The wavelength-dependent penetration depth of light into human skin is an important mark for photobiologists helping to estimate ultraviolet (UV)-induced processes in the different skin layers. For example, knowing how much radiation of what wavelength is available at a certain depth in living human skin is one prerequisite for a valid risk assessment for skin cancer based on in vitro and animal models.
Thus far, photobiologists had to rely mainly on in vitro and ex vivo data on optical properties and penetration depths because of the lack of an appropriate in vivo technique.1, 2, 3, 4 Especially in recent years, available in vivo data on optical properties and relevant chromophores have been advanced by reflectance spectroscopy, fluorescence, and optical coherence tomography for the visible and infrared wavelength range.5, 6, 7, 8 However, measurements in the UV range suffer from the strong increase of absorption toward shorter wavelengths, making purely optical techniques less and less favorable. This challenge could be met by the development of UV optoacoustics, a hybrid technique allowing noninvasive, depth-resolved determination of optical properties of human skin even in the UV.9, 10, 11 Thus far, optoacoustics has been successfully applied as a tomographic technique tracing especially chromophores, such as melanin, oxy- and deoxyhemoglobin, for image contrast.12, 13, 14
The mechanism behind optoacoustics is the absorption of radiation energy in matter, its transfer to heat and pressure, and the release of a resulting transient stress wave—thermooptical excitation of ultrasound. If the incident light pulses are sufficiently short—typically in the low, one-digit nanosecond range—the profile of the stress transient reproduces the distribution of heat sources in the sample and, consequently, the light distribution in the sample can be deduced allowing one to calculate the depth-dependent optical properties of the sample. Depending on the cumulative dose of UV exposure and the UV sensitivity of the subject, this technique may be completely noninvasive or induce a (mild) erythema at worst.
Here, we present the first in vivo data on the wavelength-dependent penetration depths of UV radiation at different skin sites which were derived from absorption coefficients of human skin measured by optoacoustics.
Materials and Methods
The optical properties of human skin in vivo were determined on the volar and dorsal aspect of the forearm as well as on the thenar in a subject study ( , 14 women, 6 men) using UV optoacoustics. The study design was approved by the local ethics committees, and all volunteers gave informed consent.
Subjects with different phototypes (I: , II: , III: , IV: ) were tested for their individual minimal erythema dose (MED) at the volar side of the forearm. The subject’s individual MED at the volar side of the forearm was determined by chromameter measurements (Minolta CR200, Minolta, Osaka, Japan) according to COLIPA recommendations, defining the erythemal threshold as a change in of units after irradiation with a solar simulator (M.U.T. GmbH, Hamburg, Germany). For subjects with Fitzpatrick phototype characteristics in-between two types, the phototype classification was decided with the help of their MED at the volar side of the forearm, resulting in a corresponding MED range for each photoype: phototype I , phototype II ca. , phototype III , and phototype IV .
Optoacoustic measurements were carried out in reflection mode. Laser radiation is delivered via a UV-enhanced optical fiber that illuminates a ca. large area of the skin through UV-transparent ultrasound gel, which also ensures acoustic coupling between the sample and a poly(vinylidene fluoride) ultrasound detector above the skin. More details of the optoacoustic sensor have been described elsewhere.15, 16
A pulsed UV laser system ( , Ekspla, Lithuania) provided wavelength tunable radiation in the range from with pulse duration of and pulse energies in the range of . The energy of each pulse was recorded by an energy monitor at the UV fiber. For one optoacoustic wavelength scan, the subject’s skin was exposed to an erythemally weighted UV dose on the order of .
Within this wavelength range, optoacoustic measurements were carried out in steps at the volar and dorsal aspect of the forearm as well as on the thenar. Three spectra were recorded for averaging within a circular area of on the relevant skin areas of each subject. On the volar aspect of the forearm, these measurements were carried out on both arms and again after the first measurements on the right arm (no indication of significant differences between these three measurement triples). If necessary, the skin area was shaved at the dorsal aspect of the forearm to minimize the influence of hair on the optoacoustic measurements. In any case, measurement spots were aimed to be in the hair-free space between follicles.
For signal processing and analysis, we used a self-developed simulation software based on the equations for thermooptical excitation of sound in a dissipative medium with inhomogenous optical properties. The program takes into account the actual geometry of our experimental setup and was calibrated using melanin dyed tissue phantoms,16 thus also incorporating sound attenuation, influences of the electronic setup, etc. For analysis, the complete optoacoustic signal is simulated. Even though the optical properties of the skin are expected to change considerably with depth—especially from stratum corneum to viable epidermis—we could not resolve a depth dependency in our measured data. Student’s t-test was used for statistical analysis of significant differences between the UV penetration depths at different skin sites.
