[Federal Register Volume 59, Number 65 (Tuesday, April 5, 1994)]
[Unknown Section]
[Page 0]
From the Federal Register Online via the Government Publishing Office [www.gpo.gov]
[FR Doc No: 94-8023]


[[Page Unknown]]

[Federal Register: April 5, 1994]


_______________________________________________________________________

Part III





Department of Health and Human Services





_______________________________________________________________________



Food and Drug Administration



_______________________________________________________________________



21 CFR Parts 352, 700, and 740




Discussion of Ultraviolet A-Protection Claims and Testing Procedures 
for Over-the-Counter Sunscreen Drug Products; Public Meeting and 
Reopening of the Administrative Record; Proposed Rule
=======================================================================
-----------------------------------------------------------------------

DEPARTMENT OF HEALTH AND HUMAN SERVICES

Food and Drug Administration

21 CFR Parts 352, 700, and 740

[Docket No. 78N-0038]
RIN 0905-AA06

 
Discussion of Ultraviolet A-Protection Claims and Testing 
Procedures for Over-the-Counter Sunscreen Drug Products; Public Meeting 
and Reopening of the Administrative Record

AGENCY: Food and Drug Administration, HHS.

ACTION: Notice of public meeting and reopening of the administrative 
record.

-----------------------------------------------------------------------

SUMMARY: The Food and Drug Administration (FDA) is announcing that a 
public meeting will be held to discuss testing procedures to 
demonstrate that an over-the-counter (OTC) sunscreen drug product 
protects users from ultraviolet A (UVA) radiation. FDA is holding this 
meeting after considering public comments regarding UVA claims and 
testing procedures received in response to the agency's notice of 
proposed rulemaking and letters sent by the agency. In addition, FDA is 
reopening the administrative record for the proposed rulemaking for OTC 
sunscreen drug products to allow comment on matters considered at the 
meeting. FDA intends to invite guests and consultants to address 
technical matters related to the questions listed in this document. 
This meeting is part of the ongoing review of OTC drug products 
conducted by FDA.

DATES: The meeting will be held on May 12, 1994, 8:30 a.m. The agency 
anticipates that the meeting will last 1 day. However, if there is 
sufficient interest in participation, the meeting will be extended an 
additional day at the discretion of the chairperson. Submit relevant 
data and notice of participation by April 29, 1994. Submit comments 
regarding matters raised at the meeting by July 31, 1994. The 
administrative record will remain open until July 31, 1994.

ADDRESSES: Submit relevant data, notice of participation, and written 
comments to the Dockets Management Branch (HFA-305), rm. 1-23, 12420 
Parklawn Dr., Rockville, MD 20857. The meeting will be held in 
conference rm. D, Parklawn Bldg., 5600 Fishers Lane, Rockville, MD 
20857.

FOR FURTHER INFORMATION CONTACT: Jeanne Rippere, Center for Drug 
Evaluation and Research (HFD-813), Food and Drug Administration, 5600 
Fishers Lane, Rockville, MD 20857, 301-594-1003.

SUPPLEMENTARY INFORMATION: In the Federal Register of May 12, 1993 (58 
FR 28194), FDA published a notice of proposed rulemaking for OTC 
sunscreen drug products. In the proposed rule, the agency discussed OTC 
sunscreen drug products that claim to provide protection from UVA 
radiation and the public health significance of UVA radiation (58 FR 
28194 at 28232 and 28233). The agency also discussed testing procedures 
for sunscreens that absorb UVA radiation (58 FR 28194 at 28248 to 
28250). The comment period for comments related to UVA ingredients, 
claims, and testing closed on November 8, 1993.
    To ensure that sunscreen drug products having UVA protection claims 
offer significant UVA protection, the agency proposed that an OTC 
sunscreen ingredient must have an absorption spectrum extending to 360 
nanometers (nm) or above in order to display UVA protection claims in 
its labeling (58 FR 28194 at 28233). The product would also have to 
demonstrate meaningful UVA protection by satisfying UVA testing 
procedures that would be included in the monograph. However, these 
procedures have yet to be established. The agency requested specific 
comments on appropriate procedures to be used. Previously, the agency 
had requested specific information on UVA protection factors (Ref. 1).
    In the proposed rule (58 FR 28194 at 28248 to 28250), the agency 
tentatively suggested that a testing method similar to the one 
described by Lowe et al. (Ref. 2) could be used to demonstrate that a 
sunscreen drug product provides protection against UVA radiation. This 
method uses 48- and 72-hour erythema reactions and 12- to 14-day 
melanogenesis in skin sensitized with 8-methoxsalen (8-MOP). However, 
because the agency did not have enough information or data to propose a 
method for determining UVA protection in the proposed rule, the agency 
stated that a method should be developed and validated in the same 
manner as was the sunscreen testing procedure for protection against 
ultraviolet B (UVB) radiation (i.e., sunscreen protection factor (SPF) 
testing). Furthermore, the agency requested comments and data regarding 
an appropriate testing methodology for OTC sunscreen drug products that 
afford UVA protection.
    The agency received a substantial amount of comments, data, and 
information regarding UVA ingredients, claims, and testing procedures. 
After evaluating the submitted material, the agency finds that there 
are two basic interrelated questions regarding testing procedures for 
determining UVA protection that must be addressed before the agency can 
complete its assessment of appropriate UVA ingredients and claims. 
These questions are:
    (1) What action spectrum best describes the biological risk of UVA 
radiation (i.e., which ultraviolet radiation wavelengths are most 
likely to cause biological damage), and
    (2) Which testing procedure best defines the UVA protection 
provided by a sunscreen drug product?

