Comparative Evaluation of R134a and HFO‐1234ze Cryogen Spray Cooling Using a Mouse Model With Controllable Epidermal Pigmentation

ABSTRACT

Objectives

Cryogen spray cooling (CSC) is critical for protecting the epidermis during laser dermatologic procedures. The widely used cryogen, R134a, has a high global warming potential (GWP), motivating interest in environmentally favorable alternatives such as HFO‐1234ze. This study evaluated whether HFO‐1234ze provides epidermal protection comparable to R134a during laser exposure and investigated the mechanistic basis underlying any differences in cooling performance.

Methods

Laser‐induced epidermal injury in conjunction with two cryogens was assessed in K14‐SCF Mc1r^e/e mice with pharmacologically controllable interfollicular epidermal pigmentation. Laser treatment sites were assigned fluences relative to baseline skin lightness (L*) to span subthreshold to suprathreshold injury conditions, with cryogen assignment randomized. Hypopigmentation at Day 7 served as the primary injury endpoint and was scored independently by three blinded raters. Surface cooling induced by each cryogen was measured separately using thin‐film thermocouples on a skin phantom, and subsurface temperature distributions were estimated using one‐dimensional transient heat‐transfer modeling.

Results

Mean hypopigmentation scores did not differ significantly between R134a‐ and HFO‐1234ze‐cooled sites across the full range of pigmentation and fluence studied (p = 0.46). Heat‐map analysis and regression modeling confirmed that hypopigmentation increased with fluence and darker pigmentation but was independent of cryogen type. Surface temperature measurements using skin phantom revealed that R134a produced more aggressive transient cooling than HFO‐1234ze, reaching approximately 10°C lower minimum temperatures. However, numerical modeling of skin cooling using heat flux inferred from phantom measurement showed that this surface advantage decayed rapidly with depth, resulting in ≤ 2°C–3°C differences within the viable epidermis.

Conclusions

Despite differences in surface cooling magnitude, R134a and HFO‐1234ze provided equivalent epidermal protection under clinically relevant conditions. These findings support HFO‐1234ze as an effective, lower‐GWP alternative to R134a for CSC in laser dermatologic applications.

1. Introduction

Cryogen spray cooling (CSC) has become an essential technique in dermatologic laser surgery for protecting the epidermis from non‐specific laser‐induced thermal injury. Introduced in the mid‐1990s, CSC enables selective epidermal cooling by applying a brief (milliseconds) cryogenic spurt immediately before laser irradiation, thereby lowering epidermal temperature without significantly affecting the targeted chromophores in the deeper dermis such as blood vessels or hair follicles [1, 2, 3]. By rapidly cooling the epidermis, CSC allows clinicians to use higher fluences with reduced risk of pain, blistering, or pigmentary alteration [4, 5, 6, 7]. Tetrafluoroethane (R134a) has historically been the most widely used cryogen due to its safety considerations, favorable boiling point at atmospheric pressure, and consistent spray formation [8, 9].

However, despite its clinical success, the continued use of R134a is challenged by environmental concerns. R134a has an exceptionally high global warming potential (GWP ~ 1300), contributing substantially to atmospheric greenhouse gas loading [10]. For comparison, the GWP of CO₂ as the reference gas is 1. International regulatory frameworks, including the Kigali Amendment to the Montreal Protocol, are driving the phase‐down of high‐GWP hydrofluorocarbons, prompting the need for environmentally sustainable alternatives. In addition to environmental considerations, these regulations are already being implemented regionally through restrictions on the production and use of high‐GWP refrigerants, a trend that is expected to affect long‐term availability, supply stability, and cost of R134a for clinical applications. Several hydrofluoroolefins (HFOs), including HFO‐1234yf (2,3,3,3‐Tetrafluoropropene) and HFO‐1234ze (trans‐1,3,3,3‐tetrafluoropropene), have emerged as promising candidates with GWP values < 1 and thermophysical properties broadly similar to R134a [11, 12, 13]. While HFO‐based cryogens offer clear environmental advantages, their slightly higher boiling points, lower vapor pressures, and altered spray dynamics raise questions about whether they can achieve the same degree of epidermal protection as R134a during laser exposure.

A major obstacle in evaluating alternative cryogens is the difficulty of assessing cooling efficacy and safety across diverse pigmentation levels. Human skin pigmentation strongly influences laser–tissue interactions, with darker skin exhibiting increased melanin absorption and a correspondingly higher risk of epidermal injury. Conventional animal models do not allow controlled modulation of epidermal melanin content, limiting the study of pigmentation‐dependent CSC performance. This gap is now bridged by the K14‐SCF transgenic Mc1r^e/e C57BL/6J mouse model which enables tunable modulation of epidermal pigmentation via topical forskolin stimulation to activate adenylate cyclase and increase intracellular cAMP to drive melanogenesis [14, 15, 16]. This model provides a unique opportunity to emulate human Fitzpatrick skin types in a reproducible small‐animal platform and to evaluate the protective performance of alternative cryogens under pigmentation conditions relevant to clinical practice.

Given the need for low‐GWP cryogens and the unique capabilities of this pigmentation‐controlled mouse model, the present study sought to directly compare the cooling performance and epidermal protective effect of R134a and HFO‐1234ze during laser irradiation. We evaluated transient surface temperatures, subsurface thermal profiles, and pigmentation‐dependent laser responses to determine whether HFO‐1234ze can serve as a functionally equivalent, environmentally responsible replacement for R134a in clinical CSC applications.

