Journal of Neurology

, Volume 261, Issue 4, pp 791–803

Do eye movement impairments in patients with small vessel cerebrovascular disease depend on lesion load or on cognitive deficits? A video-oculographic and MRI study

Authors

  • Elmar H. Pinkhardt
    • Department of NeurologyUniversity of Ulm
  • Hazem Issa
    • Department of NeurologyUniversity of Ulm
  • Martin Gorges
    • Department of NeurologyUniversity of Ulm
  • Reinhart Jürgens
    • Department of NeurologyUniversity of Ulm
  • Dorothée Lulé
    • Department of NeurologyUniversity of Ulm
  • Johanna Heimrath
    • Department of NeurologyUniversity of Ulm
  • Hans-Peter Müller
    • Department of NeurologyUniversity of Ulm
  • Albert C. Ludolph
    • Department of NeurologyUniversity of Ulm
  • Wolfgang Becker
    • Department of NeurologyUniversity of Ulm
    • Department of NeurologyUniversity of Ulm
Original Communication

DOI: 10.1007/s00415-014-7275-1

Cite this article as:
Pinkhardt, E.H., Issa, H., Gorges, M. et al. J Neurol (2014) 261: 791. doi:10.1007/s00415-014-7275-1

Abstract

Small vessel cerebrovascular disease (SVCD) is one of the most frequent vessel disorders in the aged brain. Among the spectrum of neurological disturbances related to SVCD, oculomotor dysfunction is a not well understood symptom- in particular, it remains unclear whether vascular lesion load in specific brain regions affects oculomotor function independent of cognitive decline in SVCD patients or whether the effect of higher brain function deficits prevails. In this study, we examined a cohort of 25 SVCD patients and 19 healthy controls using video-oculographic eye movement recording in a laboratory environment, computer-based MRI assessment of white matter lesion load (WMLL), assessment of extrapyramidal motor deficits, and psychometric testing. In comparison to controls, the mean WMLL of patients was significantly larger than in controls. With respect to eye movement control, patients performed significantly worse than controls in almost all aspects of oculomotion. Likewise, patients showed a significantly worse performance in all but one of the neuropsychological tests. Oculomotor deficits in SVCD correlated with the patients’ cognitive dysfunctioning while there was only weak evidence for a direct effect of WMLL on eye movement control. In conclusion, oculomotor impairment in SVCD seems to be mainly contingent upon cognitive deterioration in SVCD while WMLL might have only a minor specific effect upon oculomotor pathways.

Keywords

Small vessel cerebrovascular diseaseOculomotor functionVideo-oculographyCognitionMagnetic resonance imaging (MRI)

Introduction

Small vessel cerebrovascular disease (SVCD) is one of the most frequent vessel disorders in the aged brain in the industrialized world [1] and is one of the main risk-factors for neurological impairment in the elderly [2]. Neuropathology of SVCD shows multifocal lesions, ranging from lacunes and microinfarcts, often involving subcortical and strategically important brain areas (thalamus, frontobasal areas, limbic system), white matter lesions, and hippocampal sclerosis to multi-infarct encephalopathy and diffuse post-ischemic lesions [3]. The spectrum of neurological disturbance related to SVCD is manifold with extrapyramidal motor impairment and neuropsychological constraint in the first place. Patients frequently show gait disturbance with apraxia and instability that often is described as a feeling of dizziness. Cognitive decline, such as general slowing, exhaustibility, and apathy, together with depressive mood and increased irritability, also appears to be associated with SVCD lesions [4]. In a recent longitudinal study within the leukoaraiosis and disability (LADIS) project, Inzitari et al. [5] showed that age related changes in white matter independently and strongly predict rapid global functional decline in older people.

Another frequent, but not well understood, clinical symptom in patients with SVCD is oculomotor disturbance. In clinical investigations, patients show altered smooth pursuit with occasional saccadic intrusions as well as deteriorated voluntary and visually guided saccades. Still, it remains unclear whether these oculomotor deficits are a manifestation of cognitive constraint or reflect an impairment of the oculomotor pathway through basal ganglia.

Magnetic resonance imaging (MRI) has a crucial role in the diagnosis of SVCD. On brain imaging, discrete lacunar infarcts and more diffuse regions of white matter hyperintensities (WMH) or leucoaraiosis are seen as typical pathological findings [6]. Up to 70 % of the total population over 65 years have MRI evidence of WMH; however, the burden of WMH varies substantially across populations and disease states [7]. Attempts to quantify WMH have used either visual rating scales or volumetric analysis [8, 9].

Oculomotor disturbance in SVCD, unlike locomotion and cognition, has never been examined systematically, especially not in relation to quantified WMH lesion-load (WMLL) and neuropsychological testing. In patients with manifest subcortical vascular dementia, Rösler et al. [10] attributed impaired fixation and disinhibition of anticipatory saccades to an effect of general cognitive slowing. This proposition apparently disregards that especially the pathway from frontal areas to the colliculus superior via the basal ganglia is crucial for the executive control of oculomotor function; this pathway may well be affected by SVCD itself. Therefore, it seems of particular interest to examine the possible relationship between oculomotor performance, WMH load, and neuropsychological decline. Instrumentation based eye movement recordings have the capability to unravel subtle oculomotor as well as cognitive deficits before gait disturbance and/or cognitive impairment becomes clinically evident [11].