I. Action Spectra for UVA-Related Skin Damage

    Several comments discussed the appropriate action spectrum for 
biological risk associated with UVA radiation. One comment stated that, 
in developing a consensus regarding an acceptable assay for determining 
the UVA protection provided by a sunscreen drug product, the specific 
UVA effect that is to be blocked must be considered. These effects 
include UVA erythema, UVA-induced drug photosensitivity, immediate 
pigment darkening, delayed tanning, or other effects of photodamage.
    One comment stated that several action spectra describing different 
aspects of solar-induced skin damage have been determined in a number 
of different species and cell types. The comment described these 
aspects as photocarcinogenesis, DNA damage, photoaging, mutagenicity, 
and immunosuppression. The comment maintained that each action spectrum 
for UV-induced damage closely tracks the human erythemal action 
spectrum. The comment stated that the best summary of the biological 
risk of UV light has been published by the Commission Internationale de 
l'Eclairage (CIE) (Ref. 3). The CIE Hazard Spectrum embodies the 
comprehensive, yet normal, risks to human skin due to full-spectrum UV 
exposure and reflects the findings of the action spectra biological 
responses to UV described above. The CIE Hazard Spectrum shows that the 
damage risk for UVB at 290 nm is 100 times higher than that at 320 nm. 
The damage risk at 320 nm is 100 times higher than that at 400 nm. The 
damage risk at 320 nm is at least 10 times greater than the damage risk 
at 340 nm. This spectrum shows that the most damage potential is in the 
UVB wavelengths (290 to 320 nm), followed by the shorter UVA 
wavelengths (320 to 340 nm). The comment cited numerous scientific 
articles and submitted action spectra to support its statements (Ref. 
4).
    Another comment stated that although protection against UVA is 
important, it is not as important as protection against UVB. The 
comment argued that, based on the CIE Hazard Spectrum, the UVB 
wavelengths contribute 80 to 85 percent of the damage risk in sunlight, 
while UVA contributes 15 to 20 percent of the damage risk.
    One comment stated that some known effects on humans caused by UVA 
radiation include:
    (1) Photoaging of the skin,
    (2) UVA-induced hypersensitivities,
    (3) Augmentation of skin cancers, and
    (4) Erythema. However, the comment noted that there is no single 
known action spectrum to describe which parts of the UVA spectrum are 
most active in causing these effects. Therefore, the comment maintained 
that it is not appropriate to use the UVA-erythema action spectrum for 
testing purposes. The comment stated that UVA protection should be 
assessed in relation to UVB protection and that the assumption should 
be made that all wavelengths are equally important.
    One comment stated that the ``current test for erythema'' is 
inadequate to test for the full spectrum of UVA radiation because 
erythema is not a valid measurement of UVA exposure. The comment argued 
that using the erythema action spectrum to test for UVA protection will 
give consumers a false sense of the extent of UVA protection afforded 
by a product. The comment added that the immediate pigment darkening 
(IPD) action spectrum is preferable because it is broad enough to take 
into account almost all UVA wavelengths. Another comment stated that 
the IPD action spectrum is indicative of a true broad spectrum UVA 
response.
    One comment noted that the IPD action spectrum was first described 
to extend from 300 to 420 nm, with a broad peak between 340 and 370 nm 
(Ref. 5). The action spectrum was later described to extend from 300 to 
620 nm, with a peak effectiveness between 400 and 500 nm (visible 
light) (Ref. 6). Because of the difference between these two reported 
spectra, the comment reevaluated the action spectrum for IPD between 
310 and 400 nm and reported that the IPD action spectrum extends from 
320 to 400 nm with a low peak around 340 nm (Ref. 7).
    Concurrent with determining which testing procedure is appropriate 
for use in validating the UVA protection provided by a sunscreen drug 
product, the agency must determine what portion of the UVA spectrum 
should be blocked by the product before consumers are effectively 
protected against the hazards of UVA radiation. The agency would like 
to consider the following specific questions at the meeting:
    1. Which action spectra are the most important with respect to skin 
damage caused by UVA radiation?
    2. Is the erythema action spectrum an adequate surrogate for UVA 
biological risk, or is some other action spectrum (such as the IPD 
action spectrum) more appropriate?
    3. Can a sunscreen drug product that protects consumers against the 
shorter UVA wavelengths (320 to 340 nm) but not against longer UVA 
wavelengths (340 to 400 nm) prevent significant UVA damage?
    4. What should consumers expect from a sunscreen drug product that 
is labeled to provide protection against UVA radiation or as a ``broad 
spectrum'' sunscreen?