2. Materials and Methods

2.1. Alternative Cryogen – HFO‐1234ze

HFO‐1234ze (E‐isomer, trans‐1,3,3,3‐tetrafluoropropene) was evaluated as an alternative refrigerant with a low GWP as compared to R134a, the standard refrigerant used in CSC. HFO‐1234ze belongs to the hydrofluoroolefin class of refrigerants and has an ultra‐low GWP (< 1), representing a more than thousand‐fold reduction compared with R134a. This substantial environmental advantage has motivated its investigation as a candidate replacement in systems that traditionally rely on R134a, including dermatologic applications of CSC.

Table 1 summarizes the key thermophysical and environmental characteristics of R134a and HFO‐1234ze relevant to CSC [11, 12, 17]. The two agents differ primarily in environmental impact and volatility: R134a, a hydrofluorocarbon with a high GWP (~ 1300), has a lower boiling point and slightly higher latent heat of vaporization, resulting in somewhat stronger cooling potential. In contrast, HFO‐1234ze, a next‐generation hydrofluoroolefin with an ultra‐low GWP (< 1), exhibits a higher boiling point and modestly lower latent heat, but remains fully capable of generating rapid flash evaporation during CSC. Although HFO‐1234ze is classified as A2L (mildly flammable), the volumes used in dermatologic laser systems are small and pose negligible clinical risk [18]. These combined properties support its evaluation as an environmentally superior alternative to R134a for epidermal protection during laser procedures.

Table 1.

Key thermophysical and environmental properties of R134a and HFO‐1234ze.

Property	R134a (HFC‐134a)	HFO‐1234ze	Notes
Chemical class	Hydrofluorocarbon (HFC)	Hydrofluoroolefin (HFO)	—
Global warming potential (100‐yr)	~1300	< 1	Major motivation for replacement
Boiling point (1 atm)	−26.3°C	−19.0°C	Higher boiling point → slightly warmer spray
Latent heat of vaporization	~217 kJ/kg	~195 kJ/kg	R134a has a slightly higher cooling capacity
Safety classification	A1 (non‐flammable)	A2L (mildly flammable)	HFO has mild flammability but safe at CSC quantities

Abbreviation: CSC, cryogen spray cooling.

In this study, both cryogens were delivered using the same clinical laser, Mini GentleLASE (Candela, Marlborough, MA), a system widely used in clinical practice for laser hair removal and the treatment of vascular dermatoses, ensuring identical nozzle geometry, delivery configuration and timing. The canister pressure was matched at 120 psi (gauge) for both cryogens by adjusting canister temperature, resulting in different equilibrium canister temperatures but identical driving pressure during spray formation. This approach reflects the native operating principle of the clinical device, in which cryogen pressure is regulated thermally rather than by the introduction of inert pressurizing gases. Using temperature‐based pressure control ensured clinically relevant conditions without modifying the delivery hardware or spray composition. Any differences in epidermal cooling performance are attributable to the inherent thermophysical properties of each cryogen rather than differences in delivery hardware or operating parameters.

2.2. Mouse Model With Pharmacologically Controllable Epidermal Pigmentation

All animal procedures were conducted in accordance with the US National Research Council's Guide for the Care and Use of Laboratory Animals and the US Public Health Service's Policy on Humane Care and Use of Laboratory Animals. The study protocol was reviewed and approved by the University of California, Irvine's Institutional Animal Care and Use Committee (IACUC; Protocol No. AUP‐23‐018). All efforts were made to minimize animal suffering and to reduce the number of animals used.

K14‐SCF Mc1r^e/e C57BL/6J transgenic mice were used to evaluate the performance of CSC in conjunction with pulsed laser irradiation under controlled pigmentation conditions. The mice were originally obtained from Dr. John D'Orazio's laboratory, and subsequently bred and maintained in‐house. These mice exhibit melanocytes throughout the epidermis due to keratinocyte‐driven expression of stem cell factor (SCF), which recruits and maintains melanocytes in the stratum basale [14]. This contrasts with normal mice, in which melanocytes reside almost exclusively in hair follicles rather than the epidermis. A loss‐of‐function mutation in melanocortin 1 receptor (MC1R) shifts melanogenesis toward pheomelanin rather than eumelanin [19], producing a fair‐skinned phenotype analogous to lightly pigmented human skin. Because MC1R signaling normally elevates cAMP to promote eumelanin synthesis [20, 21] these mice can increase epidermal pigmentation when the level of cAMP is pharmacologically raised, such as through topical forskolin treatment [15, 16]. This combination of SCF‐dependent melanocyte retention and MC1R‐based pigment modulation provides a controllable epidermal pigmentation model suitable for studies of light–tissue interaction.

Gender‐matched K14‐SCF Mc1r^e/e mice (12–15 weeks old) were used for all experiments. Dorsal fur was removed under inhalational isoflurane anesthesia (1.5%–2%) using electric shears followed by brief application of a depilatory cream (Nair, Church and Dwight, Ewing, NJ) to eliminate residual hair. Because Nair can transiently irritate or disrupt the epidermal barrier, mice were allowed to recover for 1 day before initiating forskolin treatment. On the following day, mice were re‐anesthetized and 400 μL of forskolin (20%, PureBulk Inc., Roseburg OR) solution (80 μmoles in 70% ethanol and 30% propylene glycol) was applied to the dorsal skin using a micropipette and spread evenly with the side of the pipette tip to ensure uniform coverage. Treatments were performed once daily, 5 days per week, for 1–2 weeks depending on the desired pigmentation level. Prior to each application, fur was re‐trimmed if necessary to maintain direct contact between the forskolin solution and the epidermis.

