Introduction

Indocyanine green (ICG) solution distinctly stains the invisible retinal internal limiting membrane (ILM) of the human eye [5, 8]. ICG staining greatly facilitates the removal of the ILM during macular hole surgery, leading to a higher hole closure success rate and better visual acuity [1, 9].

Some investigators, however, have recently reported that the intravitreal use of ICG solution might enhance phototoxicity when used in combination with endoillumination, resulting in retinal or retinal pigment epithelial atrophy [3, 11]. The short-wavelength light emitted by endoillumination during vitrectomy is a potential source of photochemical injury; several cases of endoillumination-related retinal phototoxicity occurring during epiretinal membrane peeling have been reported [10]. Photochemical damage is produced by short wavelengths of visible light [6], and histological changes are observed in retinal pigment epithelium and photoreceptor cells [7]. Commercially available light sources for endoillumination during vitrectomy emit a wide range of light wavelengths from short (approximately 400 nm) to long (approximately 700 nm) [13], while indocyanine green exhibits strong absorption in the near-infrared part of the visible spectrum (approximately 805 nm in human plasma). Therefore, the absorbance of short wavelengths of light in ICG solutions is an important subject requiring investigation.

If indocyanine green solution exhibits a higher absorbance of short wavelengths between 400 and 500 nm than balanced salt solution (BSS), it may have an enhanced photochemical effect on the underlying sensory retina when it is applied to the ILM or retinal pigment epithelium during a vitrectomy. However, the accurate transmission characteristics of short wavelengths of light, emitted during endoillumination, in ICG solution or BSS remain unknown [4].

In this study, we quantified the absorption coefficients of short wavelengths in ICG solutions and assessed whether ICG solutions would absorb the blue light that is hazardous to photoreceptors and the retinal pigment epithelium.

Material and methods

The spectral distribution of light sources in ICG solution and BSS was determined using a standard fiberoptic endoillumination probe commonly used for vitreous surgery (Fiberoptic Endoillumination; Alcon, Dallas, TX, USA). ICG solution (0.1%, Daiichi, Tokyo, Japan) was prepared by diluting ICG in BSS.

Solutions were held in a vessel with a fused silica window at the base to maximize transmission. Above the vessel, a fused silica beam splitter was used to allow the simultaneous measurement of the initial and transmitted spectral radiation. The splitting ratio of the beam splitter was measured at each wavelength to allow the spectral radiation to be calculated.

The spectral distribution was measured with a resolution of 1 nm between 220 nm and 800 nm with a calibrated spectroradiometer (USR 40 V; Ushio Electronics, Tokyo, Japan). The measurements were repeated three times and performed at the power level recommended by the manufacturer. The output signal from the spectroradiometer was digitized and averaged by a personal computer with a digital board. The total absorbance of the sample and the sample holder in the sample chamber was measured, and the absorption coefficient of the ICG solution and the constituent components of BSS were measured by subtracting the absorbance of the sample holder from the total absorbance.

The Beer–Lambert law was used to determine the transmittance of each species contained in BSS as follows: T=I/I 0 =(1−R)2 (1−R')2e−αmCd, where I 0 and I are the light irradiation at the front and rear surfaces of the sample holder; R and R' are the reflectivity of the glass–ICG solution and the glass–air interfaces, respectively; αm is the molar absorption coefficient in moles per liter; C is the concentration in moles per liter; and d is the depth of the solution. The absorption coefficients (A) in the ICG solution and BSS were measured using the dual-beam configuration of the spectrophotometer and obtained by subtracting the absorption of the sample holder (T') from the total absorbance (T):

$$ {\rm A}\;{\rm = }\;{\rm log}\left( {T'/T} \right)\; = \;0.434\alpha \;{\rm m}\;Cd - 2\log \left( {I - R} \right) \cong 0.434\alpha \;{\rm m}\;Cd $$
(1)

In this manner, the molar absorbance coefficients in the constituent components of BSS were calculated using Eq. 1 or the short wavelengths irradiated by the light source. Finally, by summing the absorption coefficients at 400 nm and 450 nm from each species found in BSS, a total absorption coefficient for BSS at short wavelengths was obtained.