II. UVA Testing Procedures

    The agency did not propose a method for determining UVA protection 
in the tentative final monograph for OTC sunscreen drug products. The 
agency stated that a method should be developed and validated in the 
same manner as the sunscreen testing procedure for protection against 
UVB radiation (58 FR 28194 at 28250). The agency noted that any such 
method should clearly demonstrate that a particular product provides 
significant protection against UVA radiation. The method should include 
the use of a control sunscreen preparation that absorbs UVA radiation 
and that can be used to assure the reliability of the testing procedure 
and equipment. The method should demonstrate that a sunscreen 
ingredient either does or does not protect against UVA radiation. The 
agency requested comments and data regarding an appropriate testing 
method for OTC sunscreen drug products that protect against UVA 
radiation. In response, the agency received information and data 
pertaining to several UVA testing procedures, including both in vivo 
and in vitro test methods.
    One comment recommended adoption of an in vitro test method that 
does not rely upon either photosensitization or nonsolar light sources 
for determining UVA protection for normal skin (Ref. 8). According to 
the comment, this method involves:
    (1) Determining the UV absorbance spectrum of the sunscreen 
product,
    (2) Calculating a convolution spectrum by multiplying the solar 
spectrum with the current CIE Hazard Spectrum, and
    (3) Incorporating the sunscreen transmission spectrum into the 
convolution spectrum to obtain a UVA effectiveness ratio which is 
conveniently expressed as a UVA protection percentage (APP). The 
comment maintained that, unlike other methods, the APP represents the 
fraction of full spectrum UVA (320 to 400 nm) removed by a product.
    The comment stated that because the original ``full spectrum'' 
method produces an SPF value analogous to the clinically determined SPF 
number, the APP has direct relevance to the SPF determined on human 
subjects and is a subset of the full-spectrum SPF determination. The 
comment added that once an SPF has been determined clinically, it is 
simple to take the full-spectrum absorbance spectrum and calculate the 
APP based on the clinical test results. Therefore, although the 
determination of the APP does involve in vitro measurements, it also 
relies on direct clinical measurements.
    The comment contended that there are a number of advantages to 
using the APP system:
    (1) It is a subset of the existing SPF for sunscreen drug products 
and, therefore, relates to an erythemal endpoint in normal skin;
    (2) It does not unnecessarily duplicate clinical testing;
    (3) It clearly demonstrates whether a sunscreen drug product 
provides meaningful protection against UVA radiation, and it is useful 
in determining comparative UVA protection;
    (4) It avoids the deficiencies of nonsolar light sources, 
photosensitizing chemicals, the failure of dose reciprocity for human 
UVA exposures, and endpoints which do not relate to known UVA damage to 
human skin;
    (5) It is independent of exposure dose or duration;
    (6) It includes all the UVA wavelengths in their direct proportion 
and intensity as found in natural sunlight; and
    (7) It is directly relevant to overall product effectiveness. The 
comment added that in the absence of a light source specific to the UVA 
range, APP determination is the best measurement of a product's UVA 
protection level.
    One comment stated that the in vitro APP test is difficult to 
extend to a human in vivo situation and that the test cannot be used to 
study substantivity or stability. The comment added that because the 
APP test uses the erythema action spectrum and a mathematical 
extraction of the UVA segment of the solar spectrum, it overestimates 
the actual amount of UVA radiation blocked by most products. A reply 
comment argued that the APP technique is derived from well-studied and 
extensively published in vitro SPF methodology (Refs. 9, 10, and 11), 
that it is simple to evaluate water resistance using this model, and 
that data on water resistance have been published. The reply comment 
added that APP values are derived from the same spectral data (320 to 
400 nm) that provide in vitro SPF values. The final SPF value from 
clinical studies is compared to the in vitro SPF and the absorbance 
spectrum can be matched to the exact clinical SPF for UVA calculations. 
The UVA portion of the sunscreen's efficacy can then be calculated from 
the in vitro SPF data. Therefore, the comment argued that the APP has 
direct relevance to the clinical effectiveness of the sunscreen 
product, but does not require the exposure of human subjects to 
unnecessary UV radiation.
    One comment stated that an in vitro method developed by Diffey and 
Robson (Ref. 12) avoids many of the limitations of in vivo methods 
(e.g., lack of reciprocity and light sources that produce 5 to 20 times 
the intensity of the sun) and allows the correct estimation of the 
attenuating power of a sunscreen drug product. The comment described 
this method as recording photocurrent in 5 nm steps from 290 to 400 nm 
and measuring the spectral transmission of UV radiation through a 
sample of TransporeTM tape with and without sunscreen applied. 
TransporeTM tape is UV radiation transparent and has a rough 
surface that distributes sunscreen products in a way similar to the 
uneven surface of the skin. Any radiation source may be used, providing 
there is a continuous power distribution between 290 and 400 nm. This 
method assesses the SPF of a product and the UVA/UVB ratio. The UVA/UVB 
ratio compares the reduction of UV radiation in the UVA region with 
that in the UVB region of the spectrum. According to the comment, this 
ratio can be used as an indicator of the UVA protection properties of a 
sunscreen drug product.
    One comment claimed that the method developed by Diffey and Robson 
(Ref. 12) has many advantages, as well as being simple, inexpensive, 
and well correlated with clinical testing. The comment noted that the 
method does not require a biological endpoint such as erythema, 
tanning, or immediate pigment darkening. The comment stated that the 
method provides a basis for the classification of the UVA protection 
provided by a product and added that a manufacturer planned to utilize 
the Diffey and Robson method to standardize the UVA claims of its 
products. Sunscreen products would be labeled with one to four stars 
depending upon the amount of UVA protection provided by the products as 
determined by the Diffey and Robson method. The comment concluded that 
using the ``star'' rating system for UVA claims and the SPF designation 
for UVB claims provides a simple method for consumers to determine the 
protective nature of a sunscreen product. The comment submitted a 
description of the manufacturer's methodology and ``star'' rating 
system (Ref. 13).