In the present study, we investigated whether vascular lesion load in various brain regions affects oculomotor function independent of cognitive decline in SVCD patients or whether the effect of higher brain function deficits prevails. For this purpose, we examined a cohort of 25 SVCD patients and matched healthy controls using (1) videooculographic eye movement recording in a laboratory environment, (2) computer-based MRI assessment of WMLL, (3) assessment of extrapyramidal motor deficits, and (4) psychometric testing.

Materials and methods

Patients and controls

Twenty-five patients with a diagnosis of SVCD (eight male, 17 female) and 19 control subjects (seven male, 12 female) without any known neurological or psychiatric deficits underwent (1) tests of oculomotor function and eye-head coordination, (2) morphological MRI and (3) neuropsychological assessment. The mean age of SVCD patients was 75 years (range 58–91 years) and that of controls 66 years (range 48–79 years). The study had been approved by the Ethics Committee of the University of Ulm (62/08), and was, therefore, performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All subjects had given their informed consent before being enrolled.

The clinical SVCD diagnoses were made by a certified neurologist based on the clinical presentation excluding cases of acute cerebral ischemia as well as Parkinsonism other than SVCD related. Patients were included in the present study if at least two of the following three scores indicated a pathology: mini-mental state examination (MMSE), Tinetti score of fall risk [12] and classification of white matter lesions in terms of periventricular (PVH) and deep white matter hyperintensitiy (DWMH) by grades 0–3 as suggested by Fazekas [8]. Data are summarized in Table 1. Controls were examined in the same way and all were found to be free of fall risk and cognitive impairment.
Table 1

Demographic and clinical data of the SVCD patients

Age

Sex

MMSE

Tinetti

Grade

PVH

DWMH

58

M

15

12

3

3

58

F

29

25

3

3

60

F

27

25

1

2

65

F

30

26

3

2

67

F

28

18

3

2

67

F

26

26

2

3

68

F

27

23

1

2

71

M

29

20

2

1

71

F

28

23

3

1

72

M

30

20

2

1

74

M

26

18

3

3

75

F

26

20

2

1

76

F

27

25

3

3

76

F

25

20

1

2

76

M

28

23

2

2

76

M

27

23

3

2

76

F

28

25

2

2

77

M

27

21

3

3

82

F

24

20

2

2

82

F

24

15

2

3

84

F

26

20

3

3

87

M

25

18

3

2

87

F

23

16

2

2

91

F

27

23

2

3

91

F

30

20

3

2

Age (in years), sex (M male; F female); Tinetti, Tinetti’s score of fall risk [12]; PVH and DWMH, grading according to Fazekas et al. [8] of periventricular and deep white matter hyperintensities observed in MRI

MMSE mini-mental state examination

MRI data acquisition and analysis

MRI data were acquired on a 1.5 T scanner (Symphony, Siemens Medical, Erlangen, Germany) to quantify WMH. Forty T2-weighted coronar slices (Fluid Attenuated Inversion Recovery/FLAIR, TR/TE 6180/112 ms) of 3.0 mm thickness, 0.45 × 0.45 mm2 in-plane resolution and 512 × 448 voxels matrix dimension were scanned. The T2-weighted images were analysed by an in-house software based on a semi-automatic image intensity analysis [13]. The threshold for selecting WMLL voxels had to be adjusted operator dependently as T2-weighted MR images intrinsically represent no absolute intensity values. In order to improve accuracy, WMLL was determined at an image zoom factor of 8 (Fig. 1). In order to identify lesion-related voxels, voxels within an operator-defined intensity range were identified slicewise and color coded depending on the anatomical localisation so that separate counts for frontal, parietal, temporal and occipital lobes could be obtained. Brainstem and cerebellum were also included in the field of view but were not included in our analysis as the attribution to specific structures was deemed unreliable. The number of detected voxels (total amount of voxels corresponds to total volume of lesion) in any brain region was identified as the WMLL of the respective region. WMLL was then calculated as percentage of the intracerebral volume (ICV). For that purpose, ICV was estimated from the same MRI recording with a threshold based technique according to previously reported methods [14, 15].
https://static-content.springer.com/image/art%3A10.1007%2Fs00415-014-7275-1/MediaObjects/415_2014_7275_Fig1_HTML.jpg
Fig. 1

FLAIR images of a 58 year old female patient with a WMLL of 54,959 mm3, corresponding to 4.41 % of her ICV. MMSE 15, Tinetti 12, PVH Grade 3, DWMH Grade 3. a Lesion detection at an image zoom factor of 8 (slice 18), original greyscale image and lesion detection. b Slicewise identification of white matter lesion load in one subject; white matter lesions marked in red