Results

Absorbance levels in the ICG solution were almost equal to those in BSS for all the spectra measured (Fig. 1). The absorption coefficients in the ICG solution were determined to be 0.18 cm-1 at 400 nm and 0.03 cm-1 at 450 nm. Using Eq. 1, the molar absorptions for each species at wavelength of 400 and 450 nm were calculated. The results obtained for each species in BSS are shown in Table 1. On the basis of these results, the absorption coefficients in BSS were obtained by summing the varying molarity of solutions, which were determined to be 0.17 cm-1 at 400 nm and 0.015 cm-1 at 450 nm. No significant differences in absorption coefficients at 400 nm and 450 nm were observed between BSS and ICG (Table 2, P>0.05).

Fig. 1.
figure 1

Dependence of absorption in indocyanine green solutions and balanced salt solutions for short (approximately 400 nm) to long (approximately 700 nm) wavelengths

Full size image
Table 1. Molar absorption coefficient (αm, M-1cm-1)
Full size table
Table 2. Absorption coefficient (cm-1). BSS Balanced salt solution, ICG 0.1% indocyanine green solution
Full size table

Discussion

The complete spectrum of short wavelengths irradiated by endoillumination light sources was examined in ICG solution and BSS. The results showed that the absorption coefficient of short wavelengths of light in ICG solution was almost the same as that obtained in BSS. A study by Dair et al. investigated the absorption coefficients of BSS for ultraviolet (193 nm and 213 nm ) wavelengths and reported them to be 140 cm-1 and 6.9 cm-1, respectively [2]. Although the absorption of ultraviolet wavelengths in BSS has been reported, no reports have described the absorption of short wavelengths in BSS. The absorption coefficients of short wavelengths may be less than those of ultraviolet wavelengths, since the longer wavelengths have lower absorption levels. In this regard, the previous studies may support our results for the absorption rate of short wavelengths.

Recently, ILM staining using ICG has been reported to result in better visibility and the easier removal of the ILM during macular hole surgery [1, 9]. This surgical technique is a very useful tool for the surgical management of macular hole, epiretinal membrane and macular edema. The possibility of retinal phototoxicity as a result of ICG injections during vitrectomies is, however, of great concern to vitreous surgeons. Our results have clinical relevance for retinal phototoxicity during the intravitreal use of ICG.

Our results show that the absorption coefficients of short wavelengths in ICG solutions were not significantly greater than those in BSS. They suggest that the intravitreal use of ICG may not result in the higher absorbance of short wavelengths emitted by the endoillumination sources; consequently, an increase in the transmittance of short wavelengths to the sensory retina or retinal pigment epithelium is unlikely. Thus, the use of ICG on the retinal surface probably does not lead to a greater penetration of short wavelengths causing retinal pigment epithelial atrophy. Moreover, it has been recently reported that the hypo-osmolarity of the ICG solution may be responsible for the toxic effect to the retinal pigment epithelium cell, rather than a photochemical effect [12].

Although ICG is physicochemically bound to the species of BSS in the clinical use of ICG during vitrectomy, ICG may also be bound to the plasma components when the breakdown of the blood–retinal barrier occurs, resulting in an alteration of the absorption spectra. Because we speculate that the effect of components other than BSS on the absorption spectra of ICG is minimal, these results may be applied in a practical clinical setting.

In conclusion, the absorption coefficient of wavelengths irradiated by the endoillumination light source in ICG solution is almost the same as that obtained in BSS, and the absorption of short wavelengths is not greater in ICG solution than in BSS. ICG solution placed on the ILM during intravitreal use probably does not enhance retinal photochemical injury during vitrectomy with ICG-assisted ILM removal.