U TION

T IO N A L CONTRIB
EDUC A

In Vivo and In Vitro SPF Determination –

Two Sides of the Same Coin?

M. Rohr14, E. Klette2, D. Kockott11, K. Jenni10, R. Bimczok1, H. Gers-Barlag2, C. Mundt2,

S. Bielfeldt3, T. Rudolph4, F. Pflücker4, U. Heinrich5, H. Tronnier5, W. Johncock6, S. Peters6,
B. Klebon7, C. Mendrok-Edinger8, Flößer-Müller9, J. Lademann12, B. Herzog13
1Consultant, Seeheim-Jugenheim, Germany, 2Beiersdorf AG, Hamburg, Germany, 3proDERM GmbH, Schenefeld, Germany,
4Merck KGaA, Darmstadt, Germany, 5Institut für Experimentelle Dermatologie, Universität Witten/Herdecke, Germany,
6Symrise AG, Hamburg, Germany, 7L’ORÉAL Deutschland GmbH, Düsseldorf, Germany,
8DSM Nutritional Products, Basel, Switzerland, 9BASF AG, Ludwigshafen, Germany,
10Evonik Industries AG, Goldschmidt, Essen, Germany, 11UV-Technik, Hanau, Germany,
12Charité – Universitätsklinikum Berlin, Klinik für Dermatologie, Berlin, Germany
13BASF Grenzach GmbH, Grenzach-Wyhlen, Germany, 14Institut Dr. Schrader Hautphysiologie, Holzminden, Germany

DGK Task Force – Sun Protection

Keywords: Sun protection factor, in vitro SPF, in vivo SPF, sun protection, sunscreen application,
UV protection

ABSTRACT

For the purpose of evaluating the risk for
consumers who apply less sunscreen
than the standard test application
amount, the DGK Task Force - “Sun Pro-
tection” initiated a multicenter study in
accordance with the in vivo test proto-
col of the International Sun Protection
Factor (SPF) Test Method. Three differ-
ent amounts (0.5, 1.0 and 2.0 mg/cm²)
of three commercial sunscreens were
investigated in three test centers. The
main result obtained was a linear de-
pendence of the SPF on the quantity ap-
plied. The recommendation to use an
application quantity of 2.0 mg/cm² is
clearly driven by the aim to produce re-
liable and reproducible in vivo results.
In the subsequent in vitro multicenter
study, possible methodologies and
strategies for improved accuracy of the
determination of the SPF on sandblast-
ed PMMA plates were compared. Influ-
encing factors such as the amount of
product applied, pretreatment of the
samples, equilibration time and tem-
perature, and the quality of the spec-
trophotometer were taken into account.
An overall decrease in the standard de-
viations within single laboratories could
be achieved for in vitro SPF testing but
no improvement in the span of interlab-
oratory differences was obtained.
The complete findings obtained from
this multicenter study could perhaps
serve as a solid basis for increasing the
reliability of a future in vitro SPF method.

INTRODUCTION

Sunscreens are commonly used to
protect against short-term as well as
long-term skin damage such as pre-
mature skin aging and cutaneous
erythema. The sun protection factor
(SPF) of a sunscreen product is wide-
ly accepted by the general public as
a measure for the protection offered
by a sunscreen preparation against
sunburn.
The in vivo method to measure the
SPF is based on the concept of a min-
imal erythema dose (MED). The first
formalized method for SPF determi-
nation and labeling was issued in the