Oculomotor and skeletomotor measurements

Measurements took place in an optically and acoustically shielded room with subjects seated in a comfortable chair at the center of a white hemicylindrical screen with a radius of 160 cm. The screen carried pairs of adjacent red and green light emitting diodes (LEDs) placed every 5° up to ±40° in horizontal and up to ±15° in vertical direction which were invisible when not lit. In addition, loudspeakers were placed 90° to the left and right of the screen. The movements of both eyes were recorded with a video-oculography system (EyeLink I®, SR Research Ltd., Osgoode, ON, Canada) at a sampling rate of 250 Hz while the head was stabilized by a chin rest. Visually guided reactive saccades (RSac) were elicited by randomly lighting one of the red LEDs as a target, such that target steps of 5°, 10°, 20°, and 40° horizontal and of 5°, 10°, 15°, and 30° vertical resulted (32 trials in each direction); successive target steps always continued from the preceding target position and occurred at a randomly varying pace of 2.1–3.5 s. Subjects were instructed to refixate the target as accurately and as fast as possible. In the delayed saccade (DSac) task, subjects were to withhold their reaction to random target steps (similar amplitudes as with DSac; 16 trials per direction) until, after a randomly varying delay of 1.5–2.5 s, an acoustic “go” cue was sounded while the current fixation target was extinguished after a further 1.5 s. In the anti-saccade (ASac) task, a green LED was lit at eccentricities of 5°, 10°, 15°, 20° horizontal and 5°, 10°, 15° vertical, respectively, while the central red LED disappeared while the subjects were fixating; subjects then were to move their gaze immediately to the opposite empty half-field and to assume a position mirror symmetric to that of the green LED (16 trials per direction). Rapidly alternating voluntary saccades (RAVS) in horizontal and vertical direction were evoked by asking subjects to saccade as rapidly back and forth for 30 s between two red LEDs spaced 20° symmetrical about the primary direction. Smooth pursuit eye movements (SPM) were elicited by a red laser spot moving sinusoidally at frequencies of 0.125 and 0.375 Hz across the screen in either a horizontal (±20°) or a vertical (±15°) direction, resulting in horizontal peak velocities of 15.7°/s and 47.1°/s, and vertical ones of 11.8°/s and 35.3°/s. Eye-head coordination was examined by having subjects track random horizontal target steps of up to 80° with their heads free to move. Here, eye movements were recorded by DC-electrooculography using two bitemporal electrodes and two vertical electrodes placed below and above the right eye; horizontal head movement was recorded by a precision potentiometer coupled to a lightweight helmet worn by the subject. Two conditions were examined: natural gaze saccades (NGSac) where subjects were told to move, or not to move their heads, as if they observed an everyday scene, and head-pointing gaze saccades (PGSac) where subjects were to move their heads as rapidly and accurately as possible to the target so as to point their nose at it (32 trials at each condition).

Using an in-house software package, the following parameters were extracted from the eye and head movement recordings: RSac were characterized by (1) the peak velocity of saccades of 20° amplitude (read from fits to the main sequence), (2) the gain of responses to 20° target steps (amplitude of primary saccade/target step amplitude) and (3) their latency (average of all trials). SPM was characterised by (4) its gain (fundamental component of SPM velocity/target velocity). The fundamental component was obtained by identifying and cutting out saccadic velocity peaks and fitting a sinusoid to the remaining data. This procedure was repeated three times with the amplitude criterion for saccade detection being successively lowered until a limit of three times the RMS noise of the smooth epochs was reached. For ASac and DSac, error scores were obtained from (5) the percentages of prosaccades and (6) premature saccades (before “go”), respectively. For RAVS we counted (7) the number of gaze shifts containing saccades of at least 10° amplitude. All of the above parameters were calculated separately for the horizontal, upward and downward directions and are based on a cyclopean signal representing an average of the left and right eye recordings. For the head-free NGSac and PGSac, (8) peak head velocity as a function of gaze amplitude was determined so that the velocity of 30° head movements could be read from a fit to this function, and the (9) gain of the head contribution to 30° gaze shift was obtained from head amplitude/gaze amplitude.

Neuropsychological assessment and scoring

Detailed psychometric assessments regarding memory, attentional, executive, and visuospatial functions, premorbid verbal intelligence, and depression were obtained. Memory performance was analyzed by means of a validated German version of the neuropsychological assessment battery of the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) [16], using the subtests Wordlist Learning, Wordlist Delayed Recall and Wordlist Recognition. For the examination of attention and psychomotor speed, the Symbol Digit Modalities Test (SDMT) [17] and a German adaptation of the Stroop Test (“Farbworttest”, FWT) [18] were used. Assessment of executive functions included tests for verbal fluency (“Regensburger Wortflüssigkeitstest”, RWT) [19] and figural fluency (5-Point Test) [20]. Premorbid verbal intelligence was estimated with a German vocabulary test (“Wortschatztest”, WST) [21], and the mini-mental state examination (MMSE) [22] as a general screening tool for dementia was administered. Visuospatial functions were examined by means of the Hooper Visual Organisation Test (VOT) [23]. Depression was evaluated using Beck’s Depression Inventory as well as the long version of a German adaptation of the Center for Epidemiologic Studies Depression Scale (ADS-L) [24].

Individual results obtained with the various neuropsychological assessment methods were analyzed as follows: for CERAD subtest Wordlist Learning, the total sum of remembered words over three trials was taken (Score CERAD Learned). For subtest Wordlist Delayed Recall, two scores were calculated: absolute number of recalled words (CERAD Recalled) and percentage learned during last learning trial (CERAD Saved). For subtest Wordlist Recognition, the percentage of correct responses discriminated (CERAD Discrim) was computed. Psychomotor speed was measured as number of correct responses within 90 s in the SDMT and as time taken for simple color naming in the Stroop Test using FWT Table 2 (Stroop Simple). The time difference between the Stroop interference condition using FWT Table 3 and simple colour naming (Stroop Sim-Int) was taken as an indicator of susceptibility to interference. For the RWT verbal fluency subtests “P”-Words, phonemic category switch between “G” and “R”, animals and semantic category switch between sports and fruits, the number of correct words generated within 2 min were counted. To score the 5-Point Figural Fluency Task, the total number of patterns generated within 3 min was recorded (5-Point). Raw scores were taken for VOT, WST, MMSE and BDI.
Table 2