2.3. Laser Exposure Procedure and Post‐Treatment Evaluation

Once each mouse reached the desired pigmentation level following forskolin treatment, as determined by colorimetry, experiments proceeded as follows. Dorsal fur was removed under isoflurane anesthesia as previously described. Five treatment spots were marked on the dorsal skin, two on each side and one near the tail, with an inner diameter of approximately 13 mm, as shown in Figure 1. These markings guided the placement of the colorimeter probe for color measurements, standardized the field of view for dermoscopic and camera imaging, and guided positioning of the laser handpiece during treatment. Using fixed, clearly defined boundaries helped minimize variability in optical measurements and laser targeting across sites and animals.

Figure 1.

Example layout of treatment sites with assigned cryogen on the dorsal skin of a K14‐SCF Mc1r^e/e mouse. White ink circles (~13 mm diameter) demarcate the areas treated with either R134a or HFO‐1234ze cryogen spray prior to laser exposure. Each numbered circle represents a distinct test site, with cryogen assignment randomized across positions to avoid location‐dependent bias. Scale bar: 5 mm.

The pigmentation level of each marked site was quantified using a colorimeter (CR‐400, Konica Minolta, Tokyo, Japan). High‐resolution images of each spot were also acquired using a dermoscope (3Gen, Dana Point, CA) and a digital camera (PowerShot A95, Canon, Tokyo, Japan) to document baseline appearance.

Based on the measured pigmentation values, laser fluence levels were selected to span a range from no visible injury to clearly detectable epidermal hypopigmentation, rather than to define a single damage threshold. Initial fluence estimates were guided by empirical correlations derived from prior human clinical experience; however, because absolute injury thresholds differ between murine and human skin, final fluence values were refined experimentally using an iterative escalation approach. A cryogen, either R134a or HFO‐1234ze, was then randomly assigned to each site to eliminate positional or pigmentation bias. Each treatment consisted of a 50‐ms cryogen spurt followed, after 30 ms delay, by a single alexandrite laser pulse (755 nm wavelength, 3 ms pulse duration). This interval mimics clinically relevant CSC timing and allows evaluation of cryogen‐dependent epidermal protection. A brief secondary cryogen spray (~10 ms duration) was delivered immediately after the laser pulse, consistent with the clinical device's standard operating mode to suppress potential hair‐induced sparks.

On post‐treatment Days 1, 3, and 7, mice were anesthetized and imaged using the same dermoscopic and photographic setup employed at baseline. For Day‐7 imaging, fur regrowth was removed with electric shears followed by brief application of Nair depilatory cream to fully expose the treatment sites. Images were processed in Fiji (ImageJ), and for each treatment site, baseline and Day‐7 images were compiled side‐by‐side with a corresponding threshold‐based segmentation map of the Day‐7 image that highlighted areas of reduced pigmentation. To generate a segmentation map, the Day‐7 dermoscopic image was converted to 8‐bit and thresholded using a lower threshold of 180 ± 5 to visualize the hypopigmented region.

Three independent researchers, who were blinded to the cryogen type and laser parameters, scored each treatment spot based on the degree of hypopigmentation observed in the Day 7 images. As shown in Figure 2, scores followed a semi‐quantitative scale of 0, 0.3, 0.7, and 1.0, where 0 represented no hypopigmentation and 1.0 indicated clearly demarcated and pronounced pigment loss. These scores were used to evaluate the relative epidermal protective efficacy of R134a versus HFO‐1234ze across different pigmentation levels.

Figure 2.

Illustrative dermoscopic images and hypopigmentation scoring workflow for three representative treatment sites. Top row: Treatment sites on Day 0, immediately before laser irradiation. Middle row: Corresponding Day 7 images showing varying degrees of hypopigmentation following CSC and laser exposure. Bottom row: Automated threshold‐based segmentation to highlight regions of reduced pigmentation (shown in red). These segmented areas were used to guide blinded visual scoring of hypopigmentation severity (0, 0.3, 0.7, or 1). Scale bar: 5 mm. CSC, cryogen spray cooling.

A total of 22 K14‐SCF Mc1r^e/e mice were used in this study. The full distribution of baseline skin lightness (L*) values were used to categorize three predefined pigmentation levels: 35.0–41 (heavy pigmentation), 41–47 (moderate pigmentation), and 47–53.9 (light pigmentation). Based on these ranges, six mice exhibited heavily pigmented skin, 11 displayed moderately pigmented skin, and five showed lightly pigmented skin.

2.4. Statistical Analysis

Statistical analyses were designed to address three clinically relevant questions [1]: whether baseline pigmentation (L*) and applied laser fluence were comparable between cryogen groups [2]; whether cryogen type influenced epidermal injury outcomes; and [3] whether any observed differences could be explained by pigmentation or fluence rather than cryogen choice. All analyses were performed using Python (v3.12).

To first confirm that cryogen assignment did not confound baseline pigmentation, fluence selection, or injury outcomes, values from treatment sites assigned to R134a or HFO‐1234ze were analyzed as independent observations. Baseline skin pigmentation (L*) was compared to verify equivalent pigmentation distributions between cryogen groups prior to outcome assessment. Before selecting the appropriate statistical test, the distributions of each parameter were examined for normality using the Shapiro–Wilk test and for homogeneity of variance using Levene's test. Outliers were assessed using the interquartile range method (1.5 × IQR). L* values met normality and equal‐variance assumptions for both cryogen groups and therefore were compared using a two‐sample Student's t‐test. In contrast, both hypopigmentation scores and fluence values displayed significant departures from normality and, for fluence‐contained outliers, these parameters were analyzed using the nonparametric Mann–Whitney U test. A two‐sided significance threshold of α = 0.05 was used for all analyses. For descriptive comparison, means, medians, standard deviations, and interquartile ranges were calculated for each parameter.

