Retinal damage thresholds from 100-millisecond laser radiation exposure at 1319 nm: a comparative study for rabbits with different ocular axial lengths

: With the widespread use of high-power laser systems in the wavelength spectrum between 1300 and 1400 nm, the risk of ocular damage becomes more serious and concerning. Existing ocular bio-effects studies have revealed unique damage characteristics, the damage mechanisms involved, and the trends of damage thresholds in this wavelength range. However, the influence of ocular axial length on retinal damage thresholds has not been investigated quantitatively. In this paper, using a 1319 nm continuous-wave laser, the in-vivo retinal damage thresholds were determined for two groups of chinchilla grey rabbits with the ocular axial lengths of 15.97 and 17.25 mm, respectively. The incident corneal irradiance diameter was fixed at 5 mm and the exposure duration was 0.1 s. The determined ED 50 values at 24-h post-exposure for the axial lengths of 15.97 and 17.25 mm were 1.06 and 1.79 J, respectively. Detailed analysis revealed that a sufficient margin existed between the damage threshold and MPE for adult humans, but for the newborn eyes, the safety factor may be less than 2.3.


Introduction
The transitional near-infrared (NIR) wavelength range refers to the spectrum band from about 1300 nm to 1400 nm [1]. In this region, ocular media absorbs laser energy very strongly, but sufficient energy can reach the retina to pose a retinal hazard [2]. With the wavelength increase in this band, the most sensitive tissue changes gradually from the retina to the cornea [3]. Owing to the high power outputs [4][5][6], the high transmission through the atmosphere and the relatively high penetration depth in tissues, lasers in this transitional NIR wavelength region have been applied in many fields, such as military [7] and medicine [8][9][10][11][12][13]. Some typical medical applications include the removal of turmeric tissues [8][9][10], nonablative skin rejuvenation treatment [11], neural stimulation [12] and pain research [13].
Considering the widespread use of laser systems in this region and the rapid increase of laser power/energy, the risk of ocular damage becomes more serious and thus receives more concerns [3]. The published ocular bio-effects studies in past several decades have revealed some unique damage characteristics, the damage mechanisms involved and the trends of damage thresholds [1][2][3][14][15][16][17][18][19][20][21][22][23][24][25][26]. The initial experiments from Zuclich et al. revealed some damage characteristics which are obviously different from those of the laser exposures in the visible or the far-IR wavelength ranges [1][2][3]14,15,[27][28][29][30]. Firstly, the absorption of incident laser energy in this region is more evenly distributed across the ocular media. Damage may be induced in one or more of the cornea, lens, and retina/choroid, depending on the precise exposure parameters. In contrast, the most sensitive tissue is the retina for visible and near-IR wavelengths up to the frequently studied 1150 nm emission, and the cornea for infrared wavelengths longer than 1400 nm. Secondly, retinal lesions at threshold level take 24 hours to become apparent and longer to reach maximum expression. Therefore, retinal ED 50 values were invariably lower when determined based on lesion/no lesion readings collected 24-h post-exposure than when using 1-h post-exposure readings. However, for lasers in the visible wavelength range, several studies found that the damage thresholds at 24-h post-exposure are significantly and consistently lower than values at 1-h post exposure in several specific conditions including that the exposure duration is in nanosecond regime [31] or the retinal spot size is smaller than about 80 µm [32]. And for retinal spot sizes larger than about 100 µm, there is no obvious difference between 1-h and 24-h endpoints. Thirdly, a retinal or corneal lesion involves the full thickness of that layer, even at the threshold level. This trait can also found in lesions induced by lasers in visible wavelength range, but only for suprathreshold exposures. Following the threshold exposure, the damage site is either at the retinal pigment epithelium for visible and near-IR wavelengths up to the 1064 nm emission, or the anterior corneal surface for far-infrared wavelengths. Finally, the retinal lesions which appear at 24 hours increase in size and stabilize after 48 hours post-exposure. However, extensive growth in lesion size is not a characteristic trait for lesions caused by laser radiation in visible wavelength range. Further experiments from Vincelette et al. [17][18][19], in combination with the results from Zuclich et al., showed that retinal damage thresholds can be determined in rabbit models at exposure durations up to 10 s, but in nonhuman primate models, no damage thresholds are determined for exposure durations longer than 0.08 s. Through detailed simulations and experiments, they confirmed that significant presence of thermal lensing effect, an unexplored mechanism for ocular damage, affects retinal damage trends and can explain the variance [20][21][22][23]. Wang et al. determined the corneal and retinal thermal damage threshold dependences on the exposure duration from 0.1 s to 10 s in rabbit model [24,25]. They found that the corneal or retinal damage thresholds can be correlated by power law functions, having similar trends with other visible or far-infrared wavelengths. Experiments from Jiao et al. showed that strong spot dependencies exist for this wavelength region [26]. At relatively large beam spot sizes at the cornea, threshold-level damage occurs at the back of the eye, involving the retina and the nerve fiber layers. While for relatively small corneal spot sizes, threshold-level damage occurs at the cornea.
Existing reports have given considerable attentions to the ocular damage induced by transitional NIR lasers, which promotes the revisions of laser safety guidelines and standards [33][34][35]. Actually, the most significant characteristic of transitional NIR lasers is the volumetric absorption of pre-retinal ocular media. This means that the visual axial length would be an important factor for the retinal exposures, and thus the eye with shorter axis length may be more susceptible to transitional NIR laser retinal damage. As shown by Gordon et al. [36], the newborn eye has a mean axial length of 16.8 mm, while the adult value is 24.2 mm [37]. This raises an interesting question: How will ocular axial length affect the retinal damage threshold quantitatively and whether the MPEs specified in safety standards are safe enough for humans with short ocular axial length.
Considering above analysis, an experiment was performed to quantitatively determine the influence of ocular axial length on retinal damage threshold in rabbit model, using a continuous-wave laser with the wavelength of 1319 nm. The results may contribute to the knowledge base for the damage trends and the setting of laser safety standard in the transitional NIR range.