Oculomotor results, comparison between controls and patients

 

Controls

Patients

p (C vs. P)

Mean

SD

Median

Q95

Q05

Mean

SD

Median

Q95

Q05

SacGn

0.86

0.04

0.86

0.92

0.80

0.80

0.07

0.82

0.87

0.68

0.0024

SacPV

400

59

394

470

326

408

56

403

502

320

0.69

SacLat

273

27

271

318

236

332

63

319

447

261

0.0003

RAVS

57

11

58

74

41

41

13

41.5

61

20

0.0003

HdGn

0.69

0.15

0.69

0.91

0.47

0.69

0.18

0.70

0.91

0.41

0.86

HdPV

111

27

106

156

83

80

22

74

115

54

0.0003

SPM1

0.91

0.05

0.92

0.98

0.82

0.75

0.21

0.83

0.99

0.32

0.0002

SPM2

0.74

0.12

0.77

0.92

0.53

0.48

0.23

0.47

0.80

0.10

0.0000

DelErr

7

7

4

22

0

36

24

31

73

4

0.0000

AntErr

20.2

12.4

18.8

40.8

6.1

53

29

53

94

10

0.0001

SD standard deviation; Q95 and Q05 95% and 5% quantiles; p (C vs P) significance of difference between controls and patients (t tests except for Mann–Whitney U tests performed with DelErr and AntErr). SacGn saccade gain with target steps of 20°; SacPV peak velocity of 20°-saccades (°/s); SacLat latency of saccades elicited by target steps of 20° (ms); RAVS number of voluntary back and forth saccades between two fixed targets at 10° left and right within 30 s; HdGn gain of head component of head-free gaze saccades elicited by target steps of 30°; HdPV peak velocity of 30° head movements during head-free gaze saccades (°/s); SPM1 and SPM2 gain of sinusoidal smooth pursuit eye movements at 0.125 and 0.375 Hz, respectively; DelErr and AntErr percent errors in the delayed and anti-saccade tasks

Table 3

Results of neuro-psychological tests

Scores

Controls

Patients

Level of significance

N

MV

SD

Med

Min

Max

N

MV

SD

Med

Min

Max

CERAD

 Learned

18

22.8

3.8

23.5

14

30

21

17.6

5.2

17

10

29

**

 Recalled

18

7.7

2.3

9

3

10

21

4.2

3.1

4

0

9

***

 Saved

18

85.6

17.8

90

50

113

21

56.9

34.4

67.7

0

100

**

 Discr

18

97.5

4.6

100

85

100

21

88.6

15

95

55

100

– 

SDMT

18

46.7

8.3

48

32

63

20

28

12

29.5

7

47

***

Stroop

 Simple

18

21

3.9

21

16

30

20

29.2

10.9

27

15

65

**

 Sim-Intr

18

18.9

10.5

17

9

57

19

43.9

33.6

32

17

155

***

VF

 P-Words

18

16.1

7.9

16.5

4

34

20

10.8

5.4

10.5

2

24

*

 G-R

18

20.7

7.1

20

10

34

20

13.1

5.7

13

4

23

**

 Animals

18

35.6

10.4

36.5

16

51

20

23.5

8.1

23

11

39

***

 Spo-Fru

18

21.4

5.1

22

1

29

19

13.2

4.5

14

5

22

***

5-point

18

30.7

7.2

30.5

19

46

20

21

10.6

19.5

6

46

**

VOT

18

21.1

3.6

21.5

14.5

29

21

15.6

5.5

16.5

3.5

25

**

WST

18

33.4

5.3

35.5

22

40

21

27.1

 7.3

28

9

37

**

MMSE

14

29.6

1.1

30

26

30

25

26.5

3.0 

27

15

30

***

BDI

18

4.7

3.9

5.0

0.0

11

21

9.8

10.8

8.0

0.0

52

*

ADS-L

18

6.3

4.5

6.5

0

15

21

12.2

9.3

12

0

32

**

See Materails and methods for definition of scores

MV mean value, Med median value

Level of significance of difference between controls and patients (Mann–Whitney U test; * p < 0.05; ** p < 0.01; *** p < 0.001)

Statistical analysis

To reduce the number of parameters, we averaged oculomotor performance from similar experimental paradigms across stimulus frequencies and movement directions after having checked that they indeed were mutually correlated and exhibited similar differences between controls and SVCD. Thus, single measures of SPM gain, latency and gain of RSac, and number of RAVS were obtained. Parameters from functionally related paradigms (DSac and ASac; NGSac and PGSac) were averaged for the group comparisons between patients and controls. For the correlation analyses with WMLL and neuropsychological scores, the parameters in question from patients were transformed into z scores (i.e., deviations from their respective sample means expressed as multiples of σ) and then merged by averaging to yield compound z values as indicators of executive error rates and of head gain and velocity, respectively. Compound z values were also calculated for the various subscores of CERAD and of RWT. Finally, since preliminary analyses had indicated that the two scores of psychomotor speed and attention (SDMT and Stroop) were tightly correlated, they were merged by a similar procedure (the sign of the SDMT z values being inverted so that large z values corresponded to a bad performance on both tests) yielding the compound z-value iSDMT&Stroop as a score for processing speed.