To evaluate whether fluence selection depended on cryogen assignment beyond its planned dependence on baseline pigmentation, additional inferential models were used. Because fluence was intentionally scaled with pigmentation during treatment planning, simple linear regression was applied to characterize the relationship between fluence and L* for each cryogen group separately, and to obtain correlation coefficients and model fit (R²). Analysis of covariance (ANCOVA) was used to formally test whether the slope or intercept of the fluence–pigmentation relationship differed by cryogen type, with fluence as the dependent variable, L* as the covariate, and cryogen group as the fixed factor. Model assumptions were checked using standard diagnostics, and statistical significance was assessed at α = 0.05.

To distinguish true equivalence in epidermal protection from a lack of statistical power, an equivalence analysis was performed using the two one‐sided tests (TOST) procedure. Since each mouse received treatment with both cryogens, hypopigmentation scores were first averaged per cryogen within each mouse, and paired differences (R134a − HFO‐1234ze) were computed. An equivalence margin of ±0.1 in hypopigmentation score was defined a priori to represent a biologically negligible difference, as this value is less than one‐third of the spacing between adjacent categories in the semi‐quantitative scoring scale. Equivalence was concluded if the 90% confidence interval of the mean paired difference lay entirely within the equivalence bounds and both one‐sided tests were significant at α = 0.05.

Inter‐rater reliability of the hypopigmentation scoring system was assessed using intraclass correlation coefficients (ICC) derived from a two‐way random‐effects model with absolute agreement. This analysis was performed to confirm that visual injury scoring was reproducible and independent of the individual rater. Because the study endpoint was defined as the average of the three raters' scores, the primary reliability metric was ICC(2,k), which estimates how reliably the mean of multiple interchangeable raters reflects the true underlying score. For completeness, ICC(2,1), the corresponding single‐rater reliability coefficient, is also reported to characterize the performance of individual raters. Both ICC values were calculated from all sites with complete scoring, and 95% confidence intervals were derived using standard F‐distribution methods.

2.5. Surface Temperature Measurement Using Thin‐Film Thermocouples

To quantify the cooling efficiency of each cryogen independent of laser irradiation, transient surface temperature dynamics of an epoxy skin phantom were measured during cryogen spray application using thin‐film thermocouples (TFTC) [22]. These measurements provided the experimental basis for subsequent heat‐flux inference and thermal simulations.

The TFTC used in this study was fabricated by magnetron sputtering of copper and nickel thin films directly onto the epoxy phantom surface, following established microelectronic thin‐film deposition procedures. The thermocouple junction had a thickness of approximately 2 µm and lateral dimensions of approximately 1 mm, resulting in minimal thermal mass and negligible thermal contact resistance between the sensor and the phantom surface. This configuration yields sub‐millisecond response times, enabling accurate measurement of the rapid surface temperature changes that dominate cryogen spray cooling and providing a reliable experimental basis for subsequent thermal modeling [22, 23].

The epoxy skin phantom (EP30‐3, Master Bond Inc., Hackensack, NJ, USA) was positioned horizontally during experiments. All measurements were performed at an initial phantom temperature equal to ambient laboratory conditions (22°C ± 1°C). Surface temperature signals were acquired using a high‐speed data acquisition system synchronized with cryogen spurt triggering. The measured surface temperature histories, T _s (t), were subsequently used for heat‐flux reconstruction and as inputs for numerical heat‐transfer simulations, as described in the following section.

2.6. Skin Temperature Estimation Using Heat Flux Inferred From Phantom Measurements

To account for differences in thermal properties between the epoxy phantom and human skin, we employed a two‐stage computational approach. First, time‐dependent surface heat flux was inferred from phantom measurements; second, this heat flux was applied to a bioheat transfer model to calculate skin temperature under spray cooling. This approach assumes that spray‐imposed heat flux is governed primarily by cryogen spray dynamics rather than substrate properties, an assumption supported by prior experimental and modeling studies of cryogen spray cooling [24, 25, 26]. Accordingly, the epoxy phantom is used to characterize the spray‐dependent forcing function q″(t), while the skin model responds to this imposed heat flux according to its own thermophysical properties.

Surface heat flux was inferred from measured phantom surface temperatures T _s (t), which were imposed as Dirichlet boundary conditions in a one‐dimensional transient heat conduction model:

$$\frac{\partial T}{\partial t}={\alpha }_{\mathrm{phantom}}·\frac{{\partial }^{2}T}{\partial z^2},$$

where α _phantom  = k/(ρc) = 0.843 × 10⁻⁷ m²/s is the thermal diffusivity of epoxy (k = 0.14 W/(m·K), ρ = 1019 kg/m³, c = 1631 J/(kg·K)) [26]. The 10 mm phantom was modeled with a convective boundary condition at the back surface (h = 10 W/(m²·K), T _∞ = 22°C). The equations were solved using an implicit Crank–Nicolson finite difference scheme with spatial discretization Δz = 6.7 μm and temporal resolution Δt = 0.01 ms, verified through grid convergence studies.

Surface heat flux was extracted from the computed temperature field using second‐order one‐sided finite‐difference approximation:

$$q″\left(t\right)=-k_{\mathit{phantom}}\frac{-3T\mathrm{₀}+4T\mathrm{₁}-T\mathrm{₂}}{2\mathrm{\Delta }z}$$

Here, T ₀ denotes the surface temperature at the phantom‐air interface, while T ₁ and T ₂ correspond to temperatures at the first and second subsurface grid points, located at depths Δz and 2Δz, respectively.