    Another comment submitted data describing the application of this 
ratio method to the determination of the SPF and UVA/UVB ratio of 
titanium dioxide and zinc oxide dispersions (Ref. 14). The comment 
noted that the accuracy of this method is enhanced by good product 
application and that the in vitro results obtained by this method show 
good agreement with in vivo values.
    However, another comment contended that the Diffey and Robson 
method (Ref. 12) has been shown to have poor correlation with clinical 
results (Ref. 15). The comment stated that the Diffey and Robson method 
has been used as the basis for ``quantifying'' UVA protection expressed 
as ``stars'' on the package labeling of some sunscreen products sold in 
Europe (Ref. 16). Without using an action spectrum such as the CIE UV 
Hazard Spectrum or the erythemal efficacy spectrum for weighting, the 
``star'' method considers all UVA wavelengths as having the same 
erythemal effectiveness. The ``star'' value results from an unweighted 
ratio of the UVA absorbance to the UVB absorbance of the product. 
Therefore, the comment maintained that a low SPF product with a flat 
absorbance spectrum could get four ``stars'' (i.e., the highest 
rating), while a higher SPF product would get fewer ``stars'' because 
the higher SPF product would absorb disproportionately higher levels of 
UVB, similar to the action spectrum for erythema. The comment stated 
that the ``star'' concept is in direct contrast to the accepted concept 
of formulating sunscreen drug products to provide the most protection 
in the most damaging portion of the UV spectrum. The comment contended 
that the ``star'' method is misleading to consumers and added that the 
use of the ``star'' method in England has been criticized by 
dermatologists, who have asked that the system be withdrawn.
    One comment recommended that the agency adopt the current Standards 
Association of Australia (SAS) UVA (broad spectrum) test method AS-2406 
as an objective measure of UVA blocking (Ref. 17). This method measures 
the percent transmission of the test sunscreen drug product between 320 
and 360 nm. If an 8-micrometer layer of appropriately dissolved 
sunscreen product does not transmit more than 10 percent of UV 
radiation at any wavelength from 320 to 360 nm inclusive, the product 
may be considered as providing broad spectrum protection. The comment 
contended that this method has a number of advantages. UV protection 
claims are most appropriately substantiated by measuring the blocking 
of UV directly rather than measuring some consequence of UVA exposure. 
Thin film spectrophotometric evaluation of sunscreen drug products has 
reached a level of technical proficiency to permit instrumental 
evaluation of UVA blocking potential. Adopting an already accepted 
standard protocol will enhance the ability of the United States 
sunscreen industry to compete equally in foreign markets. This test 
will substantially reduce testing costs. No human subjects are used. 
The comment added that this method provides a strict criterion that 
serves to identify only the most effective UVA blockers. The comment 
submitted several UVA scans to demonstrate that the SAS method 
differentiates between the ``poorly effective'' oxybenzone-containing 
sunscreens and an assortment of products containing ``excellent'' UVA 
blockers, e.g., titanium dioxide and avobenzone (Parsol 1789) (Ref. 
17).
    Two comments contended that there are several deficiencies in the 
SAS method. The results are not correlated to a clinical SPF test. 
Numerous studies have shown that solution and thin-film spectra are not 
relevant to actual product performance on skin. The performance of the 
sunscreen is evaluated only in the limited range of 320 to 360 nm, 
rather than throughout the entire UVA spectrum (320 to 400 nm).
    Two comments recommended that the agency not adopt in vitro methods 
that rely on measuring the transmission of UVA radiation through either 
epidermis or a UV-transparent skin cast (Refs. 18 and 19). The comments 
contended that these methods are inappropriate because they use 
nonsolar UVA radiation sources and limited range UVA detectors or 
detectors without an appropriately weighted response. The comments 
stated that these limitations would cause the results to be nonrelevant 
to the actual responses of normal skin to full-spectrum natural 
sunlight. The comments mentioned that one method (Ref. 19) contains a 
small but significant contamination by UVB energy below 320 nm that 
would adversely affect the resulting efficacy values and lead to 
erroneous measures of UVA efficacy. The comments stated that the other 
method (Ref. 18) skews results toward the longer UVA wavelengths 
because of the lamp's deficiencies in the shorter, energy-rich UVA. The 
comments added that this skewing causes an overestimation of the 
protection of some products, making those with ingredients that are 
long wavelength absorbers (such as avobenzone) look unrealistically 
effective.
    One comment concluded that a rigorous and foolproof in vitro test 
method has not been established or validated. The comment submitted two 
scientific publications that discuss some of the difficulties 
associated with in vitro sunscreen testing techniques (Refs. 15 and 
20). The comment argued that none of the current in vitro methods 
adequately evaluate the photostability of sunscreens. It further stated 
that a validated in vivo human UVA test method must first be 
established. Then, future in vitro test methods can be tested and 
validated against this standard.
    Several comments urged the agency not to adopt a testing method 
that utilizes photosensitizing chemicals. The comments presented a 
number of arguments against this type of testing. It is considered 
unethical because of the carcinogenic potential of the photosensitizing 
chemical (such as 8-MOP). The action spectrum (i.e., for 8-MOP induced 
erythema) is artificial and inappropriate. The values obtained vary 
with the sensitizing chemical used and the spectrum of the irradiating 
source used. The values obtained have no relevance to real-life 
situations. The testing may result in a persistence of pigmentation or 
blistering reactions. 8-MOP sensitization exaggerates the biological 
response and presents a risk of causing severe ulcerative acute 
reactions. 8-MOP puts subjects at risk for phototoxic reactions for up 
to 6 hours after exposure.
    One comment contended that photosensitizing testing methods have a 
number of benefits, such as short irradiance times, clearly defined 
endpoints, and reproducible results. The comment added that the results 
from these test methods are relevant to patients taking 8-MOP or other 
photosensitizing drugs with similar action spectra. The comment 
concluded that photosensitizing test methods could be useful to 
determine photoprotection factors for claims against phototoxic 
reactions.
    Several comments urged the agency to accept the IPD testing method 