Statistica 10 (StatSoft Europe GmbH, Hamburg, Germany) was used for statistical testing. To detect an effect of a group (SVCD vs. controls), all parameters were subjected to t-tests or Mann–Whitney U tests, depending on their frequency distribution. To detect possible relationships between MRI, oculomotor and neuropsychological data, we first inspected scatter plots of all possible pair-wise combinations of parameters to identify single outlying data points that would be a risk to substantially bias the correlation coefficient. Five such sub-threshold data were removed (three WMLL counts and two neuropsychological scores from four subjects). To search for relations with WMLL, partial correlation coefficients were determined which controlled for the effects of age and other possibly confounding factors. One-sided significances were noted to test the hypothesis that an increase of WMLL would be accompanied by a parallel degradation of oculomotor and cognitive performance. In order to identify patterns of possibly consistent correlations between WMLL, oculomotor function and neuropsychological performance in patients, a synopsis of the correlation results was constructed by entering the significance levels of correlation coefficients into a matrix using a 3-level colour code corresponding to p < 0.05, p < 0.01, and p < 0.001, respectively (cf. Fig. 4).

Results

White matter lesion load

The distribution of WMLL in controls and SVCD patients is depicted in Fig. 2. In comparison to controls, the mean cerebral WMLL of patients was larger by a factor of almost seven (2.81 % of ICV vs. 0.41 %; Mann–Whitney U test, p < 0.0001). All subjects with a WMLL of <0.8 % belonged to the control group, whereas all subjects with a value more than 1.4 % belonged to the patient group; four controls and seven patients fell in the range between these limits. WMLL in brainstem and cerebellum was low. A separate consideration of the various regions examined (frontal, parietal, temporal, occipital lobes) indicated very similar WMLL in frontal and occipital areas (0.16 and 0.17 % in controls vs. 1.03 and 1.02 % in patients, respectively), a lower parietal WMLL (0.06 and 0.69 %, respectively), and little temporal effects (0.01 and 0.08 %, respectively), indicative of a very similar variation across areas in both patients and controls. Using population averages of the relative share of lobular volumes in total ICV [25], a normalisation of the above values to lobar volumes indicated that in patients about 3.3, 3.8, 0.5 and 11.3 % of the frontal, parietal, temporal and occipital volumes, respectively, were affected while the corresponding values for controls were 0.5, 0.3, 0.06 and 2.9 %. Across subjects, the WMLL percentages of adjacent cerebral areas were tightly correlated among each other (frontal><parietal, p < 0.001; parietal><temporal, parietal><occipital and occipital><temporal, all p < 0.01), whereas the correlations frontal><occipital and frontal><temporal did not reach significance.
https://static-content.springer.com/image/art%3A10.1007%2Fs00415-014-7275-1/MediaObjects/415_2014_7275_Fig2_HTML.gif
Fig. 2

Distribution of white matter lesion load (WMLL) in controls and SVCD patients. Each symbol represents one subject. Avg, population average; med, median of population

Eye and head movements

The results from controls and SVCD patients are listed in Table 2 and depicted in Supplementary Fig. 1. As a group, patients performed significantly worse than controls in almost all aspects, i.e., saccade gain, peak head velocity, and smooth pursuit gain were lower as was the number of voluntary saccades made within 30 s, while saccade latency and the number of errors made in the delayed and anti-saccade tasks were larger (see Table 2 for error probabilities). Saccade peak velocity and head movement gain were the only parameters not to be significantly affected.

The disturbances underlying the reduced gain of SPM were of various natures. At the lower frequency of 0.125 Hz, some patients exhibited intrusions that interrupted epochs of either perfect pursuit (Fig. 3, upper left panel) or of pursuit that repeatedly required catch-up saccades. Intrusions mostly resulted from predictive saccades anticipating the target motion and subsequent returns to the target. At the higher frequency of 0.375 Hz, SPM was mainly disrupted by predictive saccades while the intrusion pattern disappeared as there was no time for return saccades; about half of the patients made large predictive saccades during at least a third of the target’s back and forth cycles (Fig. 3, upper right panel).
https://static-content.springer.com/image/art%3A10.1007%2Fs00415-014-7275-1/MediaObjects/415_2014_7275_Fig3_HTML.gif
Fig. 3

Sample recordings of pursuit eye movements (red) in a SVCD patient (upper) and a control subject (lower) in response to horizontal sinusoidal target movements (black) of low (0.125 Hz, left) and high (0.375 Hz, right) frequency. In terms of pursuit gain, both individuals ranked second worst in their respective populations. Note the high incidence of intrusions in the patient’s low frequency response, and epochs of perfect pursuit at other times (arrows); the intrusions result from saccades anticipating the target motion followed by returns to the target. The high frequency response shows only anticipatory saccades as there is no time for returns

Neuropsychology

As a group, SVCD patients performed significantly worse than controls on all but one of the neuropsychological tests (cf. Table 3 for significance levels). Notably, the patients also exhibited a more depressive mood than controls in terms of both BDI and ADS-L (Table 3).