Selection of the experimental animals
two equal groups according to the ocular axial length. Meanwhile, the animals were examined by a slit-lamp (Topcon, Tokyo, Japan) and an ophthalmoscope (Topcon, Tokyo, Japan) to insure clear ocular media and normal fundus. The axial length of the first group was 15.97 ± 0.30 mm (mean ± SE), and the value for the second group was 17.25 ± 0.22 mm, as shown in the Table 1. All the selected animals were maintained and used in accordance with the institutional guidelines of the Animal Care and Use Committee; and the ARVO Resolution on the Use of Animals in Research.  Figure 1 shows the experimental set-up. The laser employed for this study was a diode pumped continuous-wave Nd: YAG laser at the wavelength of 1319 nm (Fujian Institute of Research on the Structure of Matter, Chinese Academy of Science, Fujian, China). The maximum output of the laser was about 80 W, with power stability within ± 2% for 1 hour operating time. The full-angle beam divergence of the laser was 19.6 mrad and the beam propagation ratio (M 2 ) was about 45, measured according to the method in the ISO 11146-1 [38]. And for a Gaussian profile the criterion for the determination of divergence is actually equivalent to the 1/e 2 divergence. An electronically-controlled mechanical shutter was employed to control the exposure duration for the 1319 nm laser. The exposure duration selected for this experiment was 0.1 s. Using a beam splitter, a constant proportion of the laser power was reflected onto a laser power meter (3A, Ophir, Jerusalem, Israel) for monitoring the stability of the laser power. Another power meter (30A, Ophir, Jerusalem, Israel) was placed at animal eye's position to measure the power arriving at the cornea. A low-power 633 nm He-Ne laser, coaxial with the 1319 nm laser, was used as a pointer for retinal exposure position. A circular variable aperture was positioned to select the central portion of the beam and control the corneal spot diameter. The diameter at the corneal plane was adjusted to 5 mm. The animal was placed such that its fundus could be observed through a direct ophthalmoscope by the investigator with a 1319 nm mirror placed just between the ophthalmoscope and the animal. This mirror was with good optical transparency in the visible allowing for a clear fundus image.

Laser damage and data analyses
To ensure the animals did not experience pain and distress, all animals were anesthetized with an intramuscular injection of a mixture of ketamine hydrochloride (40 mg/kg) and xylazine (12 mg/kg). Full pupil dilation was performed with two drops each of proparacaine hydrochloride 0.5%, phenylephrine hydrochloride 2.5% and tropicamide 1% at a 5-minute interval. The eye was held open by a wire lid speculum and irrigated with 0.9% saline solution every few seconds throughout the procedure. The anesthetized animals were placed in a conventional holder where they were positioned with the aid of the He-Ne laser. The overlapping of retinal lesions was avoided by changing the incident beam angle. Following each exposure session, lesion/no lesion determinations were made by three experienced investigators for each exposure site at 1-and 24-hour post-exposure, with at least two of three in agreement to confirm a positive reading. The retinal lesions were examined with an ophthalmoscope (Topcon, Tokyo, Japan). The lesion/no lesion data were collected and analyzed using the SAS statistical package (Version 6.12, SAS Institute, Inc., Cary, NC). Bliss probit analysis was performed to determine the ED 50 threshold, fiducial limits at the 95% confidence level and probit slope (ED 84 /ED 50 ) [39]. The ED 50 refers to the effective dose corresponding to the 50% damage probability [39]. Table 2 shows all the experimental conditions and the corresponding damage probabilities for the 1-and 24-h observational time points. According to Table 2, the damage thresholds expressed in TIE (Total intraocular energy), the fiducial limits at the 95% confidence level and the probit slopes (ED 84 /ED 50 ) are analyzed and listed in Table 3.