The inferred q″(t) was then used as a Neumann boundary condition to solve the same bioheat transfer equation over a two‐layer skin model consisting of a 50 µm epidermis with thermal diffusivity α _epi  = 0.949 × 10⁻⁷ m²/s (k = 0.34 W/(m·K), ρ = 1120 kg/m³, c = 3200 J/(kg·K)) and an underlying dermis with thermal diffusivity α _dermis  = 1.075 × 10⁻⁷ m²/s (k = 0.41 W/(m·K), ρ = 1090 kg/m³, c = 3500 J/(kg·K)) [27]. Freezing was not modeled because the simulated skin temperature remained above 0°C. The 500 μm skin domain employed an insulated back boundary and an initial temperature of 33°C. The same Crank–Nicolson scheme was used with Δz = 2 μm and Δt = 0.01 ms. Spatial resolutions were independently optimized for each domain based on grid convergence studies, with heat flux q″(t) varying by less than 0.5% upon further refinement.

All solutions were implemented in Python 3.12. Tridiagonal systems arising from Crank–Nicolson discretization were solved via LU decomposition. Code accuracy was verified against analytical solutions for heat conduction in semi‐infinite media. A 10 ms baseline period preceding spray onset ensured thermally consistent initial conditions.

3. Results

3.1. Pigmentation Levels and Fluences Applied

Baseline pigmentation levels at the treatment sites, quantified using a standard measure of skin lightness (L*), exhibited a broad and largely comparable distribution between sites cooled with R134a and HFO‐1234ze (Figure 3). Mean L* values were similar between the two cooling groups (R134a: 43.75 ± 5.76; HFO‐1234ze: 44.55 ± 6.35), with closely aligned median values (43.46 vs. 44.92). Although the distributions overlapped extensively, subtle differences were observed: the R134a group contained a modestly higher number of darker sites (L* ~37–41), whereas the HFO‐1234ze group showed a slight over‐representation of sites around L* ~50. These differences, however, were small relative to the full pigmentation range sampled, and a two‐sample t‐test confirmed no significant difference in baseline L* between groups (t = −0.68, p = 0.50). Thus, comparisons of cooling performance and hypopigmentation outcomes between cryogens were not confounded by baseline differences in skin pigmentation.

Figure 3.

Distribution of baseline skin lightness (L*) for treatment sites receiving R134a or HFO‐1234ze cryogen. Overlaid histograms show the distribution of L* measurements across all treated sites, stratified by cryogen type. The two distributions overlap substantially, indicating comparable coverage of pigmentation levels in both groups, with only minor differences in the relative frequency of some L* ranges.

Laser fluence values applied across treatment sites displayed broad and overlapping distributions for R134a‐cooled and HFO‐1234ze‐cooled sites (Figure 4). Mean fluence levels were similar between groups (R134a: 10.90 ± 3.75 J/cm²; HFO‐1234ze: 11.30 ± 4.06 J/cm²), and median fluence values were likewise comparable (9.85 vs. 10.29 J/cm²). A modest skew was observed in the lower‐fluence range: the R134a group contained a slightly higher number of sites treated with fluences below approximately 9 J/cm². This trend was expected given the mild over‐representation of more heavily pigmented skin within the R134a group, where lower fluence levels were intentionally selected to avoid overt epidermal injury. Despite this difference at the lower end of the distribution, statistical comparison using the Mann–Whitney U test demonstrated no significant overall difference in fluence between groups (U = 1323.0, p = 0.63). Accordingly, subsequent comparisons of cooling performance and pigmentary outcomes were not confounded by differences in delivered laser energy.

Figure 4.

Distribution of laser fluence applied at treatment sites cooled with R134a and HFO‐1234ze. Overlaid histograms illustrate the range and frequency of fluence values delivered in each group. Both cryogens exhibited broadly similar fluence distributions with substantial overlap, and statistical comparison indicated no significant difference between groups.

Laser fluence was deliberately assigned according to L* to span a range of exposures anticipated to fall below, near, and above the putative injury threshold. As expected, fluence showed a strong positive correlation with L* for both cryogens (R134a: r = 0.73, HFO‐1234ze: r = 0.80), and linear regression confirmed nearly identical slopes (0.48 vs. 0.51 J/cm² per L* unit) (Figure 5). This indicates that operator‐selected fluence scaled similarly with pigmentation regardless of cooling agent.

Figure 5.

Relationship between laser fluence and skin lightness (L*) for sites cooled with R134a or HFO‐1234ze. Scatterplots show fluence values delivered across the range of baseline pigmentation, with linear regression lines overlaid for each cryogen. Fluence increased with increasing L* for both groups, reflecting the experimental design in which lighter skin tolerated higher fluence. Regression slopes and intercepts were nearly identical, indicating no cryogen‐dependent shift in the fluence–pigmentation relationship. Statistical testing demonstrated no significant difference in fluence selection between the two cryogens.

To test whether the fluence–pigmentation relationship differed according to cryogen assignment, despite fluence being selected solely on the basis of L* without knowledge of which cryogen a site would subsequently receive, an ANCOVA model was fitted with fluence as the dependent variable, L* as a covariate, and cryogen group as the factor. The cryogen term was not significant (p = 0.999), and model R² was unchanged when the group factor was removed, indicating that cryogen type did not contribute to variance in fluence beyond what was explained by pigmentation alone. Effect size analysis (Cohen's d = −0.10) likewise confirmed that any difference in fluence between cryogen groups was negligible. Collectively, these statistical assessments demonstrate that operators applied comparable fluence for a given pigmentation level, with no systematic difference between R134a‐cooled and HFO‐1234ze‐cooled sites.