IFSCC Magazine 2 | 2012 97

USA by the FDA in 1978 [1]. This and However, in a number of subsequent ence of three different amounts (0.5,
all subsequently issued standards re- studies deviations from the Beer-Lam- 1.0 and 2.0 mg/cm²) of three com-
tained the application amount of 2.0 bert Law were reported. When indi- mercial sunscreen products in three
mg/cm². viduals were asked to apply commer- test centers in accordance with the in
However, under authentic conditions cial sunscreens ad libidum, the re- vivo test protocol of the International
sunbathers tend to apply less than sulting SPF was only approximately SPF Test Method (IM 2003) [5, 6].
2.0 mg/cm² of the sunscreen product 50% of the declared SPF [4]. A second multicenter study dealt with
to exposed skin in an uneven layer. It is unanimously accepted that the development of an in vitro SPF
Several studies have shown that con- SPF under real conditions is lower method.
sumers typically apply between 0.5 than the declared SPF, but there is The EU recommendation on sun pro-
and 1.5 mg/cm² [2]. Assuming the ap- some discussion on the resulting pro- tection products published in Sep-
plicability of the Beer-Lambert Law, a tection for consumers who apply less tember 2006 suggests that due to eth-
50% reduction of the thickness of the than 2.0 mg/cm². ical and practical reasons the Interna-
layer would thus reduce the SPF by Therefore, the DGK Task Force - »Sun tional Method 2006 [7] or an other-
the square root, e.g., from SPF 25 to Protection« initiated the first multi- wise appropriate in vitro SPF test
SPF 5 [3]. center study to investigate the influ- should be implemented for the as-
sessment of the SPF.
Table 1 Average SPF data and statistical parameters in three technical centers In the past, several attempts have
been made to develop in vitro meth-
Application 0.5 1.0 2.0 ods for determining protection against
amount mg/cm
UV radiation. However, due to vari-
Product A
ous practical implications like com-
N 33 35 33 plicated physical behavior of scatter-
SPF 6.0 9.7 22.8 ing films, multifaceted composition of
SD (s) 1.6 2.0 4.5 modern sunscreens and difficulty in
t 2.04 2.03 2.04 applying very thin sunscreen films in
CI95 0.6 0.7 1.6 a reproducible manner up to now no
CI95, % 9.3 7.1 7.0 broadly accepted method has been
produced. Recent attempts to use
Product B
standardized substrates such as one-
N 33 36 32 sided roughened quartz or Plexiglas®
SPF 6.9 11.7 24.2 plates [8, 9, 12] have been undertaken.
SD (s) 2.2 3.0 4.8 While working on the multicenter
t 2.04 2.03 2.04 study »Influence of Applied Quantity
CI95 0.8 1.0 1.7 of Sunscreen Products on the Sun Pro-
CI95, % 11.2 8.6 7.1 tection Factor (SPF)« [10], which is al-
so summarized in this publication, the
Product C DGK Task Force - »Sun Protection«
N 33 32 33 measured in vitro data in eleven dif-
SPF 10.1 18.5 30.7 ferent laboratories in order to com-
SD (s) 3.0 2.9 6.7 pare possible methodologies and
t 2.04 2.04 2.04 strategies for an in vitro approach to
CI95 1.1 1.0 2.4 the SPF. The data obtained from these
CI95, % 10.5 5.6 7.8 tests and several previous tests were
non-homogeneous and have given
P3 standard rise to further questions; nevertheless,
N 99 they point to technical side effects
SPF 15.1 such as the quality of the spectropho-
SD (s) 2.5 tometer used, amount of applied
t 1.99 product, and pre-treatment and equi-
CI95 0.5 libration procedure for the samples as
well as to which application process
CI95, % 3.3
may influence a chosen in vitro
N = overall number of subjects in all test sites; SPF = average of the individual values (SPF k) referring to
N; SD = standard deviation of the respective mean SPF value; t = t value from paired t test; CI 95 =
method. The findings of these ring test
confidence intervals calculated from the equation studies may prove useful for other re-
CI95 = t . s/ N1/2 (1) search groups working on similar top-

98 IFSCC Magazine 2 | 2012

ics to support development of a reli-
able and commonly accepted in vitro
SPF method.

