Effects of the lower energy and pulse stacking in carbon dioxide laser skin treatment: an objective analysis using second harmonic generation

ABSTRACT Purpose To evaluate the effect of fractional carbon dioxide (CO2) laser treatment using lower power associated with pulse stacking within collagen fibers, using second harmonic generation microscopy and computerized image analysis. Methods Twenty male Wistar rats aging eight weeks were used. Each treatment area received a single-pass CO2 fractional laser with different parameters. The 20 animals were divided into two groups and euthanized after 30 and 60 days. Second harmonic generation images were obtained and program ImageJ was utilized to evaluate the collagen organization within all areas. Collagen anisotropy, entropy and optical density were quantified. Results Increased anisotropy over time was observed in all four areas, but only reached statistical significance (p = 0.0305) when the mildest parameters were used (area four). Entropy decreased over time in all areas, but without significance(p = 0.1779) in area four. Density showed an overtime increase only in area four, but no statistical significance was reached (p = 0.6534). Conclusions When combined, the results obtained in this study regarding anisotropy, entropy and density tend to demonstrate that it is possible to achieve collagen remodeling with the use of lower power levels associated with stacked pulses.


Introduction
Skin resurfacing with carbon dioxide (CO 2 ) laser is still considered the gold standard treatment for facial rejuvenation. It has been used for this purpose since the early 90s with impressive results 1,2 . On the other hand, it presents a relatively high rate of drawbacks, as long downtime for recovery and risks for scarring and pigmentary disorders 3 .
Fractioning the laser beam with scanners was initially described by Manstein 4 with a 1500 µm laser prototype. In 2007, Hantash 5,6 described the CO 2 fractional laser. Its principle relied on creating an array of multiple micro areas of tissue vaporization (micro thermal zones -MTZ) while leaving unaffected skin around them. This allowed for faster re-epithelization and recovery time while yielding good clinical results 7,8 .
Although much safer, the fractionated mode can still present some complications, especially when higher fluences are used, as observed by Shamsaldeen 9 .
An alternative for safer fractional CO 2 laser treatments could be lowering the energy employed while stacking pulses at the same MTZ. This is already employed in clinical practice and usually delivers reliable results. An experimental study showed that using lower power associated with pulse stacking (consecutive pulses at the same location) can sustain higher macroscopic tissue contraction after 60 days compared to the use of high energy with a single pulse 10 .
In this study, we objectively evaluate this effect within collagen fibers using second harmonic generation (SHG) microscopy and computerized image analysis.
Collagen stands for the most abundant element of the extracellular matrix (ECM) and is responsible for maintaining skin tensile strength 11,12 . Due to its triple helix structure, which is not centrosymmetric, collagen is a very good SHG generator [12][13][14] and the resultant images can be evaluated by computational analysis.
This study evaluated anisotropy, entropy and optical density. Anisotropy usually quantifies the degree of collagen fibers alignment within the dermis 15 and can be used to study how it modifies, as skin ages or develops scars 16 . Entropy assesses the amount of disorderliness of a system and can be used to verify the randomness of an image. This was described previously for skin surface analysis 17,18 , as well as for the study of nerve aging 19 . Optical density is a well-known way to quantify the number of collagen fibers within an image 20 .
The aim of this experimental animal study is to use these collagen features obtained from SHG images in order to evaluate the dermal effects of using lower CO 2 laser power associated with pulse stacking.

Methods
The study was approved by the board of the Ethical Committee of Animal Research (protocol #3012-1).
Twenty male Wistar rats aging eight weeks were used. They were kept on a 12-hour light/dark cycle with free access to water and standard laboratory chow (3,100 kcal•kg -1 ). All animals were anesthetized with 80 mg•kg -1 ketamine plus 10 mg•kg -1 xylazine injected intraperitoneally and positioned on dorsal decubitus. Their abdomens were shaved and stamped with four 15 × 15 mm squares, 10 mm apart from each other to assure that one treatment area does not influence others. Then, the vertices of each square were tattooed for later area identification. Each square was assigned a number from one to four, as seen in the diagram (Fig. 1). Areas two, three and four were defined as treatment areas and area one was the control.

Groups for analysis
The 20 animals were divided into two groups. Group 1 (n = 10) was sacrificed by anesthesia overdose 30 days after laser irradiation.
Group 2 (n = 10) was sacrificed 60 days post-procedure. Each animal had all previously demarcated four areas collected for histological analysis.
Tissue specimens from both groups were fixed in 10% buffered formalin and embedded in paraffin. Then, a vertically cut 5 µm slice was obtained from each area and stained using the hematoxylin and eosin method.