The absorption coefficients of human skin in vivo were measured at the volar and dorsal aspect of the forearm as well as at the thenar (three measurements at each site per subject) in a study on 20 subjects of different UV sensitivity (phototype I: , II: , III: , IV: ). Penetration depth was calculated on the basis of these absorption coefficients.
The skin on the inner and outer sides of the arm is very much alike; it may be a little thicker and more strongly pigmented on the outer side of the arm. Figure 1 shows the average wavelength-dependent penetration depths at the three sites together with their 90% confidence intervals. As expected, the penetration depths at the volar and dorsal forearm show fairly similar characteristics. Ultraviolet radiation can penetrate deeper into the skin at the volar side throughout the spectrum ( for 335–305 and , for 329–311 and ; one-tailed paired t-test). This is in accordance with the consideration that skin on the dorsal forearm of the same subject should display the same pigmentation as on the volar side, plus possibly some amount of extra facultative pigmentation because the outer side of the forearm is more strongly exposed to the sun than the inner side. Equal or stronger pigmentation should result in equal or shallower penetration depths comparing these two sites for the same subject. Minimal penetration depths of ca. are found at for both sites. Toward longer wavelengths, the penetration depths are slowly increasing in both spectra (up to , see Fig. 1 for details). At the dorsal side, maximal penetration depths are found at followed by a weak decrease toward . At the volar side, the penetration depths do not show a maximum but reach a plateau in the UVA (315–499 nm).
The spectrum of penetration depths at the thenar differs significantly from those at the forearm (volar/thenar: for 341–332 and , one-tailed paired t-test; dorsal/thenar: for 341–329 and , for 341–332 and , one-tailed paired t-test): whereas in the UVB (290–315 nm) the penetration depth only slowly rises from well above , it increases steeply in the UVA so that radiation even penetrates deeper at the thenar than at the forearm sites at wavelengths longer than ca. .
As already indicated when comparing volar and dorsal forearm spectra, a one-tailed paired t-test was used because we expect a change of optical properties from skin site to skin site caused by UV adaptation and/or a thicker horny layer to go in the same direction for each subject. Intraindividual comparison of the site spectra supports this assumption. However, even if this is not assumed and a two-sided test is used, the results substantially stay the same: for the difference between volar and dorsal forearm spectra, is then slightly larger than 0.05 for 335 and and drops from highly significant to significant for ; no changes in the level of significancy occur for the t-test of volar/thenar, and for dorsal/thenar, only turns from highly significant to significant at .
Figure 2 takes a closer look at the phototype dependence of the penetration depths at the three sites. This time the average penetration depth was calculated separately for the different phototypes. Of course, the 90% confidence intervals are fairly large here because of the small number of subjects: for phototype I, for phototype II, for phototype III, and for phototype IV. Still, the breakdown of the data into the four phototype groups allows an interesting perspective.
At the volar side of the forearm [Fig. 2a], penetration depths of phototypes I–III are very close in the UVB. Penetration depth is minimal at and rises constantly up to ca. at . It is maximal in the range from and decreases again toward longer wavelengths. In the UVA, the close agreement between the three phototypes is lost to some degree and the curves are less smooth. Penetration depths are now higher (approximately ) in phototypes I compared to phototypes II and III (well above to ca. ). Phototype IV behaves a little differently from the other three types. Penetration depth stays at a low level of from up to . Then it rises to another plateau at from onward. This implicates higher penetration depths in phototypes I and IV in the UVA range above ca. than in phototypes II or III.
At the dorsal side of the forearm [Fig. 2b], phototypes I and II as well as III and IV seem to be grouped throughout the UVB and up to . Penetration depth is again rising from shorter to longer wavelengths, and it is higher in the light skin types than in the darker ones. Phototypes behave differently toward longer wavelengths. In type IV, the penetration depth is constantly rising from at up to above at . Types III and II reach a kind of plateau in the UVA. However, this level is reached earlier in type II (at ca. ) than in type III (at ca. ) and it is as high as in skin type II and in type III. In phototype I, penetration depths reach maximal levels of in the range of and then decrease toward longer wavelengths down to .
