one cone type in the first-order kernel are generally smaller in amplitude than those of the dichromats, indicating that the missing receptor type in these subjects may be replaced by a functioning photoreceptor.

The results of this experiment thus indicate that the first mfOP in the first-order kernel may stem from the additive activity of the L or M cones. This type of cone-specific stimulation can additionally give a very quick diagnosis of an L or M color vision deficit.

Congenital Stationary Night Blindness

In another experiment, we looked at the mfOP activity when rod function is altered [25]. Congenital stationary night blindness (CSNB) is a genetically heterogeneous retinal disorder of X-linked inheritance. In the complete form (CSNB1), rod responses to dim flashes are not detectable, whereas in the incomplete form (CSNB2, residual rod responses are found. The cone system is also affected in CSNB: There is a selective defect of the ON response in CSNB1 patients, leaving only the cone OFF responses intact [26]. The defect in CSNB2 patients, on the other hand, appears to affect both ON and OFF cone pathways, leaving only residual rod activity [26, 27]. In Fig. 4 we show a comparison of typical recordings from a 14-year-old patient with CSNB1, a 17-year-old patient with CSNB2, and a healthy 15-year-old control subject.

In the CSNB1 patient, the mfOP peak amplitudes of the first-order kernel response show a significant reduction of the first peak without significant reduction of the second, whereas in CSNB2 both peak amplitudes are severely compromised. In the second-order kernel, only the third potential is prominent in CSNB1 patients, and again none are discernable from noise in CSNB2 patients. Implicit times are not significantly altered.

The difference in mfOP amplitude between CSNB1 and CSNB2 patients probably reflects the different molecular mechanisms underlying the two types of disease, which differentially affect the postreceptoral pathways of cone signal processing. The wellpreserved peak 2 amplitudes of the first-order mfOPs and peak 3 amplitudes of secondorder mfOPs in CSNB1 patients point to a major impact of OFF pathway components on these responses that are not present in CSNB2 patients.

Thus, in the clinic, mfOP recordings are a quick way of distinguishing between the 2 CSNB types.

Taken together, these experiments suggest that the first peak in the first mfOP kernel is affected to a large extent by additive cone ON responses, whereas the second is heavily influenced by cone OFF responses. In the second-order kernel, the third peak is also dependent on cone OFF responses, whereas peaks 1 and 2 have a component based on cone ON responses. However, in all second-order kernel peaks, rod activity is also necessary to obtain potentials of normal amplitude [4].

Topographical Alterations

As for fOPs, mfOPs can be a sensitive indicator of inner retinal function. The advantage is that it is a sensitive method that may be able to detect local retinal changes, thus giving the potential of discovering pathological conditions early in their pathogenesis. Hypoxia of the retina plays a role in many retinal diseases and causes a reduction in the mean amplitude of mfOPs [28].

CSNB

Control

CSNB1

First order kernel

CSNB2

Second order kernel

1 nV/deg2

10ms

Fig. 4. Average of all 61 traces from one 15-year-old control subject (dashed lines), one 14-year-old patient with congenital stationary night blindness 1 (CSNB1; thin continuous lines) and one 17-year-old patient with CSNB2 (thick continuous lines). The first-order kernel is shown in the upper traces and second-order kernel in the lower traces. The second potential of the firstorder and the third potential of the second-order kernel response remain prominent in the recordings from the CSNB1 patient. No decipherable potentials were found in the recordings from the CSNB2 patient.

Diabetes

High blood glucose levels cause an altered metabolism in retinal cells. Alterations in the mfOP recordings of patients with diabetes mellitus can be found in the absence of any diabetic complications. In a group of 12 insulin-dependent (type 1) patients without evidence of diabetic retinopathy, most potentials were significantly delayed by 1–2 ms compared to those of an age-similar control group [15]. We show an example of this in Fig. 5. The typical results for a 20-year-old patient (continuous lines) who had had diabetes for 7 years are depicted alongside those of a healthy subject, also aged 20

Temporal

Superior

Nasal

Inferior

2 nV/deg2

10 msec

Fig. 5. Comparison of traces from a 20-year-old normal subject (dashed lines) with those of a 20-year-old diabetic subject with no clinical evidence of a retinopathy (continuous lines). The recordings are grouped into a central region and four retinal quadrants (see upper right). The diabetic subject showed delayed potentials throughout the retina compared to those of the control.

years (dashed lines). We show an example of a quadrant analysis of the retina, which also illustrates the nasotemporal and superior-inferior asymmetries in the second -order responses (right) as discussed earlier.

These small early delays cannot easily show local topographical alterations. Other studies have looked at some measure of the summed amplitudes of all potentials and have detected local areas of reduced function in diabetic patients without retinopathy

[29] and reduced amplitudes along with implicit time delays in peripheral regions in patients with nonproliferative diabetic retinopathy (NPDR) [29, 30]. Bearse et al. [29] have also been able to detect abnormalities of the second-order kernel, which are associated with local retinal sites containing NPDR. Panretinal photocoagulation reduces amplitudes significantly in peripheral regions [30].

The results of recording mfOPs in diabetic patients indicate that there is an early alteration in retinal sensitivity causing a short delay in most potentials. A longer implicit time has been shown to be caused by decreased rod activity [4], indicative perhaps of an impaired rod-cone interaction at this early stage of the disease. Even before the development of retinopathy, the method can provide information about local areas of dysfunction and can therefore serve as a method for detecting early changes in retinal metabolism.

Retinal Vessel Occlusion

Topographical alterations in the summed mfOP amplitudes from mfERG waveforms have also been shown in patients with branch retinal artery occlusion [31]. Again, decreased amplitudes were found in affected areas, which corresponded to perimetric measurements of sensitivity. MfOP recordings can further differentiate between patients with ischemic and non-ischemic branch retinal vein occlusion [32], demonstrating the importance of an intact retinal circulation in their formation.

Glaucoma

In the paragraph describing the mechanisms underlying mfOP generation, we saw that there is evidence for a ganglion cell component in the mfOPs that is responsible for the nasotemporal asymmetry in mfOP amplitude. Glaucoma patients suffer from ganglion cell damage, often caused by an increased intraocular pressure, and first experiments indicated that the presence of this component can be used diagnostically in the clinic.

The selective loss of an oscillatory feature in the temporal retina has been shown in mfERG recordings from primary open angle glaucoma patients, which may indicate loss of the optic nerve head component [33]. However, it cannot be ruled out that the feature arises in the inner plexiform layer. Experimental glaucoma induced in monkeys also reduces the nasotemporal asymmetry as well as the amplitudes of high-frequency oscillatory potentials, derived from a Fourier fast transform of a slow-sequence mfERG, which are thought to be generated to a larger extent by ganglion cell activity than the low-frequency potentials [18].

MfOPs appear to be especially sensitive to retinal alterations caused by normaltension glaucoma. A scalar product of the mfOPs extracted from mfERGs recorded from normal-tension patients was shown to be significantly different from control patients in the central 7.5° and in the nasal field. The traces from patients with high-tension glaucoma, on the other hand, differed significantly from the control only in the central 7.5°. The mfOPs were able to detect 85% of the normal-tension glaucoma patients and 73% of the high-tension patients [34].

General Alterations

Whereas the previous paragraphs have dealt mostly with alterations that affect specific potentials or with topographical alterations in mfOP recordings, there are