Correlation analyses

Two partial correlation analyses of patient’s data were run, one controlling only for age and the second one in addition for processing speed (as reflected by iSDMT&Stroop) as well as for depressive mood (measured by ADS-L). The matrices of significant associations between WMLL, oculomotor parameters and neuropsychological scores are shown in Fig. 4. First note that the sign of the correlations in Fig. 4 depends on the type of scores being paired whereas, functionally, all significant correlations indicate a parallel deterioration of oculomotor and cognitive scores with increasing WMLL and among each other. As long as the results were controlled for age only, almost no, and if present only weakly significant, correlations of either oculomotor parameters or neuropsychological scores with WMLL were seen, whereas there were strong associations between nearly all oculomotor and neuropsychological data. Particularly clear were those between oculomotor on the one side, and depressive mood (ADS-L), verbal fluency (RWT), and psychomotor speed and attention (reflected by iSDMT&Stroop) on the other side (cf. Fig. 4, upper). Most of these associations disappeared when correlations were controlled for psychomotor speed and depressive mood in addition, while some correlations with WMLL now became evident (Fig. 4, lower). Notably, there emerged an association of the executive error rate with WMLL which appears to be consistent in that it applied to three areas (i.e., frontal, parietal and occipital) with frontal and parietal significance levels reaching p < 0.01. Likewise, there may be a consistent relation of verbal fluency to fronto-parieto-temporal WMLL.
https://static-content.springer.com/image/art%3A10.1007%2Fs00415-014-7275-1/MediaObjects/415_2014_7275_Fig4_HTML.gif
Fig. 4

Matrix of significance levels of correlations between oculomotor (OMOT) and skeletomotor (SMOT) performance, neuropsychological scores (NPSY) and white matter lesion load (WMLL) in SVCD patients. Upper panel, analysis corrected for age only; lower panel, corrected for age, depressive mood (ADS-L) and iSDMT&Stroop. All parameters except Hypokinesia (UPDRS III hypokinesia score), VOT (Hooper Visual Organisation Test), MMSE (Minimental State Examination), and Depression (ADS-L score) are compound values obtained either by averaging across movement directions and stimulus frequencies (SPM, gain of sinusoidal smooth pursuit eye movements; SacLat and SacGn, latency and gain of visually guided reactive saccades; RAVS, number of voluntary back and forth saccades within 30 s) or by averaging z values across related tasks or subscores (ExecErr, errors in DSac and ASac tasks; HdPV, peak head velocity during natural and head pointing gaze saccades; CERAD, subscores Recalled, Saved and Discriminated; VerbalFluenc, subscores correct P-words, G-R, Animals and Spo-Fru of verbal fluency test), or across different but tightly correlated scores (iSDMT&Stroop, average of sign inverted SDMT z values and Stroop z values)

Discussion

The present study was the first to examine oculomotor disorders associated with SVCD in detail. Its aim was to investigate if the clinically apparent eye movement deficits in SVCD are predominantly a consequence of deteriorated oculomotor pathways, caused by WMLL itself, or if they result from deficits of higher brain function in SVCD and, hence, correlate with cognitive decline. Taken together, the deficiencies of our SVCD patients in terms of WMLL, neuropsychological and oculomotor performance would seem to suggest that oculomotor impairment in SVCD is mainly contingent upon cognitive deterioration in SVCD while WMLL, as assessed in the present study, seems to have only a minor specific effect upon oculomotor pathways. We are aware of the methodological limitations of the present approach which, per se, cannot unambiguously separate cause and effect. Yet, the main challenge here is to understand why also the cognitive deficits apparently were almost unrelated to WMLL although WMLL would seem to ultimately cause these deficits—an issue to be considered further below.

There are no well-defined diagnostic criteria for SVCD on cliniconeurological grounds given that the clinical significance of WMLL is still incompletely elucidated [26]. Individuals with a virtually identical WMH burden present with a wide variance in cognitive and motor performance ranging from no complaints at all over subjective cognitive complaints and mild extrapyramidal signs to dementia and parkinsonism [4]. Hence, in order to make a diagnosis of SVCD, it appears reasonable to demand for the presence of cognitive and/or motor impairment in addition to WMLL. Accordingly, in the present study, SVCD was diagnosed if patients presented with either cognitive impairments and/or motor impairment in addition to a WMLL exceeding grade 1 in the score by Fazekas et al. [8]. On the basis of this definition of SVCD, the frequency distributions of cerebral WMLL (percentage of affected volume) exhibited only small overlap between patients and controls which occurred in a low range of WMLL between 0.8 and 1.4 %, while the large majority of patients exhibited much larger WMLL ranging from about 2 % of ICV up to 6 %. We conclude that the threshold where WMLL becomes clinically significant is low. This observation is in line with recent findings that the impact of vascular lesions in the elderly in terms of mild cognitive impairment is more important as assumed to date [27]. One caveat has to be acknowledged, though: controls were on average younger than patients by about nine years since it had proved difficult to recruit enough subjects of advanced age free of any neurological problems that could at least be possibly related to WMH.

Cognitive deterioration

Given that cognitive impairment was an inclusion criterion for patients to be classified as SVCD, the clear degradation in terms of neuropsychological performance observed in patients with respect to controls had to be expected. Indeed, if clinically manifest, cognitive decline associated with WMLL has been reported to find its expression in a predominant disturbance of executive control and deficits in attention and psychomotor speed with subsequent memory decline [28]. The present neuropsychological data were in accordance with this pattern in that they indicated clear deficits of attention and psychomotor speed (as reflected by SDMT and Stroop) which conceivably reflect frontal lobe dysfunction. In addition, also other aspects of cognition such as verbal fluency and declarative memory (probed by CERAD) were significantly impaired pointing to a malfunction of a multitude of cortical networks, caused most likely by the occurrence of WMH in all cortical lobes. The observed increase of the depression scores (BDI, ADS-L) in SVCD patients was in agreement with observations by Mueller and coworkers [29] in SVCD patients and neuroimaging studies that suggest a possible role of WML in late onset depressive symptoms in SVCD [30, 31]. Also, a causal relationship of WMLL to depression is suggested by studies of individuals with risk for familial depression [32]. For screening of cognitive impaiments and depression in future studies in SCVD, we would suggest to use other, possibly more appropriate, screening tools such as the Edinburgh Cognitive and Behavioural ALS Screen (ECAS) and the ALS depression inventory (ADI12), respectively.