The phototype-dependent penetration depths calculated for the thenar [Fig. 2c] show an interesting trend: the penetration depths in the horny layer are considerably higher in phototype I than in the other skin types throughout the measured spectral range. The confidence intervals of phototype I data are relatively large due to the small number of subjects in this group; thus, this difference is not significant. If, however, the variance in a larger group of phototype I subjects turned out to be equally small, as that of the group containing phototypes II–IV, the difference between the thenar absorption spectra would become significant for wavelengths other than 335–323 and .
Discussion and Conclusions
Our new data show the natural variance of in vivo properties between individuals and between different skin sites and provide enhanced wavelength resolution. Thus, these first in vivo results add important new information to the ex vivo results1, 2, 3, 4 and answer the call for an in vivo validation of earlier ex vivo data. Anderson and Parrish4 estimate the penetration depth of UV radiation in fair Caucasian skin to be at , at , and at . According to Jacques,2 penetration depths for a moderately pigmented adult with a 10% volume fraction of melanosomes can be expected to rise from ca. at at . Our results are in the same order of magnitude and also agree with respect to the wavelength dependency of the penetration depths but provide a more differentiated view.
The comparison of penetration depths at different skin sites hint at keratin as an important natural UVB filter apart from the skin pigment melanin. At the forearm, the keratin-containing stratum corneum at the skin surface is only ca. thick, whereas it is several thick at the thenar. Keratin is a potent short-range UVB absorber and a strong scatterer throughout the spectrum, whereas melanin is a broadband UV absorber.17, 18 Melanin also acts as a scatterer, but its scattering coefficient only contributes a few percent of the total attenuation coefficient in the UV.19 Thus, the thenar absorption spectrum is dominated by the characteristics of keratin, whose absorption coefficient is rapidly decreasing toward longer wavelengths. This is not compensated for by melanin absorption at the palm. In contrast to this, the much higher melanin concentration at the forearm limits UV penetration throughout the spectrum.
The subdivision of the skin site spectra according to the subjects’ phototypes shows interesting trends, even though the relatively small number of subjects per phototype impedes significant results. Apart from different values for the penetration depth at the same wavelength, the spectral course may be different for different phototypes as well. At the volar side of the arm, for example, the penetration depth reaches its maximum at in phototype II, whereas maximal penetration depths for phototype IV are found at and above. Comparing volar forearm spectra and the horny layer spectra from the thenar suggests that the comparably small penetrations depths and the shift of maximal penetration depth toward longer UV wavelengths found at the volar forearm of phototype IV subjects might be due to a stronger influence of the horny layer on the optical properties, possibly because of a thicker horny layer. A thicker horny layer could also provide an explanation for phototype IV showing a trend to higher penetration depths at the volar forearm than phototype II and III for wavelengths longer than .
Melanin is not distributed equally in the epidermis, but for white subjects, 56% is contained in the lower (basal) layer, 30% in the middle (spinous) layer, and only 14% in the upper (granulosum) layer.20 We were not able to differentiate optically different layers of the epidermis by our optoacoustic measurements. Thus, if the upper layer of phototype IV skin is thickened, the measured optoacoustic data are more strongly influenced by the properties of the rather weakly pigmented upper skin layers, showing in comparably low measured absorption coefficients in the UVA and consequently higher penetration depths. However, this would have to be seen in the perspective of a thicker horny layer and thus a thicker epidermis (i.e., the viable epidermis may still be well shielded as would be expected for phototype IV). The decrease of penetration depth for wavelengths above ca. , which is seen at the volar side for at least phototypes I–III and at the dorsal side for phototype I, could possibly be due to a maximum in the absorption of constitutive melanin.17 The spectral characteristics of this pigment could play a role at sites with low UV exposure or for subjects with a very weak ability to tan. If the trend of horny layer spectra splitting up in phototype I and phototypes II–IV can be confirmed in measurements on more subjects, then it will give way to some interesting speculations. The measurements at the thenar are representative for pure horny layer data. The differences between phototypes I and II–IV indicate that the horny layer of type I is more transparent for UV radiation than that of the other three. There is no reason why this should be a characteristic unique to very thick horny layers only. Consequently, a horny layer of the same thickness would provide much less UV shielding in type I than in the other three. Taking into account that the significance of the stratum corneum in photoprotection is higher in fair-skinned individuals than in pigmented individuals,21 this deficiency would represent an important cause for the high UV sensitivity of type I.
This work was partially supported by a grant from the DFG (German Research Foundation).