Laser procedure
Each treatment area received a single-pass CO 2 fractional laser (Smartxide Dot; DEKA, Florence, Italy) with different parameters.
The 120 µm spot size was used, as well as a 500 µm spacing between MTZs for all areas in this study.
Other parameters, such as power, exposure time, stacking, fluence and energy per each MTZ, are summarized in Table 1.

Second harmonic generation image acquisition
Second harmonic generation images were utilized to evaluate the collagen organization within all four areas. Images were acquired with an inverted Z.1 Axio Observer microscope equipped with a Zeiss LSM780 NLO confocal scanning head (Carl Zeiss AG, Jena, Germany) at the National Institute of Photonics Applied to Cell Biology. All samples were evaluated according to a protocol previously described by Utino 21 . To obtain a complete image of the slide, we acquired tile scans (512 × 512) that were stitched in larger mosaics, as seen in Fig. 2.

Image evaluation
All images were analyzed using the free software ImageJ (National Institutes of Health, USA. http://www.imagej.nih.gov/ij) for collagen morphometric features. Collagen anisotropy, entropy and optical density were quantified. For each of these collagen features, a specific software plug-in is needed. All of them are already previously described for collagen analysis.
In order to measure anisotropy, we used the FibrilTool plug-in applied without any image pre-processing, as indicated by Boudaoud 22 .
Either for quantifying optical density and entropy, all images were split into color channels to obtain only the red channel, which is specific for the SHG signal.
To quantify collagen optical density, we clicked the "measurement" button under the "analyze" menu and the results were presented in the "results" window.
For entropy analysis, we used the grey level co-occurrence matrix (GLCM) texture analysis plug-in.
Each whole image was measured three times for each feature and averaged.
All data were tabulated on a sheet for further statistical analysis.

Statistics
The SAS system was used for statistical analysis. Areas within a group were compared using the Friedman test, while intergroup comparison was conducted by the Mann-Whitney test. The variables studied were anisotropy, density and entropy. The level of significance used in this study was 5%.

Results
All images were assessed for three collagen features: anisotropy, entropy and density.

Anisotropy
In group 1 (30 days), all treatment areas showed decreased anisotropy when compared to the control area, but without statistical significance (p = 0.7891).
In turn, in group 2 (60 days), we observed an increased anisotropy for treatment areas three and four when compared to the control area. On the other hand, anisotropy values were lower in the treatment area two than in the control area. Again, no statistical significance was found (p = 0.7014). These results are shown in Fig. 3. When comparing areas between the two groups, we show an increase in anisotropy overtime for all of them, including the control area. This increase was more evident in areas three and four, but only reached statistical significance in area four (p = 0.0305).
The results for anisotropy comparison between the two groups studied are represented in Fig. 4.

Entropy
In group 1, there were no statistical differences between treatment areas and the control area (p = 0.8497). The same pattern was observed in group 2 (p = 0.4551). These results are demonstrated in Fig. 3.
However, when comparing the groups, we noticed an overtime decrease in entropy for all areas. This decrease showed statistical significance for the control area (p = 0.004), area two (p = 0.0113) and area three (p = 0.0013). Area four also showed a decrease in entropy, but without statistical significance (p = 0.1779). These results are demonstrated in Fig. 4.

Density
In group 1, treatment areas two and three showed a slight increase in density compared to the control area. On the other hand, the treatment area four values were lower than the control area. However, no statistical significance was reached when comparing all areas (p = 0.4551) (Fig. 3).
Group 2 demonstrated that all treatment areas increased density against the control, especially treatment area four. Despite this major increase in density in the treatment area four, no statistical significance was reached when comparing areas (p = 0.2059) (Fig. 3).
In intergroup comparison, we observed a decrease in density in the control area (p = 0.9674) and treatment areas two (p = 0.6232) and three (p = 0.7337). Treatment area four, in its turn, showed an overtime increased density (p = 0.6534). No statistical significance was reached (Fig. 4).