Oculomotor deficits

In contrast to cognitive criteria, oculomotor parameters were not considered when assigning patients to the SVCD group. Yet, also an impairment of oculomotor performance could be expected, in agreement with clinical experience. SVCD in fact significantly affected nearly all investigated aspects of oculomotor control. The extent of these deteriorations (Supplementary Fig. 1) followed a pattern that is reminiscent of what has been described for Parkinson’s Disease (PD) [3335] featuring (1) visually guided reactive saccades with slightly reduced gain, a slightly longer latency, but normal peak velocity, (2) less voluntary back and forth saccades in a given time interval, (3) reduced sinusoidal pursuit gain, and (4) elevated error rates during delayed and antisaccade tasks. In PD, these deficiencies become more prominent as the disease progresses and, noticeably, some are seen only if cognitively impaired patients are included (e.g., increase of saccade latency [36]). A number of other neurodegenerative disorders, such as Alzheimer’s disease, also show qualitatively similar patterns of oculomotor deterioration whereas frontotemporal dementia appears to share only impaired pursuit and increased antisaccades errors with our SVCD patients [37].

A first, straightforward conclusion from the current data is, thus, that the pulse generating part of the brainstem circuitry for saccade generation [38] is unaffected in SVCD since saccades had normal velocity. On the other hand, oculomotor control merges a multitude of influences from cortical (mostly frontal and parietal) areas as well as from the basal ganglia with ascending and descending pathways to brainstem and cerebellum [38]. For example, the reduced gain of our patients’ SPM could result either from an affection of the brainstem and cerebellar structures controlling the pursuit motor signal or from impaired parietofugal target velocity signals or from fronto-striatal dysfunction leading to attentional deficits and interruptions of SPM by the release of inappropriate saccades. Many patients exhibited catch-up saccades, a pattern reminding one of theponto-cerebellar affect, but which could also result from impaired parietal target velocity signals. In other SVCD patients, SPM was interrupted by predictive saccades that anticipated the target’s excursions and that were followed, at low stimulus frequencies, by saccades returning to the target, thus, giving rise to a pattern of intrusions. This type of SPM pattern, as well as elevated error rates of delayed and anti-saccades has been linked to deficiencies of the prefrontal-striatal network [33]. Without additional cues, it cannot be ascertained whether the deficient inhibitory control underlying these behaviours was caused by a primary malfunction of the frontal cortex rather than by one secondary to a disturbed input from basal ganglia or by a basal ganglia impairment perse. Hikosaka and colleagues have demonstrated a gating function of the basal ganglia that could control input to the superior colliculus in a task dependent way [39]. Thus, on the basis of oculomotor data alone, a basal ganglia dysfunction in SVCD patients as the source of the increased executive errors rates in SVCD cannot be excluded a priori, especially since SVCD also affects patients’ basal ganglia [3]. Given the coexistence of forth and back connections (via the thalamus) between frontal lobe areas and basal ganglia [40] with direct fronto-tectal pathways, the difficulty of distinguishing between cortical and basal ganglia effects is still a challenge in many neurodegenerative conditions when searching for the reasons of altered functions.

Correlation between WMLL, oculomotor performance, and psychometry

A joint occurrence of oculomotor and cognitive impairment has been repeatedly noted in neurodegenerative disorders. In Alzheimer’s disease [41], fronto-temporal dementia [42] and bulbar amyotrophic lateral sclerosis [43], cognitive decline is accompanied by increased error rates during antisaccades, a deficit that has been attributed to dysfunction of the DLPFC. The present data confirm these observations. When correlations were corrected for age only, significant associations between all oculomotor parameters and most neuropsychological scores were found (Fig. 4). A consistent worsening of all oculomotor parameters and most neuropsychological scores—however, with the notable exception of processing speed—with depressive mood became evident. Thus, depressive symptoms could have interfered with both, patients’ oculomotor and cognitive testing and should be eliminated as a confounding factor [44, 45]. Moreover, in order to identify the possible contributions of WMLL and higher brain functions to oculomotor performance, we chose psychomotor speed and attention performance as represented by the compound score iSDMT&Stroop as a likely representative of these functions. The iSDMT&Stroop was chosen for its apparent independence from ADS-L according to the age corrected correlation matrix. Upon controlling for these factors, most of the associations between oculomotor and neuropsychological disappeared or weakened while a few associations of oculomotor and neuropsychological performance to WMLL were unravelled. However, in view of the relative low number of subjects, the threefold correction for confounding influences and the mostly low (p < 0.05) significance levels of these associations, a direct interference of WMLL with oculomotor or neuropsychological parameters should only be conjectured for executive control and verbal fluency which consistently exhibited associations with three of the four investigated cerebral areas.