for determining UVA protection. One comment stated that the IPD method 
is generally recognized by a substantial body of scientists as the 
preferred UVA testing method. Several comments provided a number of 
benefits for using the IPD method. They claimed that this test method 
is the one that is most representative of true conditions because it is 
an in vivo determination that accounts for biological reactions that 
can occur on living skin. Unlike the testing procedure using skin 
photosensitized with 8-MOP, the IPD test is indicative of a true broad 
spectrum response of normal healthy skin. Unlike erythema that is a 
response only to the shorter wavelength UVA radiation, IPD is a 
response to broad spectrum UVA radiation. The IPD method requires the 
use of considerably lower doses of radiation energy, thus exposing 
subjects to less risk. The IPD method uniquely compliments the current 
SPF system by accurately reflecting the actual amount of long wave UVA 
radiation attenuated by a sunscreen product. The IPD method is 
reliable, accurate, and reproducible. The IPD test can be performed in 
a standard sunscreen evaluation laboratory with minimal adaptation of 
existing equipment. The comments concluded that until well-established 
action spectra for specific UVA damage are established, IPD is the best 
method currently available because it reflects broad spectrum UVA 
protection equally across the entire UVA spectrum.
    One comment submitted a testing protocol using IPD as the endpoint 
(Ref. 21). The comment stated that the suggested testing procedure 
fulfilled a number of criteria. The resulting protection factor gives 
the consumer additional information about the sunscreen number, 
complementing the SPF value. The response variable has a relatively 
flat action spectrum (i.e., 320 to 400 nm, with a low peak at around 
340 nm) throughout the region of interest. Using this spectrum results 
in UVA protection values that closely reflect the actual amount of 
radiation reduced. The testing response obeys dose reciprocity over the 
anticipated irradiance range. The test is practical with minimal risk 
to subjects. The comment added that an eight-center clinical test has 
validated this method as acceptable for determining UVA protection over 
the entire UVA spectrum, including long wavelength UVA (i.e., 340 to 
400 nm).
    The comment submitted clinical test results from the eight test 
sites (Ref. 21). Each testing facility completed between 10 and 20 
subjects. Subjects with skin types III and IV were used. Four sunscreen 
formulations and a vehicle control were tested. The sunscreen products 
contained:
    (1) 7-percent padimate O,
    (2) 2-percent oxybenzone,
    (3) 5-percent oxybenzone, and
    (4) 4 percent titanium dioxide.
    The protocol used a randomized, complete block design with all 
subjects at each testing center receiving all five test materials. The 
comment stated that the studies were conducted similar to an SPF test, 
but a detailed protocol was not submitted. For example, UVA dosages and 
application density of the test sunscreens were not noted. Six sites 
used a 150-Watt (W) Xenon lamp with 3-millimeter (mm) WG335 and 1-mm 
UG11 filters. One site used a 1,000-W Xenon lamp with 3-mm WG335 and 2-
mm UG11 filters. One site used a Krypton lamp (i.e., Dermlite) of 
unspecified wattage. The comment noted that neither of the last two UVA 
sources are recommended, and the results obtained using these lamps 
were included for information purposes only.
    The IPD threshold dose of UVA radiation was first measured on 
unprotected skin, then on protected skin. The ratio of these two doses 
was then calculated to derive the UVA Protection Factor (UVA-PF). The 
IPD was graded immediately after UV exposure, allowing complete testing 
in a single visit.
    The comment stated that comparison of test products indicated that 
the mean UVA-PF of the vehicle (1.7), the 7-percent padimate O product 
(1.8), and the 2-percent oxybenzone product (1.8) were similar. The 5-
percent oxybenzone product (2.1) and the 4-percent titanium dioxide 
product (3.0) were both significantly greater than the other three 
products. The titanium dioxide product was significantly greater than 
the 5-percent oxybenzone product. Although overall statistical analysis 
detected significant site-by-product interaction, the individual 
results indicate that this was primarily a quantitative interaction 
effect. The comment maintained that the consistency of the results was 
encouraging, considering that this was the first experience in reading 
the IPD response for most of the participating sites. The comment 
stated that these data indicate that the IPD procedure can reliably 
discriminate among products that provide meaningful long wavelength UVA 
protection. The comment proposed using a base size of 20 subjects per 
test and a 4-percent titanium dioxide product as a control to be run 
concurrently with each subject. The comment proposed that the UVA-PF be 
the lower 95 percent confidence interval subtracted from the mean.
    Three comments recommended using an IPD test based upon two recent 
publications (Refs. 22 and 23). One study (Ref. 22) used a Xenon lamp 
equipped with a dichroic mirror filtered with 1-mm WG345, 1-mm WG320, 
and 1-mm UG11 filters. The exposure increments were programmed in 
arithmetic increments of 2 Joules per square centimeter (J/cm\2\). On 
protected skin, the increments were 3 or 4 J/cm\2\. Application density 
of the test sunscreen products was 2 milligrams per cm\2\. Visual 
assessment of pigmentation was done immediately after exposure and was 
performed on the basis of a homogenous pigmentation with well-defined 
borders as endpoints. The UVA-PF is determined as the ratio of minimal 
IPD dose with protection to the minimal IPD dose without protection.
    The other study (Ref. 23) recommends the use of xenon or metal 
halide sources, or a xenon/metal halide combination, with continuous 
spectra restricted to the UVA spectrum (320 to 390 nm) with filters, 
such as 3-mm WG335 and 1-mm UG5. Six UVA doses ranging from 4 to about 
30 J/cm\2\ are applied in 50 percent increments to subjects with skin 
types II, III, and IV. The study states that with potential free 
interpolation, this is actually a 25-percent progression. The doses 
applied on the protected skin are multiplied by the expected UVA-PF of 
the product under test. Observation of the responses are delayed for at 
least 1 hour, and typically 2 hours, after exposure. The study states 
that results are less variable if read at 2 hours. A simultaneous 
determination of the minimal IPD doses on protected and unprotected 
skin is done at the same time in standard room and illumination 
conditions. Other parameters concerning the test area, size of test 
sites, product application density, selection of volunteers, etc. 
follow the same current standards as for the SPF determination.
    Several comments objected to the use of the IPD testing method. One 