Discussion
Our study investigated the possibility of using lower power whereas associating pulse stacking to achieve collagen remodeling. For this purpose, we analyzed SHG images with ImageJ free software, quantifying three collagen features: anisotropy, entropy and density. Together, they enable a structural evaluation of collagen changes after fractional CO 2 laser treatment.
Since the early 90s, the CO 2 laser has evolved to become the gold standard for facial resurfacing 1,2 . It is possible to achieve impressive results in a single session due to its capability of skin contraction e collagen remodeling 23 . However, the long downtime for recovery, the risks for pigmentary disorders and unaesthetic scars rendered CO 2 laser resurfacing a less useful tool for facial rejuvenation 3 .
The concept of fractional lasers emerged in 2004 4 and its applicability to the CO 2 laser was described by Hantash 5,6 . Since then, many authors have studied its ability to successfully treat photoaging 7,8,24,25 , as well as other disorders like hypertrophic scars 26,27 and acne scars 28 . The idea of deep ablating dermal tissue whereas leaving intact surrounding skin made it possible to deliver an excellent result while reducing significantly the risks and downtime 5,29 . However, achieving deep ablation usually requires higher power levels 30 . Despite its superior side effect profile over full ablative CO 2 lasers, the fractional mode still presents some drawbacks, especially when higher power levels are used 9 . Avram 31 advised for the risks of hypertrophic scaring when fractional CO 2 laser resurfacing is used on the neck.
In the search for reducing, even more, the overall risk profile of fractional CO 2 laser, an option could be reducing power levels while increasing the number of pulses delivered to each MTZ (stacking). Pulse stacking has been studied since long before the advent of the fractional laser. Fitzpatrick 32 , using a full ablative CO 2 laser, found a greater potential for scaring due to the increased thermal injury observed with pulse stacking. On the other hand, when using a fractional CO 2 laser, the use of consecutive pulses seems to be beneficial. Oni 33 demonstrated that, by doubling a pulse using half the energy, the tissue does not dissipate heat between pulses. It results in deeper and narrower columns of ablation with relatively wider zones of coagulation. This pattern could yield good results with less downtime than single pulses.
To investigate this concept, a previous study from our group 10 concluded that it is possible to achieve similar MTZ dimensions with lower power and pulse stacking. The same study showed increased tissue contraction overtime when these less aggressive parameters had been used. Prignano 34 , studying cytokine responses in tissue remodeling, also demonstrates that it is possible to achieve good biological results using lower power levels.
In order to corroborate our previous results, in this present experimental study we aimed to histologically characterize changes in collagen structure that could confirm the effects of using lower power and pulse stacking in collagen remodeling.
There are many methods to evaluate collagen within the dermis, including electron microscopy, biochemical and immunohistochemical analysis, among others 35 . A study conducted by Reilly 36 demonstrated that the molecular effects of both fractional and fully ablative CO 2 laser are very similar. This could explain the consistent rejuvenation obtained with the fractional mode. Other authors examined collagen behavior after CO 2 exposure employing specific stains 37 or even by describing collagen changes in simple hematoxylin-eosin stains 38 . These methods, although largely used, many times are subject to subjective analysis 14 .
To overcome subjectivity, during the last decade several optical methods also had been developed, as confocal laser scanning, optical coherence tomography, and multiphoton microscopy, especially SHG microscopy 39 . All these methods provide images that can be further analyzed by image software, resulting in a more objective way to assess collagen behavior 40 .
In this study, SHG microscopy was employed to analyze collagen response to a fractional CO 2 laser single treatment. Due to the collagen non-centrosymmetric molecular structure, it is an effective SHG generator 41 . The images obtained from SHG can be used to provide information on collagen structure within the skin 42 . Guo 43 used SHG to investigate skin rejuvenation after treatment with a 1550 µm fractional laser. They concluded that it is an appropriate technique to evaluate collagen regeneration by the fractional laser treatment. The same study observed that lower power treatment associated with a higher density of MTZ induces faster collagen regeneration.
The anisotropy index quantifies the degree of preferred alignment of collagen fibrils within the dermis 44 . This is important for understanding skin behavior. In this study, anisotropy was quantified using a plug-in from ImageJ software named FibrilTool. It was designed based on the concept of nematic tensor from the physics of liquid crystals to measure how well collagen fibrils are aligned. This enables image analysis without the complex processing needed with other techniques. Values obtained vary from 0 (random orientation -isotropic) to 1 (total alignment -anisotropic) 22 .
As skin ages, collagen fibers become more parallel 16,45 , and the same happens to the scar tissue 14 . This results in higher anisotropy values. Interestingly, the same pattern is present in newly formed collagen after photothermal treatments as observed by Wu 44 . Thus, we expected an increase in anisotropy for the laser-treated areas as indicative of neocollagenesis. We did not observe this pattern 30 days after treatment. In fact, a decreased anisotropy compared to the control area was noted. Dainichi 46 also observed that an increased collagen alignment was present only after 35 to 58 days from laser treatment due to the initial collagen degeneration. This parallel arrangement of collagen fibers was also observed in vivo 90 days after fractional CO 2 laser treatment using confocal microscopy 47 .
