐FP‐CIT SPECT of 78% and mean specificity of 96.8% in a multicenter study;28 and in another multicenter study, including 158 PD patients and 27 patients with essential tremor and with the consensus blinded visual evaluation of ¹²³I‐FP‐CIT SPECT made by a panel, the sensitivity for PD was 95% and the specificity for essential tremor was 93%.29

Our earlier reported results for the sensitivity and specificity of the semiquantitative analysis of DaT‐SPECT made on two different cameras of >90%22 are in concordance with the visual evaluation of ¹²³I‐FP‐CIT SPECT.

In a situation where there is just a slight decrease in the putaminal uptake and uncertainty in the visual interpretation, a follow‐up scan is of value, if the clinical suspicion of degenerative parkinsonian disease remains. However, an issue is the entity of SWEDD. Earlier studies have suggested that patients with a repeatedly normal/nondeteriorating DaT‐SPECT and no further deterioration upon withdrawal of anti‐PD medication may not have a degenerative parkinsonian disorder.25, 30 In this study, there were at least 5 conceivable SWEDDs according to the result of the DaT‐SPECTs.

Even if the visual interpretation of a DaT‐SPECT examination is often fairly straightforward, with obvious putaminal uptake deficits, there will always be some challenging borderline cases. Nevertheless, this and previous studies show that DaT‐SPECT provides good support to the clinician in the diagnostic process, with a low risk of erroneously ruling out a parkinsonian disease when the scan is normal, and provides good support for such a diagnosis if the scan is abnormal.

Conclusion

Visual interpretation of dopamine transporter imaging with ¹²³I‐FP‐CIT offers reliable support in the early diagnosis of parkinsonian diseases.

Author Roles

(1) Research Project: A. Conception, B. Organization, C. Execution; (2) Statistical Analysis: A. Design, B. Execution, C. Review and Critique; (3) Manuscript Preparation: A. Writing of the First Draft, B. Review and Critique.

S.J.M.: 1A, 1B, 1C, 2A, 2B, 3A

J.L.: 1A, 1B, 1C, 2C, 3B

L.F.: 1A, 1B, 1C, 2C, 3B

K.R.: 1A, 1B, 1C, 2A, 2C, 3B

Disclosures

Funding Sources and Conflicts of Interest: This work was supported by grants from Västerbotten County Council (ALF), the Medical Faculty of Umeå University, the Swedish Medical Research Council, the Parkinson Foundation in Sweden, King Gustaf V's and Queen Victoria's Foundation, the Swedish Parkinson's Disease Association, and The Swedish Parkinson Foundation. The authors report no conflicts of interest.

Financial Disclosures for previous 12 months: S.J.M. has received support from Umeå University and Västerbotten County Council (ALF). J.L. has received support from The Swedish Parkinson Foundation, The Foundation for Clinical Neuroscience at Umeå University Hospital, Umeå University, and Västerbotten County Council (ALF). L.F. has received research support from the Swedish Medical Research Council, The Swedish Parkinson Foundation, The Swedish Parkinson's Disease Association, Umeå University, The Kempe Foundation, the Erling Persson Family Foundation, The Foundation for Clinical Neuroscience at Umeå University Hospital, and Västerbotten County Council (ALF); he has received honoraria from Orion Pharma. K.R. has received support from Umeå University and Västerbotten County Council (ALF).

Financial Disclosures related to this work: S.J.M. has received support from Umeå University and Västerbotten County Council (ALF). J.L. has received lecture and consultant honoraria from AbbVie, H. Lundbeck AS, Nordic InfuCare, IPSEN, and Abbott Scandinavia and research grants from the Swedish Parkinson Foundation, the Swedish Parkinson's Disease Association, the Medical Faculty at Umeå University, the Foundation for Clinical Neuroscience at Umeå University Hospital, and Västerbotten County Council (ALF). L.F. has received support from the Swedish Medical Research Council, The Swedish Parkinson Foundation, The Swedish Parkinson's Disease Association, Umeå University, and Västerbotten County Council (ALF). K.R. has received support from Umeå University and Västerbotten County Council (ALF).