comment stated that the action spectrum for the IPD response is flat 
and quite dissimilar from the action spectrum for damage to the skin 
from ultraviolet light for erythema, skin cancer, or photoaging of the 
skin. The comment contended that the IPD response has not been 
demonstrated to be a direct or surrogate endpoint for biological damage 
and, therefore, there is no relationship between a product's ability to 
prevent IPD and to prevent damage to the skin. The comment added that 
IPD has been shown to be unstable, variable, and nonlinear. Another 
comment stated that the IPD reaction shows nonreciprocal behavior, 
i.e., the severity of the reaction depends upon the time taken to 
deliver a certain dose.
    One comment noted that there are two methods of assessing UVA 
protection that are referred to as IPD. The comment stated that these 
two methods differ in the amount of energy needed to produce a response 
and the time after irradiation at which the endpoint is read. In one 
method, the response is read at 45 seconds after exposure. This 
response is transient and has been shown to be highly variable and 
nonreproducible. The response is oxygen-dependent and can only be 
elicited in darker skin types. In the second method, the response is 
read at 2 to 4 hours after exposure and uses a much higher dose of UVA. 
The test causes a persistent pigment response in the skin that may last 
up to several hours. The comment maintained that the action spectrum 
for the persistent pigment endpoint has been neither determined nor 
published.
    The comment argued that the threshold problem of not having a truly 
solar UVA-only light source further complicates the results obtained 
using either IPD method. The comment contended that even filtered Xenon 
lamps contain significant amounts of visible radiation which, while not 
harmful to the skin, may cause the IPD reaction to occur. The comment 
pointed out that a sunscreen's ability to block visible light should 
not be confused or combined with its ability to provide UVA protection. 
In addition, the comment argued that the light sources used for both 
IPD methods lack significant energy in the shorter UVA wavelengths, 
which are present in sunlight and which are responsible for the 
preponderance of UVA damage to the skin.
    Stating that there is great demand among the ``sunbather'' 
population for a ``great looking'' tan and for an indicator to predict 
how good a tan can be obtained with a product, one comment argued that 
tanning tests like the IPD are not appropriate for measurement of the 
damage caused by UVA radiation. The comment contended that measures of 
melanogenesis would be misinterpreted by consumers as indicators of 
efficacy of tanning and that consumers would soon be choosing products 
with the lowest IPD rating to help get the deepest tan.
    One comment recommended that the agency adopt the Protection Factor 
in UVA (PFA) test method (Refs. 24, 25, and 26). This method is similar 
to the SPF testing procedures with a modification to the light source 
to virtually eliminate UVB radiation and thus expose subjects to UVA 
radiation (greater than 99 percent). The PFA test uses subjects with 
skin types I, II, and III. The UVA source is a continuous UVA spectrum 
(preferably xenon arc) filtered with a 3-mm Schott WG335 filter that 
eliminates 99 percent of the UVB radiation, with less than 1,500 W per 
square meter (W/m\2\) irradiance. UVA exposures are delivered at 25-
percent increments to skin above and below the expected UVA protection 
level of the sunscreen product times the minimal response dose. The 
endpoints measured in this testing method are delayed erythema or 
tanning, whichever is present, observed 16 to 24 hours after UV 
exposure. The comment stated that these acute responses have similar 
action spectra to the chronic action spectra for nonmelanoma skin 
cancer (as determined in animals), solar elastosis, and skin wrinkling. 
The comment added that the data indicate equivalent results with either 
response parameter. Minimal response doses are elicited with UVA 
exposure ranging from approximately 80 to 250 J/cm\2\. The PFA is the 
ratio of the minimal response dose on protected skin to the minimal 
response dose on unprotected skin.
    The comment submitted the results of a multicenter evaluation of 
sunscreens using PFA methodology (Ref. 25). Sunscreens containing 2 or 
5 percent oxybenzone, 7 percent padimate O, and a placebo were tested 
in five laboratories using a PFA protocol. All the solar simulators had 
intrinsic UV reflecting/IR absorbing dichroic mirrors and were fitted 
with Schott 3-mm WG335 and 1-mm UG11 filters. The comment stated that 
the PFA test method yielded reproducible results between test centers 
and was capable of distinguishing between the three levels of UVA 
protection provided by the placebo sunscreen and the sunscreens 
containing 2- or 5-percent oxybenzone. The test was incapable of 
distinguishing between the UVA protection provided by the placebo and 
the 7-percent padimate O (a strong UVB absorber with little UVA 
absorbency). The comment stated that these results indicate that the 
PFA test method is not influenced by the presence of a strong UVB 
blocker in the formulation and is specific in identifying UVA 
protection. The comment added that the data show that the level of 
irradiance of the light sources (i.e., 300 to 1,200 W/m\2\) did not 
influence the protection factors of the sunscreens.
    Two comments stated that testing procedures using modified lamps 
that produce mostly UVA wavelengths are unsatisfactory for evaluating 
the UVA protection afforded by a sunscreen drug product because the 
filters required for such testing can remove 40 percent or more of the 
critical, damaging wavelengths between 320 and 340 nm. In addition, the 
comments pointed out that some of these modified lamps contain UVB 
wavelengths below 320 nm that can overwhelm UVA effects. However, 
another comment stated that PFA values obtained using modified lamps 
and delayed tanning or erythema as endpoints weigh the UVA II (320 to 
340 nm) heavily and, for the most part, ignore the contribution of the 
longer UVA wavelengths (360 to 400 nm). One comment stated that failure 
of reciprocity may occur with very long exposures and that making each 
test exactly the same for each differently configured UV source used 
and its particular energy distribution would be impossible. Another 
comment stated that dose reciprocity for the endpoints of delayed 
tanning or erythema has been reported to fail at irradiances between 10 
and 50 milliwatts per cm2 and below. One comment noted that the 
interval after exposure at which the responses are evaluated can bias 
the results. The comment added that the infrared energy or heat 
delivered to the skin during these exposures can affect and alter the 
results.
    The comment stated that no currently existing lamps accurately and 