Discussion
The rhesus is the most widely used animal model for the study of retinal damage threshold due to its eyes' close similarities to human eyes [36]. While the rabbit, a lesser species, is suitable for corneal damage studies and retinal damage trends studies [28][29][30][39][40][41]. For transitional NIR wavelengths, thermal-mechanism-dominated retinal damage thresholds have been determined for rhesus at exposure durations of 350 μs, 650 μs and 80 ms [2,17]. Experiments at longer exposure duration up to 10 s were also conducted but no retinal damage were found [1,2]. The rabbit is also used for retinal damage threshold study in this wavelength range. Retinal damage thresholds have been determined in chinchilla grey rabbits for exposure durations of 0.1, 1 and 10 s at the wavelength of 1319 nm, revealing the dependence of damage threshold on exposure duration [25]. Compared with the adult rhesus, the rabbit has a shorter ocular axis and the retina is more sensitive to transitional NIR laser damage [25]. Furthermore, for the transitional NIR lasers, retinal damage is mainly dominated by the water absorption, and the absorption of the retinal pigmented epithelium layer has relatively little effect [42]. From these points of view, the rabbit is more proper for the damage trends study in this wavelength range and avoids use of valuable non-human primates, following current trends toward the reduction, refinement, and replacement philosophy in animal research. Therefore, we used the rabbit model to quantitatively investigate the influence of ocular axial length on the retinal damage threshold. For the expression of the ED 50 values, two methods can be employed, including the radiant exposure on the cornea and the total intraocular energy (TIE) [33]. For the transitional NIR wavelength range, the radiant exposure is more suitable when comparing the risks of retinal and corneal damages, while the TIE is preferred when comparing retinal damage thresholds for different corneal spot sizes. In following analysis, the retinal damage thresholds from different reports would be discussed. Therefore, we used the TIE expression in this paper. Another important point to note, when comparing different reports, is that there are two methods for specifying the slope of the probit plot [39]. The first definition is the conventional mathematical meaning of slope, i.e., the change in probability of an effect divided by the change in dose, and termed the "real slope" (RS). The other method is to define the probit slope S as the ratio of the ED 84 value divided by the ED 50 value. The two definitions of slope are related by RS = 1/log S [39]. In determining ED 50 values, the probit slope reflects the total experimental uncertainty. A probit slope S = 1 corresponds to a real threshold, i.e., to a vertical line, and a larger value of S corresponds to a shallower slope. From Table 3, we can see that the probit slope is between 1.2 and 1.3, which is a good indication of the overall uncertainty and quality of the experimental data.
For 100-millisecond laser radiation exposures, the retinal damage threshold data depends on three fundamental characteristics, including the transmission of laser radiation to the retina, absorption of laser radiation by the retina, and the diameter of the laser-spot formed at the retina [18]. The retinal laser-spot diameter is influenced by various factors, such as the animal species, the accommodation state of eye, laser divergence and wavelength, and corneal spot size. In our experiment, only one animal species (chinchilla grey rabbits) was employed and thus the retinal absorption coefficients for the two groups of rabbits are same. The obvious difference for the damage thresholds, shown in Table 3, must be induced by two other factors. The laser energy incident on the retina is mainly determined by the pre-retinal absorption and only the directly transmitted portion contributes to retinal damage. Using onedimensional Beer's law, linear absorption coefficient of water (1.79 cm −1 at 1319 nm) [43], ocular axial lengths, and TIE thresholds at 24-h post-exposure shown in Table 3, the energy incident on the retina can be approximately predicted. According to this method, the estimated values for the energy incident on the retina are about 60.8 mJ for the group 1# and 79.4 mJ for the group 2#, respectively. Obvious difference could still be found for the two values and the retinal laser-spot size should be compared between the two groups of rabbits. As shown by Lund et al. [32], the retinal irradiance diameter (D) is dependent on the source angle (α) and the effective focal length of the eye in air (f e ) as D = αf e (assuming a relaxed eye). This equation is correct only for ocular accommodation to infinity. Basically, f e is relevant with animal species and the eye with a shorter axial length would have a smaller f e . For example, the