Yet, the question then arises why also the degradation of neuropsychological scores appears to be essentially unrelated to WMLL in the age-corrected data although ultimately WML would seem to cause them. While another study also failed to demonstrate an association between the amount of leukoaraiosis and cognitive functioning [46], many previous studies report significant correlations between WMLL and various aspects of higher brain functions [5, 47, 48]. Several reasons can be invoked to explain this discrepancy. All of these studies have examined populations that were considerably larger than ours, thus, increasing the chance of unravelling also weak correlations. A weakness of the observed associations between WMLL and cognitive functioning is often acknowledged [49]. Moreover, many of the above studies (e.g., [47, 49]) have examined populations of randomly selected elderly subjects rather than focusing on clinically defined SVCD patients. This may have facilitated the detection of covariations with WMLL. In fact, when we approximated this situation by merging the control and patient data and thereby accentuated the two ends of the natural distribution of WMLL in elderly subjects, many significant correlations between WMLL and both oculomotor and neuropsychological were observed after correction for age. After the additional correction for ADS-L and iSDMT&Stroop, all oculomotor parameters then exhibited significant correlations with WMLL, quite in contrast to the few WMLL-oculomotor associations seen in Fig. 4. Most consistent was again the increase of executive errors with WMLL, but now also SPM degradation exhibited a higher association (p < 0.01) with parietal WMLL, pointing to a possible interruption of pursuit related parietofugal fibres.

These observations suggest that the deteriorations of oculomotor and cognitive functioning most probably depend on which fibres are hit by WMLL rather than the amount of fibres affected—especially corticofugal and long distance cortico-cortical association fibres rather than local U-fibres constitute the critical substrate [50]. Thus, the lack of a separate analysis of periventricular leukoaraiosis in the present study may also have contributed to the scarcity of significant associations of oculomotor and neuropsychological parameters to WMLL.

Limitations

Our exploratory study has several limitations. There is the relatively low number of subjects which was due to the fact that a thoroughly defined clinical sample had to be found which was not simply determinated by the presence of WMH in the MRI of elderly people but contained patients with a clinical correlate as described above. Patients with highly severe SVCD who might have pointed to clearer correlational analysis results could not be included due to their lack of compliance with the study protocol and especially the oculomotor tasks. In addition, the separation of patients and controls had to be done by clinical rather than MRI-based parameters—a fact which we consider a strength of our study, but which might be a matter of controversy with the definition of SVCD. In close context with this definition, the control group could not be matched for age. Future studies should validate these findings by contrasting SVDC patients with age-matched controls in larger samples. Finally, the study per se is a correlational investigation with several co-factors with the inherent difficulty of such studies to distinguish between cause and effect.

It could have been useful to focus WMLL analysis on areas that are specifically involved in ocular motor control (such as parietal or frontal eye fields). However, in order to accurately segment single lobar substructures, a spatial normalization, e.g., MNI normalization, where one anatomic structure definition is valid for all subjects under investigation, would have been necessary. Since a non-affine stereotaxic normalization would change absolute WMLL volumes (with a different scaling compared to lobar substructure volumes), a more specific categorization into substructures was not performed. As a further limitation, the analysis was limited to WM, i.e., fibres leaving and entering the cortical layers, including those in close vicinity to these layers. While juxtacortical lesions were identified by the data processing, lesions located fully in the cortical gray matter—which also occur in SVCD—could not be determined from the FLAIR images in sufficient accuracy.

Whether future investigations into the factors determining the oculomotor alterations observed in SVCD patients will benefit from advanced MRI techniques, such as diffusion tensor imaging (DTI), cannot be answered with certainty as yet. DTI has the potential to unravel also the microstructural integrity of white matter that appears unaffected with fluid attenuated inversion recovery imaging, and has been applied to elucidate the causes of cognitive decline in SVCD. Here, single studies support the hypothesis of a causal role of disconnection of white matter tracts in the pathogenesis of cognitive impairment [51]. However, the conclusions drawn from the additional information provided by DTI parameters are mixed, ranging from “limited” [52] to “promising” [49].

Summary

In summary, the results of this study show that a particular pattern of oculomotor deficits in SVCD exists. Correlation analyses suggest that these deficits are an expression of the patient’s cognitive decline. Paradoxically, but in keeping with clinical evidence, both the oculomotor and the cognitive deficiencies were only loosely related with WMLL. Yet, we do not exclude the possibility that such effects might be demonstrated in a larger sample with advanced methods (e.g., DTI) focusing on functionally relevant fibre tracts.

Conflicts of interest

There are no conflicts of interest to report for all authors.

Supplementary material

415_2014_7275_MOESM1_ESM.tiff (3.9 mb)
Supplementary Fig. 1. Pattern of oculomotor differences between SVCD patients (red) and controls (blue); median values with 90 % ranges (black bars). Note different scalings depending on parameter type (cf. Table 2 for exact values). Significant differences marked by * (p < 0.01) or ** (p < 0.001). SacGn, saccade gain with target steps of 20°; SacPV, peak velocity of 20°-saccades; SacLat, latency of saccades; RAVS, number of rapidly alternating voluntary back and forth saccades between two fixed targets at 10° left and right within 30 s; HdGn, gain of head component of head-free gaze saccades elicited by target steps of 30°; HdPV, peak velocity of 30° head movements during head-free gaze saccades; SPM1 and SPM2, gain of sinusoidal smooth pursuit eye movements at 0.125 and 0.375 Hz, respectively; DelErr and AntErr, percent errors in the delayed and anti-saccade tasks. Saccade and SPEM measures represent averages of horizontal and vertical movements (TIFF 4,004 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2014