fully reproduce the UVA spectrum of sunlight. Pure UVA I lamps, e.g., 
the UVASUN series, are used primarily for photochemotherapy where only 
longer wave UVA (above 340 nm) is wanted. Xenon arc lamps modified with 
a WG345 filter are deficient in UVA energy below 335 nm and do not 
accurately reflect the energies or damage risk of natural sunlight UVA. 
According to the comment, such lamps remove too much of the energy at 
the short end of the spectrum and result in overestimating the UVA 
effectiveness of sunscreens. Xenon arc lamps filtered with 1- or 2-mm 
WG335 filters contain UVB wavelengths below 320 nm and thus can greatly 
affect test results. Thicker WG335 filters cut off too much lower UVA 
energy to accurately represent UVA risk.
    One comment submitted a method that assesses the attenuation of the 
incident solar radiation on human skin by a sunscreen (Ref. 27). This 
method utilizes the principles of diffuse reflectance and fluorescence 
excitation spectroscopy. The method directly measures the optical 
properties of the skin decoupled from its biological responses. Both 
procedures are based on the same principle, any modification of the 
surface of the skin will produce changes in its absorption properties. 
Application of a sunscreen modifies the surface of the skin by 
providing an additional barrier through which solar radiation must 
penetrate before reaching the skin. Measurement of the absorption 
properties of the skin before and after sunscreen application yields 
the transmission spectrum of the product and permits calculation of it 
solar protection value for the wavelength range investigated. The 
comment stated that, because this method allows repetitive testing, 
evaluations of substantivity and water resistance are possible. The 
comment contended that, because the testing is done on human skin, 
questions of binding, distribution, and photodegradation can be 
answered. In addition, the comment maintained that this testing 
procedure, like the Diffey and Robson method (Ref. 12), does not suffer 
from a lack of dose reciprocity, as observed with UVA-induced acute 
skin reactions. The comment concluded that this procedure allows for 
the correct estimation of the attenuating power of a sunscreen; thus, 
the protection potential of products in sunlight will be correctly 
estimated.
    The agency would like to discuss the advantages and disadvantages 
of the various recommended testing procedures and the following 
specific questions regarding these test methods at the meeting:
    1. Which of the in vitro test methods described above would be 
adequate to evaluate the UVA protection provided by a sunscreen drug 
product? Why would the others not be appropriate?
    2. Are the results of in vitro UVA testing methods relevant to the 
UVA protection provided to consumers by a sunscreen drug product during 
normal use?
    3. Does the APP test method demonstrate whether a sunscreen drug 
product provides meaningful protection against harmful UVA radiation?
    4. Do the data show that the APP test does not overestimate the 
actual amount of UVA radiation blocked by most sunscreen drug products? 
Identify the data.
    5. Describe the specifications for an appropriate light source for 
the IPD testing method, e.g., spectral distribution, intensity, etc.
    6. What UVA radiation doses are appropriate for use in the IPD 
test?
    7. When should the IPD response be read--immediately, or 1 or 2 
hours after UVA exposure?
    8. Is the IPD reaction relevant to protection of the skin from UVA 
damage?
    9. Do the available data demonstrate that the IPD test is stable, 
nonvariable, and reproducible? Identify the data.
    10. Do the available data demonstrate that the IPD testing response 
obeys dose reciprocity over the anticipated irradiance range?
    11. Are results of the PFA testing procedure relevant to protection 
of the skin from UVA damage? Identify the results.
    12. Do the data show that the PFA test obeys dose reciprocity? 
Identify the data.
    13. Can the interval after exposure at which PFA responses are read 
affect the results?
    14. Does the heat or infrared energy delivered to the skin during 
PFA testing exposure affect the results?
    15. Describe the specifications for an appropriate light source for 
the PFA testing method, e.g., spectral distribution, intensity, etc.
    The agency has concluded, under 21 CFR 10.65, that it would be in 
the public interest to hold a public meeting to discuss the many 
questions and topics associated with UVA testing for OTC sunscreen drug 
products. The proposed rulemaking involves 21 CFR parts 352, 700, and 
740; however, the discussion at the public meeting will be limited to 
part 352.
    The agency requests information regarding UVA protection claims and 
UVA testing procedures from any interested person. However, the agency 
requests that only new or additional information not previously 
included in the rulemaking be submitted. Data should be specifically 
limited and relevant to the questions asked. Any individual or group 
may, on or before April 29, 1994, submit to the Dockets Management 
Branch (address above), comments and data relevant to the questions and 
topics on UVA protection and testing procedures contained in this 
document. Two copies of any comments are to be submitted, except that 
individuals may submit one copy. All comments are to be identified with 
the docket number found in brackets in the heading of this document. It 
is not necessary to resubmit data and information submitted previously 
to this docket.
    Any individual or group interested in making a presentation at the 
meeting should contact Jeanne Rippere (address above). Presentations 
should only address the questions and topics listed previously. Persons 
interested in participating in the meeting must also send a notice of 
participation on or before April 29, 1994, to the Dockets Management 
Branch (address above). All notices of participation submitted should 
be identified with the docket number found in brackets in the heading 
of this document and should contain the following information: Name, 
address, telephone number, business affiliation, if any, of the person 
desiring to make a presentation, summary of the presentation, and the 
approximate amount of time requested for the presentation.
    Groups having similar interests are requested to consolidate their 
comments and present them through a single representative. Depending on 
the time available and the number of participants, FDA may require 
joint presentations by persons with common interests. After reviewing 
the notices of participation, FDA will notify each participant of the 
schedule and time allotted to each person.
    The administrative record for the OTC sunscreen drug products 
rulemaking is being reopened to specifically include only the 
proceedings of this public meeting. The administrative record will 
remain open until July 31, 1994, to allow comments on matters raised at 
the meeting.