3.2. Skin Injury

3.2.1. Skin Injury Pattern

Representative images of the treatment sites illustrating the range of skin responses observed in this study are shown in Figure 6. Baseline images demonstrate the variation in induced pigmentation (L* ≈36–51), while Day‐7 images reveal the extent and spatial distribution of epidermal injury, specifically, hypopigmentation, under different combinations of pigmentation and fluence. In the darker‐skinned examples (L* ~36–37), relatively low fluences (6.0–6.2 J/cm²) produced mild to moderate hypopigmentation, consistent with the lower energy threshold expected for these sites. In contrast, the lighter‐skinned examples (L* ~48–51) tolerated markedly higher fluences (14.5–15.0 J/cm²), but also resulting in more conspicuous and well‐demarcated hypopigmented zones.

Figure 6.

Representative examples of Day‐7 epidermal injury across pigmentation levels, fluence settings, cryogen types, and mean pigmentation scores. Each column illustrates a treatment site with its corresponding baseline image (top row) and Day‐7 post‐treatment appearance (bottom row). Scale bar: 5 mm.

Across these exemplars, the pattern, severity, and morphology of injury were comparable between R134a‐cooled and HFO‐1234ze‐cooled sites when pigmentation and fluence were matched. These examples illustrate the broader dataset trend: hypopigmentation severity scaled predictably with both L* and fluence, and there was no observable qualitative difference attributable to cryogen type.

3.2.2. Inter‐Rater Reliability of Skin Injury Scoring

A total of 106 treatment sites contained complete scoring from all three raters. The averaged hypopigmentation score demonstrated excellent inter‐rater reliability, with an ICC(2,k) of 0.93 (95% CI: 0.90–0.95), indicating that the mean of the three raters provided a highly stable and reproducible estimate of laser‐induced hypopigmentation. The reliability of an individual rater, assessed using ICC(2,1), was 0.82 (95% CI: 0.77–0.87), reflecting good agreement at the single‐rater level and indicating that no individual rater introduced systematic bias or drift. The relatively small difference between ICC(2,k) and ICC(2,1) suggests that the raters exhibited similar scoring tendencies with limited systematic disagreement. Collectively, these findings demonstrate that the visual scoring method is robust, internally consistent, and suitable for quantifying differences in epidermal response across treatment conditions.

3.2.3. Dose–Response Characterization of Epidermal Injury for the Two Cryogens

After illustrating representative examples of laser‐induced hypopigmentation and establishing strong inter‐rater reliability in the visual scoring method, we next quantified outcomes across the full dataset to examine how hypopigmentation varied with fluence, baseline pigmentation, and cryogen type. As shown in Figure 7, laser‐induced hypopigmentation in the mice skin increased systematically with fluence for both cryogens and showed a secondary dependence on baseline pigmentation. The two‐dimensional heat maps (Figure 7, left and middle) illustrate the combined influence of fluence and L* on the severity of hypopigmentation with darker colors indicating greater laser‐induced hypopigmentation. Within each cryogen group, sites treated at higher fluence displayed progressively greater hypopigmentation, and this trend was consistent across the full pigmentation range. In both panels, the transition from minimal to pronounced pigment loss occurred at similar combinations of fluence and L*, indicating comparable injury thresholds between R134a‐cooled and HFO‐1234ze‐cooled skin.

Figure 7.

Mean hypopigmentation score as a function of baseline skin pigmentation (L*) and applied fluence for sites cooled with R134a (left) and HFO‐1234ze (middle), displayed as two‐dimensional heat maps. Higher scores (darker color) indicate greater laser‐induced hypopigmentation. Box‐and‐whisker plots (right) summarize the distribution of mean hypopigmentation scores for each cryogen group, overlaid with individual data points and group means. No significant difference in hypopigmentation scores was observed between cryogens (p = 0.455).

For darker skin (lower L*), hypopigmentation remained modest at lower fluence levels but increased sharply once fluence exceeded the pigmentation‐dependent threshold. In lighter skin (higher L*), the fluence required to induce similar pigment loss was higher, reflecting the expected influence of melanin content on epidermal laser energy absorption. Despite these physiological effects, the overall pattern of injury was nearly indistinguishable between the two cryogens: both demonstrated increased hypopigmentation with increasing fluence, and both produced comparable score distributions within corresponding L*–fluence regions.

The box‐and‐whisker plot (Figure 7, right) summarizes the mean hypopigmentation scores across all treatment sites by cryogen type. The score distributions overlapped extensively, with nearly identical medians and similar overall ranges. Although the interquartile range for R134a appeared narrower, approximately 60% of that observed for HFO‐1234ze, this difference did not reflect a statistically meaningful separation between groups. A Mann–Whitney U test, which does not assume equal variances, confirmed that hypopigmentation severity did not differ significantly between cryogens (U = 1517, p = 0.455). Thus, despite minor visual differences in distributional spread, both cooling methods produced comparable degrees of epidermal injury across the range of pigmentation and fluence levels evaluated.