RESULTS AND DISCUSSION

In vivo multicenter study

The average SPF data and the statisti-
cal parameters obtained in the in vivo
multi-center study in the three techni-
cal centers are listed in Table 1. Fig. 1
shows the SPFs of the three products
determined with an application
amount of 2.0 mg/cm² according to
the IM 2003.
In accordance with the aim of the
multicenter study, further investiga-
tions were carried out with samples A,
B and C using reduced product appli- Fig. 1 SPFs of products A, B and C determined in three technical centers (TC1, TC2 and
cation amounts of 1.0 and 0.5 TC3) in an application amount of 2.0 mg/cm² (experimental errors indicated as confidence
mg/cm². intervals, labeled SPF values drawn as horizontal, dashed lines; labeled SPF = 20 for prod-
Fig. 2-4 clearly show the expected de- ucts A and B, and 25 for product C).
crease in SPF when using smaller
amounts of sunscreen products. At
first glance the mathematical correla-
tion of SPF and applied amount ap-
pears to be linear for all samples. In
order to obtain a closer examination,
linear regression lines were calculat-
ed; a correlation coefficient of R >
0.98 was calculated for all types of
products (product A: R = 0.981, prod-
uct B: R = 0.997 and product C: R =
0.995). With only one exception
(product A), the mathematical model
of a linear correlation was within the
confidence intervals of the individual
in vivo SPF values measured.
The impact of the application amount
on the 95%-CI of the in vivo SPF is
shown in Fig. 5. There is a clear ten-
dency of an increased confidence in- Fig. 2 SPF values of product A with varying application amounts measured in three tech-
terval with smaller amounts of ap- nical centers (TC1, TC2 and TC3) (statistical experimental errors indicated as confidence
plied sunscreen. Taking into account intervals; labeled SPF values drawn as horizontal, dashed lines)
that the tested products represent a
wide range of UV filters described as ducing reliable and reproducible re- center in vitro study to investigate the
‘easy’ to apply, the correlation sults. influence of different amounts ap-
demonstrated in Fig. 5 is a clear and In this study, a correlation (close to plied on PMMA plates on the result-
powerful argument for the for use of linearity) of SPF and the amount ap- ing in vitro SPF.
2.0 mg/cm² as a standard amount for plied was obtained. This linear de-
the applied product. Therefore, justi- pendence found in the in vivo inves- CONCLUSION (in vivo)
fication for the relatively high amount tigations demonstrates that the Beer-
required by each of the generally ac- Lambert Law is obviously not appli- A reasonable explanation for the lin-
cepted SPF methods, cannot come cable. ear relationship obtained for the SPF
from daily usage by the consumer but The same three sunscreen products A, and product amount could be an ef-
is clearly driven by the aim of pro- B, and C were used later in a multi- fective absorbing product thickness

IFSCC Magazine 2 | 2012 99

 to the consumer as well as to the man-
ufacturers of sunscreens. Bearing this
in mind, the described multicenter
study contributes to clarification of
the different views concerning the risk
to consumers who under real condi-
tions apply less than the 2.0 mg/cm²
used under laboratory conditions.

RESULTS AND DISCUSSION

In vitro multicenter study

For the first in vitro study, dealing with
different amounts of application, the
same products A (SPF 20), B (SPF 20),
and C (SPF 25) as in the in vivo study
were chosen. Three further products
(D, F and E) with increasing UV pro-
Fig. 3 SPF values of product B with varying application amounts measured in three tech- tection (SPF 10, 26 and 29, respec-
nical centers (TC1, TC2 and TC3) (statistical experimental errors indicated as confidence tively) were chosen for all following
intervals; labeled SPF values are drawn as horizontal, dashed lines) studies (optimization of the method).