References

    (1) Letters from W. E. Gilbertson, FDA, to T. P. Koestler, 
Westwood Pharmaceuticals, Inc., K. M. O'Brien, Schering-Plough 
Corp., N. J. Lowe, UCLA School of Medicine, and M. A. Pathak, 
Harvard Medical School, coded LET45, LET47, LET50, and LET52, 
respectively, in Docket No. 78N-0038, Dockets Management Branch.
    (2) Lowe, N. J. et al., ``Indoor and Outdoor Efficacy Testing of 
a Broad Spectrum Sunscreen Against Ultraviolet A Radiation in 
Psoralen-sensitized Subjects,'' Journal of the American Academy of 
Dermatology, 17:224-230, 1987.
    (3) McKinlay, A. F., and B. L. Diffey, ``A Reference Action 
Spectrum for Ultraviolet Induced Erythema in Human Skin,'' CIE 
Journal, 6:17-22, 1987.
    (4) Comment No. C104, Docket No. 78N-0038, Dockets Management 
Branch.
    (5) Reference 29, Comment No. C128, Docket No. 78N-0038, Dockets 
Management Branch.
    (6) Reference 30, Comment No. C128, Docket No. 78N-0038, Dockets 
Management Branch.
    (7) Figure 2, Comment No. C128, Docket No. 78N-0038, Dockets 
Management Branch.
    (8) Comment No. 135, Docket No. 78N-0038, Dockets Management 
Branch.
    (9) Sayre, R. M. et al., ``A Comparison of In Vivo and In Vitro 
Testing of Sunscreening Formulas,'' Photochemistry and Photobiology, 
29:559-566, 1979.
    (10) Sayre, R. M. et al., ``Sunscreen Testing Methods: In Vitro 
Predictions of Effectiveness,'' Journal of the Society of Cosmetic 
Chemists, 31:133-143, 1980.
    (11) Cole, C. A., and R. L. VanFossen, ``In Vitro Models for UVB 
and UVA Photoprotection,'' Comment No. RC1, Docket No. 78N-0038, 
Dockets Management Branch.
    (12) Diffey, B. L., and J. Robson, ``A New Substrate to Measure 
Sunscreen Protection Factors Throughout the Ultraviolet Spectrum,'' 
Journal of the Society of Cosmetic Chemists, 40:127-188, 1989.
    (13) Reference 13, Comment No. C257, Docket No. 78N-0038, 
Dockets Management Branch.
    (14) Comment No. C140, Docket No. 78N-0038, Dockets Management 
Branch.
    (15) Kelley, K. A. et al., ``In Vitro Sun Protection Factor 
Evaluation of Sunscreen Products,'' Journal of the Society of 
Cosmetic Chemists, 44:139-151, 1993.
    (16) Reference 33, Comment No. C135, Docket No. 78N-0038, 
Dockets Management Branch.
    (17) Comment No. C171, Docket No. 78N-0038, Dockets Management 
Branch.
    (18) Lowe, N. J., M. M. Mobayen, and T. Bourget, ``UVA 
Protection in Human Epidermis: Comparison of Three Sunscreen 
Formulations,'' The Journal of Investigative Dermatology, 94:551, 
1990.
    (19) Stockdale, M., ``A Novel Proposal for the Assessment of 
Sunscreen Product Efficacy Against UVA,'' International Journal of 
Cosmetic Science, 9:85-98, 1987.
    (20) Diffey, B. L., ``Pitfalls in the In Vitro Determination of 
Sunscreen Protection Factors Using Broad Band Ultraviolet Radiation 
Detectors and Solar Simulating Radiation,'' International Journal of 
Cosmetic Science, 11:245-249, 1989.
    (21) Comment No. C128, Docket No. 78N-0038, Dockets Management 
Branch.
    (22) Gonzenbach, H. U., and R. E. Romano, ``UVA Sunscreen In 
Vivo Effectiveness Measurements,'' Cosmetics & Toiletries, 106:79-
84, 1991.
    (23) Chardon, A. et al., ``Method for the UVA Protection 
Assessment of Sunscreens Based on Residual Immediate Pigment 
Darkening,'' Comment No. C104, Docket No. 78N-0038, Dockets 
Management Branch.
    (24) Cole, C. A., and R. VanFossen, ``Testing UVA Protective 
Agents in Man,'' Comment No. C137, Docket No. 78N-0038, Dockets 
Management Branch.
    (25) Cole, C. A., ``Multi-Center Evaluation of Sunscreen UVA 
Protectiveness using the PFA Test Method,'' Comment No. C137, Docket 
No. 78N-0038, Dockets Management Branch.
    (26) Cole, C., and R. VanFossen, ``Measurement of Sunscreen UVA 
Protection: An Unsensitized Human Model,'' Journal of the American 
Academy of Dermatology, 26:178-184, 1992.
    (27) Kollias, N. K., and R. R. Anderson, ``The Non-Invasive 
Determination of UV-A Sunscreen Effectiveness In Vivo,'' in 
``Biological Responses to Ultraviolet-A Radiation,'' edited by 
Urbach, F., Valdenmar Publishing, Overland Park, KS, pp. 371-376, 
1992.

    Dated: March 25, 1994.
Michael R. Taylor,
Deputy Commissioner for Policy.
[FR Doc. 94-8023 Filed 4-4-94; 8:45 am]
BILLING CODE 4160-01-P