To assess whether cryogen type contributed to differences in hypopigmentation after accounting for the two principal determinants of injury, fluence and baseline pigmentation, we fitted an ordinary least squares model with mean hypopigmentation score as the dependent variable and cryogen group, fluence, and L* as predictors. Cryogen type showed no significant effect on hypopigmentation (β = 0.050, i.e., switching from one cryogen to the other changed the mean score by only 0.05 units, p = 0.30). In contrast, hypopigmentation increased significantly with higher fluence (β = 0.060 per J/cm², p < 0.001) and decreased with lighter skin (β = ‐0.027 per L* unit, p < 0.001). These findings indicate that, after accounting for pigmentation and fluence, cryogen type did not measurably influence epidermal injury severity.

To determine whether this lack of difference reflected true equivalence rather than limited statistical power, a paired equivalence analysis was performed across the 22 mice that received both cryogens. The mean paired difference in hypopigmentation score (R134a − HFO‐1234ze) was 0.037 (SD = 0.169). Using an equivalence margin of ±0.1, the two one‐sided tests procedure demonstrated equivalence between cryogens (TOST p = 0.048), with the 90% confidence interval of the mean difference (−0.025 to 0.099) lying entirely within the predefined equivalence bounds. Taken together, the regression and equivalence analyses support the conclusion that R134a and HFO‐1234ze provide equivalent epidermal protection within biologically relevant limits, with injury outcomes driven primarily by fluence and baseline pigmentation rather than cryogen choice.

3.2.4. Measured and Simulated Temperature Induced by Two Cryogens

Following the assessment of laser‐induced injury, we next characterized the intrinsic cooling performance of each cryogen without laser heating. Surface temperature measurements acquired using TFTCs revealed distinct but clinically comparable transient cooling profiles for R134a and HFO‐1234ze (Figure 8). Both sprays produced an immediate temperature drop from a baseline of approximately 22°C, reaching sub‐zero levels within milliseconds of spray onset. At the end of the 50‐ms primary spray, R134a achieved substantially colder temperatures (‐46.6°C ± 0.49°C), approximately 10.4°C below those produced by HFO‐1234ze (‐36.2°C ± 0.54°C). A secondary temperature dip beginning at ~80 ms was observed in both groups and was caused by a brief (~10 ms) cryogen spurt delivered at the time a laser pulse would normally occur; this auxiliary spurt is routinely used to suppress potential sparks from laser‐induced singeing of hair. Overall, these measurements confirm that, independent of laser irradiation, R134a provides more aggressive surface cooling than HFO‐1234ze, offering a mechanistic context for interpreting differences in subsequent laser‐tissue injury responses.

Figure 8.

Surface temperature dynamics during cryogen spray cooling measured with thin‐film thermocouples (TFTC). Mean temperature traces (±1 SD) for R134a and HFO‐1234ze during a 50‐ms primary cryogen spray, followed by a brief secondary spray (~10 ms) beginning at ~80 ms used to extinguish potential hair‐induced sparks. Temperature curves represent measurements obtained without laser irradiation to isolate the intrinsic cooling performance of each cryogen.

Despite these differences in peak surface cooling, both signals demonstrated similar recovery trajectories, with temperatures rising gradually toward baseline over the subsequent 150 ms. By 200 ms, the temperature difference between the two cryogens narrowed markedly (R134a −24.9°C ± 2.49°C; HFO‐1234ze −19.6°C ± 0.63°C), yielding a reduced inter‐cryogen difference of only −5.2°C. The combined TFTC profiles thus show that while R134a generates deeper transient supercooling at the surface, the overall temporal cooling dynamics remain qualitatively similar, and both sprays provide robust and sustained epidermal temperature reduction during the clinically relevant time period surrounding laser exposure.

Numerical modeling of transient heat conduction and tissue freezing in the superficial skin layers were used to examine how the observed differences in surface cooling between the two cryogens translate into subsurface temperature distributions. At 80 ms, the time when a subsequent laser pulse would be delivered, R134a spray produced consistently lower temperatures than HFO‐1234ze throughout the 0–100 µm depth range (Figure 9a). Neither cryogen cools the tissue below the freezing threshold. The largest temperature differences occurred at the surface, where R134a‐cooled skin was approximately 3.7°C colder than skin cooled with HFO‐1234ze. This difference decreased to ~2.7°C at 50 µm depth, corresponding to the upper viable epidermis, and by 100 µm the difference was only ~2°C between the two conditions (Figure 9b). Thus, although R134a delivered stronger surface cooling, the thermal advantage diminishes with depth due to the dominant role of tissue thermal diffusion and the short duration of cryogen exposure.

Figure 9.

(a) Simulated temperature profiles within the superficial 0–100 µm of skin at 80 ms after the onset of a 50 ms cryogen spray, comparing R134a and HFO‐1234ze. (b) Differential temperature profile (ΔT = T_R134a − T_HFO‐1234ze) at the same time point. Although R134a attains lower surface temperatures, the temperature difference remains small across the viable epidermis, indicating that both cryogens may provide comparable thermal protection during laser irradiation.

4. Discussion

This study directly compared two commercially viable cryogens, R134a and HFO‐1234ze, in their ability to protect epidermis from laser‐induced injury across a physiologically relevant range of skin pigmentation. Although R134a produced colder surface temperatures and greater transient cooling intensity than HFO‐1234ze, these differences did not translate into measurable differences in hypopigmentation outcomes under controlled laser exposure. The results therefore indicate that, within the treatment parameter space, HFO‐1234ze provides epidermal protection functionally equivalent to R134a despite its less aggressive surface cooling profile.