Basic Application Procedures

Fig. 4 SPF values of product C with varying application amounts measured in three TCs
technical centers (TC1, TC2 and TC3) (statistical experimental errors indicated as confi-
dence intervals; labeled SPF values drawn as horizontal, dashed lines))
that may not be linearly linked to the
amount of product applied by weight.
Erythemal protection can be described
by the convolution of the absorption
spectrum of an applied sunscreen with
the erythemal effectiveness curve
(product of action spectrum and solar
radiation), which is influenced by ho-
mogeneity and texture of the protect-
ing film covering the skin.
Due to the changing amounts of ap-
plied product the ratio of the applied
amount from height to valley may
possibly also change. Therefore, by
varying the film texture the expected
exponential link between the amount
of product and SPF could be com-
pensated and an experimental linear
correlation obtained.
Thus, the linear manner of consumer
thinking can also be applied to the
SPF, which is an important result of
this study and should be made known
The first procedure for product appli-
cation can be described as follows:
1. The PMMA plate was weighed.
2. The product was applied using a
pipette to homogeneously apply
droplets. The applied product (wet)
was measured using a balance.
3. The product was distributed using
a fingertip, which had been satu-
rated with the product prior to dis-
tribution. During the first phase of
product distribution, typically 20
to 60 s, the product was distributed
quickly until breaking of the emul-
sion can be observed. In a second
phase lasting about 20 to 30 s, the
product was rubbed into the rough-
ened surface of the PMMA plate
with increased pressure (> 200 g).
4. A PMMA plate with a few micro-
liters of glycerin was used as a ref-
erence sample.
5. The equilibration time and temper-
ature were varied as specified in
Table 2.
6. After equilibration for further stan-
dardization, the remaining amount
of dry product was measured.
7. The number of plates measured
and minimum size of the measured
area are given in Table 2.
8. The acceptance limit for the opti-
mized method was linked to a 15%

100 IFSCC Magazine 2 | 2012


Fig. 5 Confidence interval as a function of the applied amount (fit calculated on the ba-
sis of a monoexponential decay function)
confidence limit, which had to be
reached before the plate measure-
ment results were accepted. If the
95% confidence interval was high-
er than 15%, the number of meas-
ured plates was increased until the
CI fell below this limit.
To investigate the influence of the
product application procedure, a sec-
ond (optimized) design was intro-
ations in the calculated results.
Unfortunately, when the linear slope
of this relation was considered for all
laboratories, no uniform correlation
could be found between the in vitro
SPF and applied amount of product
unlike that previously shown in the in
vivo investigations. Linear regression
analysis of any of the three products
of the first experimental set produced
R² values below 0.4, indicating that
no correlation exist when the data of
all laboratories is taken into account.
The slopes of the curves for the ap-
plied amount versus the calculated in
vitro SPF strongly vary among the dif-
ferent test laboratories. This is shown,
for example, for Product A in Fig. 6.
Attempts to harmonize
the application procedure
amount of product applied. An in- Due to this variation among laborato-
creasing amount of product resulted ries a new set of experiments was per-
in an increase in interlaboratory vari- formed as described by the optimized
Table 2 Application Parameters for the DIN and Optimized Methods
Item DIN Optimized
application [13] method
Am ount applied 0.5, 0.75, 1.0 0.75 mg/cm
mg/cm

duced. As described in Table 2, the PMMA-plates (Schönberg, Germany, WW2) Ra=2 m, Ra=2 m,
sandblasted sandblasted
size of the measured surface was in-
creased as well as the number of Accepted tolerance of applied am ount - 3%
plates. In addition, the equilibration E q u i l i b r a t i o n t e m p e ra t u r e R o o m t e m p e ra t u re 40 °C
temperature and time were increased E quilibration tim e 1 5 m in 3 0 m in
in order to quantify any possible in- Num ber of plates 4 8
fluence. Minimum measured surface per plate - 7 cm
2

M in i m u m m e a s u r i n g p o i n t s p e r p l a t e 3 -
Correlation of applied amount
of sunscreen and SPF in vitro Accepted 95% CI calculated over number of plates - < 15%

For the several ring tests, the 11 par-

ticipating laboratories had to prove Table 3 Summary of Results for Different Amounts of Product with DIN Application
their instrumentation fulfilled the Amount 0.50 0.75 1.00 0.50 1.00 2.00
technical requirements by means of in mg/cm
the double-plate test and the holmi- Sample in vitro in vitro in vitro in-vivo in-vivo in-vivo
SPF SPF SPF SPF SPF SPF
um wavelength test as described in Product Mean 6.8 13.7 20.4 6.0 9.7 22.8
[11]. A
In a first set of experiments (DIN ap- SD 1.2 4.0 8.2
SD [%] 17.9 28.9 40.5
plication) the link between applied Product Mean 12.0 27.0 36.5 6.9 11.7 24.2
amount of product and in vitro SPFs B
was analyzed. A summary of the re- SD 2.6 7.8 14.5
SD [%] 21.3 28.7 39.7
sults calculated for all three investi- Product Mean 16.3 26.9 44.3 10.1 18.5 30.7
gated products as the mean of all lab- C
oratories is given in Table 3. SD 3.7 6.3 14.8
SD [%] 22.7 23.5 33.5
The in vitro SPF depended on the