However, after 60 days, the two areas treated with lower power and pulse stacking increased anisotropy compared to the control area. This pattern was not observed in the area treated with the highest power in a single pulse. Although these results did not reach statistical significance, they could indicate that milder parameters can stimulate collagen synthesis, as observed by Prignano 34 .
We also compared anisotropy of all areas between 30 and 60 days. All areas including the control area increased anisotropy overtime. Concerning the control area, the increase, although not significant, is expected due to collagen alignment seen with aging 16 . The laser-treated areas experienced a greater increase in anisotropy from 30 to 60 days, but it only reached statistical significance (p = 0.0305) in the area treated with the mildest parameters. We hypothesize that these findings were detected only 60 days after laser treatment due to the prolonged effect on collagen structure seen after fractional laser treatments.
These long-term effects have been demonstrated immunohistochemically to last for up to six months 6,48 . Our results show a greater increase in anisotropy using less power and pulse stacking are in line with other authors. Yuan et al. 28 demonstrated that lower fluences can provide a significant efficacy for acne scars whereas Prignano 38 found no advantage, both clinically and histologically, to use treatment with 4.15 J•cm -2 laser irradiation compared to those obtained with 2.07 and 2.77 J•cm -2 .
Entropy is one of the features of the textural analysis, known as the GLCM, an effective method for quantitative analysis of skin texture 49 . It can be used to assess the level of randomness of a digital image 18 and has been associated with the degree of fiber organization 21 . We used entropy to quantify collagen structure behavior in the SHG images obtained after CO 2 fractional laser irradiation. With aging, the skin (and consequently collagen fiber arrangement) loses randomness, and entropy values decrease 19 . In this way, CO 2 fractional laser treatment ideally should slow down this collagen arrangement loss of complexity. Analyzing entropy 30 days after laser irradiation, we noticed almost no differences among all areas. In turn, after 60 days, it was possible to observe an increased entropy, although not statistically significant, for the treatment areas two and four. These are, respectively, the areas where the highest and the lowest laser parameters were employed. Important data was found when we analyzed entropy from a temporal perspective. From 30 to 60 days after laser treatment, all areas including the control area showed decreased entropies. These results confirm the observations of Silva 19 , who demonstrated a loss of randomness with aging. The only area in which the entropy decrease was not statistically significant (p = 0.1779) was the area where de mildest parameters were employed. This result could indicate that by using lower power and associating stacked pulses, it is possible to slow down entropy decline overtime. To our knowledge, no previous study has evaluated collagen entropy on SHG images after fractional CO 2 laser irradiation.
Integrated density is one of the most used parameters in SHG image analysis and an excellent form of collagen quantification. Its measures (the product of area and mean gray level) can be associated with the number of collagen fibers 20 . Collagen density decreases with aging, as it has been demonstrated both using regular stains 50 and SHG microscopy 51 . In our study, almost all treated areas showed an increased collagen density when compared to the control area, but without statistical significance. But when density was analyzed over time, we observed that only the area where the mildest parameters were employed (area four) had an increase, whereas all the other areas showed the expected decreased density with aging 39,51 . These results also did not reach statistical significance, probably due to the small sample studied. Nevertheless, it still might indicate that it is possible to induce neocollagenesis using less aggressive laser settings.
It has been already demonstrated that fractional CO 2 laser is clinically effective 52 . Tierney 53 showed a 65.3% mean clinical improvement in lower lid laxity six months after two to three treatments. Another study with the same equipment used in this paper found a 52.4% mean improvement for overall cosmetic outcome after two to three sessions treating moderate to severe photoaging 54 . However, most of these studies show some degree of subjectivity when evaluating the results obtained.
Attempts to relate the clinical effects obtained with the application of fractional CO 2 laser to changes in the dermis were also carried out with different methods. For example, Ozog 37 used the Herovici stain to differentiate types I and III collagen using computation analysis of digital images evaluated after treating mature burn scars. Prignano 38 provided a clinical to histological correlation using hematoxylin-eosin stains after a single treatment with a fractional CO 2 laser device. Even the molecular mechanisms underlying fractional CO 2 laser clinical outcomes were already studied 36 . The use of SHG signal to evaluate collagen responses to laser treatments was already studied by some authors 43,44 .
Our study, in its turn, uses SHG images to objectively evaluate collagen changes after fractional CO 2 laser irradiation with lower power and pulse stacking. Although not conclusive, our results show a tendency that is in line with other authors 33,38 . Besides that, our findings corroborated our previous study 10 on the macroscopic skin tightening observed using the same concept.
Further studies should investigate this theory within the human dermis to determine ideal laser settings balancing good results and low risks. In this way, SHG microscopy provides quantifiable measures for evaluation and can also be applied in vivo without the need for skin biopsies.

Conclusion
The combined results regarding anisotropy, entropy and density tend to demonstrate that it is possible to achieve collagen remodeling with the use of lower power levels associated with stacked pulses.

Authors' contribution
Intellectual and scientific content of the study: Motta

Data availability statement
Data will be available upon request.

Funding
Not applicable.