Several factors likely contribute to the observed equivalence in epidermal injury outcomes. Numerical modeling demonstrated that the difference in cooling power between cryogens decays rapidly with depth. Even though simulated skin surface temperatures differed by ~3.7°C, temperature differences in the upper viable epidermis were ~2.7°C and diminished to only ~2°C by 80–100 µm. Because epidermal injury thresholds depend primarily on temperatures reached within these deeper layers rather than at the immediate surface, attenuated thermal contrast at depth reduces the functional impact of stronger surface cooling. It is also possible that the incremental advantage of R134a is too small to be detected by the hypopigmentation scoring method or that biological variability in melanocyte response masks subtle thermal differences. Together, these factors help explain why cryogen choice did not significantly influence epidermal injury outcomes, despite being thermophysically distinct at the surface.

In evaluating alternatives to R134a, HFO‐1234ze is among the most promising candidates due to its dramatically reduced GWP (< 1) and favorable safety profile [28, 29, 30], including low flammability under clinical spray conditions. From a practical standpoint, adoption of HFO‐1234ze may involve considerations related to cost, supply chain availability, and system‐level compatibility. Since HFO‐1234ze operates at pressures comparable to R134a under clinically relevant conditions, existing cryogen delivery architectures are expected to be broadly compatible, although validation of long‐term material compatibility (e.g., seals and valves) would be required at the manufacturing level. Detailed cost analyses and engineering qualification are beyond the scope of the present study. A related hydrofluoroolefin, HFO‐1234yf, has very similar thermophysical properties as compared to R134a, though its higher flammability index may limit clinical adaptability without engineering modifications. Other non‐halocarbon cryogens such as liquid CO₂ have been proposed, but their markedly different spray characteristics and more complex delivery systems present significant practical challenges. HFO‐1234ze therefore occupies a practical balance between environmental sustainability, safety, and predictable cooling performance. If further cooling enhancement is required, auxiliary approaches such as cold air jets have been shown to improve spray cooling performance of low‐GWP refrigerants [31].

Environmental considerations also extend to atmospheric degradation products of HFO‐1234ze and their indirect climate impacts. Laboratory and atmospheric modeling studies indicate that trace amounts of HFC‐23 can form as a minor byproduct during the oxidation of HFO‐1234ze [32, 33, 34]. Although HFC‐23 has an extremely high 100‐year global warming potential (GWP ~ 14,700), its formation yield is very small, resulting in an estimated indirect GWP contribution of approximately 3.3 for HFO‐1234ze [34]. This indirect contribution remains negligible compared with the direct GWP of R134a, whose continued clinical use is increasingly difficult to justify from a regulatory and environmental standpoint. The present work supports the transition toward HFO‐based cryogen spray cooling by demonstrating comparable epidermal protection while offering a dramatically reduced overall climate footprint.

We acknowledge structural differences between murine and human skin that are relevant to interpretation of cryogen spray cooling performance. The murine epidermis is ~25 μm thick [16], which is thinner than the human epidermis and places melanin closer to the skin surface than in human skin. As a result, this model represents a more sensitive test of cryogen‐dependent epidermal protection, since surface cooling differences attenuate rapidly with depth. The absence of cryogen‐dependent differences in injury under these conditions therefore supports, rather than limits, the relevance of these findings for protecting human epidermal melanin. Although murine skin is thinner than human skin overall, the dorsal dermal thickness of ~500 μm [35] is sufficient for the skin to be treated as a semi‐infinite medium during transient CSC, because the thermal penetration depth at 80 ms after spray initiation is ~185 μm. Consequently, the temperature distributions relevant to epidermal protection are not expected to differ substantially between murine and human skin.

Differences between the mouse model and human clinical response must also be acknowledged. In humans, crust formation has historically been used as an indicator of excessive epidermal heating [36], whereas in this mouse model hypopigmentation served as the primary injury endpoint. Human epidermis is thicker, contains a more complex rete ridge architecture, and has a different melanin distribution, all of which influence thermal injury thresholds. Moreover, topical forskolin application was discontinued after laser irradiation, a condition expected to return melanocytic cAMP signaling toward baseline and thereby limit new eumelanin synthesis during the post‐treatment period when damaged melanosomes would otherwise be replaced. This experimental condition differs fundamentally from human skin, in which melanocytes remain metabolically active after laser exposure and can progressively regenerate melanin. In the mouse model used here, the absence of ongoing melanin synthesis likely limits the replacement of damaged pigment, thereby lowering the apparent fluence threshold for visible hypopigmentation. Consistent with this interpretation, we observed that laser‐induced hypopigmented sites in mice readily repigmented following re‐initiation of topical forskolin (data not shown), a response that contrasts sharply with laser‐induced hypopigmentation in human skin, which often persists for weeks to months [37, 38]. This distinction highlights an important limitation of the model for studying long‐term pigmentary outcomes, while reinforcing its utility for controlled, mechanistic comparisons of acute epidermal injury and cryogen‐dependent protection.

Despite these limitations, the strong inter‐rater reliability of the hypopigmentation scoring system, combined with consistent findings across multiple statistical analyses, strengthens confidence in the study's conclusions. The similarity in injury outcomes between R134a and HFO‐1234ze cannot be attributed to possible bias in fluence selection, as analyses confirmed no cryogen‐dependent differences in applied fluence. Moreover, a paired equivalence analysis demonstrated that hypopigmentation outcomes for R134a and HFO‐1234ze were statistically equivalent within a predefined, biologically negligible margin, indicating that the observed similarity reflects true equivalence rather than limited statistical power. Together, these results confirm that the two cryogens provide equivalent epidermal protection under matched laser conditions and indicate that HFO‐1234ze is a viable, environmentally responsible replacement for R134a that preserves the protective function essential to safe laser dermatologic procedures.

Conflicts of Interest

The authors declare no conflicts of interest.