IFSCC Magazine 2 | 2012 101

Fig. 6 In vitro SPF as a function of an applied product
method in Table 2. In this experimen- veloped and tested by simultaneously
tal set, the »human« part of applying measuring three products using the
sunscreens was investigated. Before same spectrophotometer. The variation
launching a new ring test, 17 experi- in the dry application amount (amount
enced technicians met in a workshop after equilibration, the amount »seen«
to optimize the different steps of the in by the instrument) could be decreased
vitro procedure. They demonstrated to an acceptable range between 3.1
their individual methods of product ap- and 6.8% (Ci%(0.95).
plication. During this workshop an op- Although the dry amount was in good
timized application procedure was de- agreement, the 17 technicians
demonstrated individual in vitro SPF
Table 4 Relative Confidence Intervals patterns: Certain individuals generat-
(%Ci(0.95)) for the DIN Application Versus ed higher in vitro SPFs than others for
the Optimized Method (mean of all labs) all three samples despite a similar ap-
plication amount. The individual tex-
DIN Optimized ture of the applied film seems to play
a major role concerning the measured
%Ci(0.95) %Ci(0.95) UV-protection of a sunscreen. There-
Product fore, this attempt to harmonize the ap-
19.8 10.5
A plication procedure did not result in
Product comparable in vitro SPF values. For
15.9 9.6
B example, the values for a product with
Product an in vivo SPF of 29 varied from less
23.7 10.6
C than 20 to more than 50.
102
Even introduction of the »optimized
application« for a repeated test of
three products in the “home” labora-
tories did not result in further inter-
laboratory harmonization of the in vit-
ro measurements, but an overall de-
crease in the 95% CI could be ob-
tained (Table 4).
Although the confidence intervals for
a certain laboratory and a certain
product could be improved with the
new application procedure, differ-
ences between laboratories could not
be alleviated.
CONCLUSION (in vitro)
One aim of the in vitro multicenter
study was to improve the reliability
and reproducibility of the commonly
implemented in vitro SPF method ac-
cording to the DIN application pro-
cedure [13]. The operators dealt very
differently with the task of accom-
IFSCC Magazine 2 | 2012
plishing the requirements of the prac-
tical implementation of the method.
Therefore, the technicians were in-
structed in the same laboratory using
the same spectrophotometer with an
optimized application.
These improvements led at the most
to an internal unification of the stan-
dard deviations within laboratories,
but not between the different partici-
pants. At the least, enhancement of
the dynamic range of the instruments
and harmonization of the application
protocol led to intralaboratory confi-
dence intervals between 3% and
15%, which fall into the same com-
parable range as the given 95% CI of
± 17% for the in vivo determination of
the SPF in one testing laboratory ac-
cording to the IM 2003.
All the described drawbacks lead to
the conclusion that the development
of a valid and reliable SPF in vitro test
method still remains a challenge. Am-
bitious further efforts will be neces-
sary to achieve a satisfactory method
in the future in order to replace the
question mark in the title with an ex-
clamation mark.
References
[1] Department of Health, Education and
Welfare, FDA, USA, Sunscreen drug
products for over-the-counter human
drugs; proposed safety, effective and
labeling conditions, Federal Register