typical f e values for the rabbit, monkey and human are 9.9, 13.5 and 17.0 mm, respectively. Above analysis only apply to the different animal species. For the same animal species, the relevance of the f e with the ocular axial length L is a determining factor to the explanation of the experimental results. Our additional experiment showed that the f e values were approximately 9.20 and 10.57 mm (measured assuming a relaxed eye and accommodation to infinity) for two rabbits with ocular axial lengths of 15.9 and 17.0 mm, respectively, which demonstrated that f e was positively related to the ocular axial length. However, one important fact should be noted that all the f e values above are apply to visible light. For the 1319 nm in this experiment, significant chromatic aberration will enlarge the retinal laser-spot size and this effect could be induced in both groups of the rabbits. Considering above analysis, the actual retinal irradiance diameter D should be smaller in the first group of rabbits (L = 15.97 mm) than in the second group of rabbits (L = 17.25 mm). Furthermore, according to Schulmeister et al. [32,44,45], retinal damage threshold ED 50 , expressed as intraocular energy, is proportional to D for 0.1 s exposures when D is less than D max corresponding to α max (100 mrad). Therefore, the damage threshold expressed in terms of energy incident on the retina should be smaller in the group 1# than that in the group 2#, which qualitatively explained the difference between the two values (60.8 vs. 79.4 mJ). However, the values of D could not be determined quantitatively due to the lack of the precise ocular parameters of the rabbits employed in the experiment.
In the report by Vincelette et al. [17][18][19], the damage threshold for 0.1 s exposure determined in rhesus model at 1319 nm was 1.37 J and the ocular axial length was about 19.5 mm, so the energy incident on the retina was about 41.8 mJ by employing the same calculation method mentioned above. Obvious difference could be found between the rabbit and rhesus. This may be induced by various factors. For 0.1 s laser radiation exposures, the retinal damage thresholds are determined not only by the transmission of laser radiation to the retina, but also by absorption of laser radiation by the retina and the laser-spot diameter formed at the retina. Comparing with the rabbits, the rhesus has a higher density of retinal pigments (especially in the macula), leading to higher retinal absorption. Other possible factors include the difference for the laser divergence and ocular refractive parameters.
As we know, the laser-induced tissue damage thresholds are normally obtained by exposing laboratory animals under controlled conditions to simulate "worst-case" human exposure conditions. Thus, we should analyze whether the MPEs specified in safety standards are safe enough for people with short ocular axial length. By using the data from Vincelette et al. [17][18][19], one-dimensional Beer's law, linear absorption data of water [43] and setting the retinal threshold energy of human equal to that of rhesus, the TIE for the newborn eye with mean axial length of 16.8 mm is estimated at 0.84 J for exposure duration of 0.1 s. While for the adults with mean axial length of 24.2 mm [37], the TIE value is about 3.18 J. In the IEC-60825-1 standard [35], the MPE value for 0.1 s at 1319 nm is 0.36 J. As shown in Table 4, a sufficient margin exists between the damage threshold and MPE for adults, but for the newborn eyes the safety factor is only about 2.3. One important fact should be noted that above analysis is very simple and rough. As indicated by the experiments for the rabbits, significant difference for the retinal laser spot-size exists between the rabbits having different ocular axial length. For newborn eyes, the ocular axial length is obviously less than the value of rhesus, thus the retinal damage threshold in Table 4 may be overestimated and the actual safety factor may be less than 2.3. A more accurate analysis for the damage of human eye should consider retinal spot size, laser parameters, eye chromatic aberration, corneal spot size and pupil diameter. Such a complex model has not been established at present.

Conclusion
The rabbit retinal damage thresholds at 24-h post exposure for the ocular axial lengths of 15.97 and 17.25 mm were 1.06 and 1.79 J respectively. The obvious difference for the damage thresholds resulted from the dependences of pre-retinal absorption and retinal spot size on the ocular axial length. Detailed analysis indicated that a sufficient margin existed between the damage threshold and MPE for adult humans, but for the newborn eyes the safety factor may be less than 2.3. The obtained results could be used in the refinement of the safety standards for transitional NIR lasers.

Disclosures
The authors declare that there are no conflicts of interest related to this article.