43/166, pp. 38206-38269, 1978.
[2] Bech-Thomsen, N., and Wulf, H.C.,
Sunbathers' application of sunscreen
is probably inadequate to obtain the
sun protection factor assigned to the
preparation, Photodermatol. Pho-
toimmuno.l Photomed., 9 (1993)
242-244.
[3] Farr, P.M., and Diffey, B.L., How reli-
able are sunscreen protection fac-
tors?, Br. J. Dermatol., 112 (1985)
113-118.
IFSCC Magazine 2 | 2012
[4] Brown, S., and Diffey, B.L., The effect
of applied thickness on sunscreen
protection: in vivo and in vitro stu-
dies, Photochem. Photobiol., 44
(1986) 509-513.
[5] COLIPA: International Sun Protection
Factor (SPF) Test Method. COLIPA,
2003.
[6] DGK-Task Force X – »Sun Protection«:
Bimczok, R., Gers-Barlag, H., Mundt,
C., Klette, E., Bielfeldt, S., Rudolph, T.,
Pflücker, F., Heinrich, U., Tronnier,
H., Johncock, W., Klebon, B., Wes-
tenfelder, H., Flösser-Müller, H., Jenni,
K., Kockott, D., Lademann, J., and
Herzog M.R.B., Influence of applied
quantity of sunscreen products on the
sun protection factor – A multicenter
study organized by the DGK Task
Force Sun Protection, Skin Pharma-
col. Physiol., 20 (2007) 57-64.
[7] COLlPA, JCIA and CTFA SA: Interna-
tional Sun Protection Factor (SPF) Test
Method, 2006. DGK-Task Force X –
»Sun Protection«: Gers-Barlag, H.,
Wendel, V., Klette, E., Bimczok,R.,
Springob, C., Finke, P., Rudolph, T.,
Gonzenbach, H.U., Westenfelder,
H., Schneider, P., Kockott, D., Hein-
rich, U., Tronnier, H., Johncock, W.,
Langner, R., Driller, H., Pflücker,
F.,and Wünsch T., The Reproducibili-
ty of an in vitro Determination of the
UVA INDEX Describing the Relative
UVA Protection of Sun Care Products,
IFSCC Magazine 5 (2002) .
[8] Pissavini, M. F., Alard,L, Heinrich, V.,
Tronnier, U., Kockott, H., Lutz, D.,
Tournier, D., Zambonin, V., and Me-
loni, M., Determination of the in vitro
SPF, Cosmetics and Toiletries, 118
(2003) 63-72.
[9] DGK-Task Force X – »Sun Protection«:
Gers-Barlag, H., Wendel, V., Klette, E.,
Bimczok,R., Springob, C., Finke, P.,
Rudolph, T., Gonzenbach, H.U.,
Westenfelder, H., Schneider, P., Kock-
ott, D., Heinrich, U., Tronnier, H.,
Johncock, W., Langner, R., Driller, H.,
Pflücker, F.,and Wünsch T., The Re-
producibility of an in vitro Determi-
nation of the UVA INDEX Describing
the Relative UVA Protection of Sun
Care Products, IFSCC Magazine 5
(2002)
[10] DGK-Task Force X – »Sun Protection«:
Rohr, M., Klette, E., Ruppert, S., Bimz-
cok, R., Klebon, B., Heinrich, U.,
Tronnier, H., Johncock, W., Peters, S.,
Pflücker, F., Rudolph, T., Flösser-
Müller, H., Jenni, K., Kockott, D.,
Lademann, J., Herzog, B., Bielfeldt,S.,
Mendork-Endrok, C., Hanay, C., Zas-
trow, L., In vitro sun protection factor:
still a challenge with no final answer,
Skin Pharmacol. Physiol., 23 (2010)
201-212.
[11] Klette, E., Wendel , W., Wittern K.-P.,
and Gers-Barlag H., A quick, practi-
cal test procedure to evaluate the per-
formance of instruments used for in
vitro UV protection measurements,
Int. J. Cosmet. Sci., 24 (2002) 323-329
[12] Heinrich, U., Tronnier, H., Kockott,
D., Kuckuk, R., Heise, H.M., Com-
parison of sun protection factors de-
termined by an in vivo and different
in vitro methodologies: a study with
58 different commercially available
sunscreen products, Int. J. Cosmet.
Sci., 26 (2004) 79-89.
[13] DIN 67502, Characterization of UVA
protection of dermal suncare pro-
ducts by measuring the transmittance
with regard to the sun protection fac-
tor, 2005.
Corresponding author
Mathias Rohr
Institut Dr. Schrader Hautphysiologie
Holzminden, Germany
mathias.rohr@schrader-institute.de

103